Deelip.com

Thursday, July 10, 2008

Solid Edge with Synchronous Technology - Part 5

In this part I will discuss the fourth and final concept of Synchronous Technology: Procedural Features. I also intend to answer the question: "Is it really a hundred times faster?". Before we get to procedural features, let's spend some time on ordinary features.

Traditional parametric modelers have a feature tree that contains a set of features in the order they were created. The problem with the feature tree is that the child features depend on their parents for their existence and/or shape. Modifying a parent feature can affect it's children, and not always positively.

Solid Edge ST creates features the same way they are created by a traditional parametric modeler, but does not order them using parent-child relationships. It's like plucking out all the features from a feature tree and dumping them into a feature basket. If you edit one particular feature in the basket, the rest adjust themselves as necessary. If you remove a feature from the basket, again the rest of the features adjust themselves as necessary. Now this is what everyone raving about Synchronous Technology seems to believe.

However, I feel the need to point out something here. The features that are seen in the Solid Edge ST Path Finder (the equivalent of the traditional feature tree) are very different from the features we see in the traditional parametric modelers. In a traditional parametric modeler, you can edit a feature by specifying new values for it's parameters. For example, for an extrusion feature, you can modify the extrusion distance and the feature will be updated. You cannot do this with a feature in Solid Edge ST. This is because the features here are dumb and have no parameters to edit. The only way to change the extrustion distance is to pull/push the face you extruded in the first place or set up a dimension and edit the dimension. The features in Solid Edge ST are merely a collection of related faces that make up the feature. They do not even know which sketch was used to create them. You can go ahead and delete all the sketches in a part and the model will hold up just fine.

People using traditional parametric modelers may consider this as a drawback, when in fact, it is the greatest advantage. This is precisely what gives Synchronous Technology to power to edit features in any order. Let me try and explain this by means of an simple example. The figure below shows a part consisting of three boxes created as extrusions one on top of the other. In a traditional parametric modeler, creating such a part would automatically set up the parent-child relationships from bottom to top.


When I delete the middle protrusion, the top protrusion gets orphaned and is left dangling in space, yielding a multi body part. Doing the same thing in Solid Edge ST gives a different result.


Now watch what happens if I try to increase the height of the middle box.

The parametric modeler (left) pushes the top box upwards, whereas Solid Edge ST (right) keeps the top box where it was and simply reduces it's height. Which effectively means that I have edited the middle feature without interfering with the feature above and below it. How this happens is quite interesting and needs a little explanation.

Every edge in a water tight solid model has exactly two faces that touch it, one on either side. Moving the edge (which is what happenned in the figure above) affects both faces and they are trimmed or extented as necessary. If one of the faces is deleted (as was the case when I deleted the middle box) the other face spills over the edge and proceeds to find a face that will trim it. When all such faces get trimmed we end up with a watertight solid once again. Even if one of the faces does not get trimmed, the software reports a modeling error.

So as you can imagine we need not be bothered about the order in which the features were created. We are only concerned with the faces of the model (irrespective of the feature that they belong) which will need to be extended or trimmed in order to keep the model water tight.

So to answer the question: "Is this a hundred times faster?", the answer is "For large models, yes. Maybe a lot lesser for smaller and simpler models". Imagine you have a large part with more than a hundred features and you wish to change a parameter in one of the features high up in the feature tree. Your traditional parametric modeler will need to rebuild the entire model while you sit and twiddle your thumbs. Synchronous Technology will only look for faces in the vincinity that will be affected and extend/trim them quickly. It does not matter how high or low in the feature bucket the feature lies. To Synchronous Technology a feature is just a bunch of related dumb faces.

Using the benefit of hindsight, I now wonder why PTC took the heirarchal feature tree approach dedades ago and everyone followed them blindly. Seems like we have been unnecessarily taking the longer route all these years.

Anyways, so does Synchronous Technology employ parametric features at all? Yes, they do and they call it Procedural Features, most probably to distinguish them from ordinary features discussed above. Although pulling, pushing and rotating faces does the trick in most cases, there are times when you need to rely on good old parametric features. For example, I cannot think of any face to pull, push or rotate in order to double the number of holes in the flange below.

This is what traditional parametric modeling is famous for and Synchronous Technology has adopted some of these key aspects. Rectangular and circular patterns are examples of procedural features and work more or less the same way as in traditional parametric modeling. Other examples are holes, thin walls (shell) and rounds. Like I said before, Synchronous Technology offers the best of both worlds and I like the mix.

Here is a video that shows the circular pattern procedural feature.





This wraps up this series and my understanding of Synchronous Technology. But there is still one unanswered question that I asked in Part 1: "Does the industry really need Synchronous Technology?" If you have been following this series then I guess you already know my answer. I would like to know yours. What do you think?

22 Comments:

  • "Does the industry really need Synchronous Technology?"

    Yes it very badly needs it.

