Tuesday, June 21, 2011

Is Your Investment in the Structured Model Generating a Positive Return?

As I've mentioned before, in my current job I get to visit and interact with many people and companies involved with product development; several every week. Most of these people/companies use history-based modeling of some kind as their primary design tool. As I walk through their processes and tool usage,I am continually amazed at how little value most companies get from their history based design tool and the resulting structured model, considering the level of investment that goes into the development and manipulation of the structured model. It seems that rarely is the potential value of the structured model realized. I say this based on a few observations:
  • I've had the opportunity to look at many history based models/assemblies and review their history trees and structure, coming from a variety of different CAD tools. While I see some models that are well structured with a good representation of design intent, most models are very poorly structured and really make no sense. They are clear examples of people just trying to get their job done as quickly as possible; get the models built, evaluate the design, make drawings and make parts. And a lot of these models are coming from companies that I would consider high tech and leaders in the industry.
  • For further review I often load these models into my CAD tool of choice (Creo Elements Direct Modeling) and run a simple interference check. Since they are structured models I assume that errors will not exist, as many people believe that the structured model keeps them from making errors. Unfortunately, and all too often, I find many errors.
  • I also find that recreating models is not too uncommon. This doesn't indicate that they are not getting value from the models, it just indicates that the cost of a well structured model may be much higher than most realize.
Why are we investing so much time, effort and money into history-based modeling and the structured model, when so often we are not getting the value from it that we could/should? Here are the most common answers I hear, or can assume based on what I see, with #1 being the most frequent.
  1. They don't know of, or recognize, any alternative. Many people truly don't know that there is an alternative to history-based modeling and the structured model. Value is a relative thing, and if there is no recognition of alternatives, there is no recognition of opportunity.
  2. Many companies assume that strict modeling practices are being followed and as such assume they are getting value from it, but have really not done any work to validate this.
  3. The company may not be open to an alternative as migration costs are assumed higher than the potential return. Of course understanding potential return will require a clear understanding of alternative solutions.
  4. Individuals may not be open to an alternative as they have bettered their career based on their knowledge and skill with a particular tool. Learning a new one will set them back.
  5. User preference often gets higher priority than process improvement when it comes to tool selection. The cost of training can contribute to this. A lack of understanding in how the tool impacts process also contributes to this.
  6. Others?
Of course the alternative I am referring to is direct modeling and the unstructured model. Most all CAD companies are working on this technology and are presenting it in some form or another. This alternative is becoming much more prevalent in our industry.
Before you start jumping to conclusions about the lack of value in an unstructured model, be careful not to assume that the unstructured model cannot be predictably controlled and/or represent design intent. Creo Elements Direct Modeling, for example, has a powerful 3D parametric solver built into it. Users can add control and design intent to geometry and assemblies at anytime in the process, independent of where or how the geometry was created. An IGES file can even be parametrically controlled and easily be made intelligent if desired. Assemblies and mechanisms can easily be brought to life for virtual prototyping. And don't confuse design intent with modeling intent. Here's an old example using imported models (although I should update this to Creo):


So the question is, are you really getting value from your investment in history-based CAD tool and the structured model? If so, what value? Do you understand the total investment, and does the value justify the investment? Are you sure you understand the alternatives?
There is certainly value that can come from the structured model, but many times this value comes at a great cost. I could go on about the various perceived values that people think they are getting from the structured model. Some are valid, some are not. Perhaps that is a topic for a different post.
Paul

7 comments:

Jon Banquer said...

How long would it take to update this video to Creo Elements Direct 18?

How long before I can download the trial version of Creo Elements Direct 18?

Jon Banquer
San Diego, CA

Blake Courter said...

Nice post, Paul.

What do you think is the value in the ad-hoc feature-based direct modeling in Creo Direct? Seems to me that it's a step in the wrong direction.

-Blake
(A SpaceClaim employee)

Paul Hamilton said...

Jon, The free edition of #Creo Elements Direct Modeling v18, called Modeling Express 4.0, due out sometime in July.

Hope I can update some of the videos, just need more time :)

Paul

Paul Hamilton said...

Blake,

The “ad-hoc feature-based direct modeling” (if you want to call it that) in Creo Direct will be useful for people that need to consume and contribute to the design process where Creo Parametric is the primary design tool used in product development. As a complementary tool, Creo Direct can provide the flexibility and ease-of-use that non, or infrequent CAD users need, without any data lose or disconnect from the engineering master.

For those that want to use Creo Direct as an art-to-part standalone design tool, they may want to turn this feature off. The user won’t know the difference - other than the file size being a bit smaller.

Paul

Anonymous said...

Is the new term you are introducing "structured" just a synonym for "history-based"? If so, are are you assuming that anyone who uses a history-based modeler should create a well-constrained (i.e., a "structured" model), and designed in a way so that all the expected changes can be done easily?

I would grant that, such a goal may be desirable and that it can be costly. But what about sloppy history-based modeling? Can such modeling be quite appropriate in some situations? If so, could such modeling be any more time consuming, or expensive, than direct modeling? If you disagree with this statement, could you please show why direct modeling is inherently more efficient than history-based modeling? All of the tools that are available to direct modeling can be (theoretically, at least) made associative, and thus become part of the history-based toolset. As far as I understand, this is how Creo Direct apparently attempts to merge direct and history-based (So, you should have a more intimate knowledge of this idea than the rest of us).

It seems that you are being critical of those who have used history-based modeling in less than ideal terms, at least in your eyes. Perhaps those users have chosen to work at a suitable point in the sloppy-structured continuum so that they get some benefit of history based modeling (e.g., being able to easily change a sketch driving an extrusion) without paying the big price for observing all the rules and practices of structured history-based modeling. What is wrong if someone had chosen to do that? Do all models have to be equally well-structured?

Jon Banquer said...

Thanks for the specifics, Paul. Very much looking forward to learning it and seeing your videos done with it.

I've waited this long for a new user interface so I certainly can wait till July.

Jon Banquer
San Diego, CA

Paul Hamilton said...

Wow, Anonymous, I apologize for not responding to the comments and questions. I totally missed it. And your questions and comments are great and right on topic.

Yes, I use the term “structured” to refer to the history tree with its parent/child structure and relationships. And, yes, I do assume that users of history-based systems should create well “structured” and constrained models.

I certainly know that these types of systems could be used to create geometry, with no regard to structure. The problem comes if we realize that the real design process, including modeling, is a very iterative process. This process combines an equal amount of