    However:

    Synchronous Technology is not going to sell itself. Siemens will need to offer big incentives to get SolidWorks, Inventor, etc. users to switch. Without an aggressive marketing program its going to be very slow going and Siemens will lose the significant time advantage they now enjoy with far better technology than their competition has.

    In the industry I'm involved in Solid Edge with Synchronous Technology will need to be bundled with UG CAM Express and Siemens will have to show why this package from one vendor is a much better solution compared to two separate solutions from two vendors that run inside a parametric-history based modeler.

    Jon Banquer
    San Diego, CA

    By Blogger Jon Banquer, At 6:37 PM, July 10, 2008  

  • My impression: it's an advance in user interface, but not a revolution. I suspect how much of an advance depends on the designer's approach.

    For example, in your example, neither result of increasing the middle box's height is necessarily correct. In the Parametric Tree (PT) version, the total height increases, in the ST version it stays the same. But what if the designer wanted to increase the total height? And, if the designer needed the overall height to stay the same, couldn't they setup that relationship in a PT modeler?

    I suspect part of the problem is that many designers do not fully understand how best to use parametric modelers (just like many programmers still haven't "gotten" Object Oriented programming, let alone stuff like functional programming, DSL's, Unit Testing, and closures). (And maybe that's a problem with PT modeling...)

    Thank you for the series -- it's made ST much clearer.

    --Tony

    By Anonymous Anonymous, At 11:38 PM, July 10, 2008  

  • Anonymous,

    You raised an excellent point. If the designer wanted to raise the overall height, he would check the "Parallel" live rule and the top face would move as well. But in the traditional parametric modeling system, the user has no choice. The top box wil move up whether he likes it or not. Synchronous Technology gives the user options, which he can choose to use or ignore.

    By Blogger Deelip Menezes, At 12:26 AM, July 11, 2008  

  • The changes appear quick in the video, but are they correct. If I am seeing it right, all the corner fillets just became elliptical. This makes them pretty difficult to manufacture.

    Design intent is key...

    By Anonymous Anonymous, At 1:28 AM, July 11, 2008  

  • I am trying to reply to Tony's comments: "I suspect part of the problem is that many designers do not fully understand how best to use parametric modelers (just like many programmers still haven't "gotten" Object Oriented programming..."

    As a Designer do i really need to bother and learn about parametrics before choosing a 3D CAD system?

    As a 2d user looking to migrate to 3d, i would be keen to change the model the way i want to, not the way the system would like to and that is where i can look upto ST with it's advantages of 'Live Rules and 3d driving dimensions'

    A2Z

    By Anonymous Anonymous, At 1:25 PM, July 11, 2008  

  • Minor quibble - if you delete the middle box feature, you'll not get multiple bodies in Solid Edge, but a single body that's disjoint.

    By Anonymous Anonymous, At 9:29 PM, July 11, 2008  

  • A2Z,
    Pick what seems best for you. I'm not a CAD person (my day job is software), but I've always heard good things about Solid Edge (and know enough good Solidworks & Pro/E users to respect those systems, too).

    What was trying to say is that some tools (say Object Oriented programming or Parametric Modeling) require a certain approach to be successful (and if you follow a correct approach, the benefits are there), but don't work well with the "quick and dirty" approach (for example, that Visual Basic lends itself to -- the initial development is quick, but maintenance is a nightmare).

    Deelip's series is interesting, but to really evaluate ST someone needs to model a complex object in Solid Edge with ST, in Pro/E, in Solidworks, etc -- and be proficient in all of those programs. Then they need to start revising the model -- remember, it's not just how quick you can make the initial design, but how easily you can update it, extend it, etc.

    --Tony

    By Anonymous Anonymous, At 10:48 PM, July 11, 2008  

  • Tony,

    I find your point about modification of an existing model particularly interesting. I believe that it is one of the main selling points of Direct Modeling systems. In my opinion, the time taken to create a model from scratch using both techniques may not vary a great deal. It's the modification part that gives direct modeling an edge over the traditional parametric technique. In ST if you want to modify a feature, just do it. If the model has design intent encoded in the form of dimensions or procedural features, they will kick in and ensure that you do not screw things up.

    By Blogger Deelip Menezes, At 12:32 AM, July 12, 2008  

  • I'll have to disagree about "the time taken to create a model from scratch using both techniques may not vary a great deal". With a traditional History based Parametric modeler, everything depends upon the sketch and it's constraints. Because of this, a great deal of time is consumed "fully constraining" sketches. This does not need to be done in ST because Live Rules does that work for you. I also see situations where actual design innovation work makes it necessary for the designer to perform what I call "hack and whack" modelling. This is where the form of the part is the driving reason to model and many design alternatives/changes are occuring to the point where the designer does not care about building in "design intent" because it would most likely change. These parts would then require re-modelling to build in design intent and remove redundant features with a conventional modeller. With ST, kiss remodelling goodbye because now it's all about moving/placing dimensions where you need them.

    By Blogger PellaKen, At 9:59 AM, July 14, 2008  

  • pellaken, you make a good point.

    By Blogger Deelip Menezes, At 10:04 AM, July 14, 2008  

  • This site is new to me, so hello to everyone.
    As I see it, with the traditional parametric approach to the box example, you can have the total height change when you increase the height of the middle section OR you can have the total height remain the same, depending on how you construct the extents of the features. If you want to change the result you would have to re-define the extents.
    However, with ST you can do EITHER one, on the fly, just by setting a couple of parameters. This has to be a vast improvement.

    By Anonymous Anonymous, At 5:41 PM, July 16, 2008  

  • what if i have a imported part with 4 hole features in a circular pattern and i want to make it 8 now? can i use procedural features? if yes, how?

    By Anonymous Anonymous, At 1:56 PM, July 22, 2008  

  • Not directly. You may need to do some feature recognition first. I have not delved into that part yet. Maybe someone from Siemens keeping track of thsi thread may want to ship in here.

    By Blogger Deelip Menezes, At 8:43 PM, July 23, 2008  

  • Where I see problems with this is in the marketing.

    The industry is finally letting go of autocad or 2d in general and a lot of people have moved to Solidworks which was a big move to begin with, and now Solidedge is going to try to swing the solidworks userbase to swtich? In the US, Solidworks seems to be the leader right now and I am sure they want to keep it that way. If for some reason they start losing numbers to solidedge you can bet that Dessault Systems will put something just like ST in Solidworks and call it something else. If you havn't taken a look at Solidworks 2009 and its new features go check it out. In fact in Solidworks 2008 you can actually do most of what you were doing in that video by picking faces and dragging them around to modify your model. Granted its still paramteric and you can only modify the seed of a pattern etc etc. My guess is that they will keep expanding on that. Thier focus seems to be leaning towards exactly what ST gives the user. Time will tell I guess.

    By Anonymous Anonymous, At 5:44 PM, September 25, 2008  

  • Does ST recognize thread features of imported files, and can you modify feature?

    By Anonymous Anonymous, At 3:21 PM, October 10, 2008  

  • I don't believe it can.

    By Blogger Deelip Menezes, At 11:50 AM, October 11, 2008  

  • Hello,
    Back to the pattern feature. Can I change the shape of all instances by creating a constraint on one instance and another constraint on another instance. The wheel scenario does not show that. Am I wrong ?

    By Anonymous Anonymous, At 7:10 PM, October 20, 2008  

  • Hi,
    One more question about pattern. Can I create a dimension between two instances ?
    Regards.

    By Anonymous Anonymous, At 7:47 PM, October 31, 2008  

  • Let me call the traditional parametric model as "what you see is what you belive" and ST as "What you see is what you get".

    Small difference there, but crucial.

    Suresh

    By Anonymous Anonymous, At 5:23 PM, November 18, 2008  

  • Actually in a parametric based modeler, the top face would be createds from an offset workplane, thus the removal or editing of the middle section is not going to affect the top section. this of course would be PLANNED and must be know in advance, and it is exactly this where ST wins, no planning required. As mentioned above, you wouldnt require remodeling to make a model show design intent. The biggest benifit that I can see is when Im sat down with a client making edits on the fly. There is nothing worse than when a client asks for a 'live' edit, the horror on my face must be clear, when i know the thing is going to fall apart and make me have to explain why I cant 'just do it'

    By Anonymous Anonymous, At 8:20 PM, January 17, 2009  

  • I've used both Solid Edge V20 and Solidworks 2008 extensively, but preferred SolidWorks because you have the ability to use the dimensions with tolerances etc from the model and bring them into a drawing. You can alter dimensions of the model from the drawing environment and when entering dimensions in the model you can put your tolerances in there and then.

    Solid Edge is (or certainly used to be) different. You can import the dimensions, but it ends there- there is no real interaction with the modeling environment- you certainly don't put tolerances in whilst making the model- so you effectively have to put all the dimensions of the model onto the drawing all over again, and if you change the model you often lose the dimensions on the drawing which when you are doing a large drawing can be a real PITA. None of the reviews cover this area, and I don't see how ST can help this, it could only make it worse surely? When i was working for SIEMENS VAI (Steel Mills)we used to waste so much time replacing lost dimensions on drawings it was ridiculous.

    By Anonymous Anonymous, At 8:09 PM, February 26, 2009  

  • Hi
    1. What about all these pattern modifications in imported models. Does it work. I think It's not working. If there is a way, please let me know.
    2. Can we convert a blind hole/pocket into a through hole/pocket by pulling the bottom face (Face Move).

    By Anonymous Anonymous, At 5:12 PM, March 05, 2009  

Post a Comment

Subscribe to Post Comments [Atom]



<< Home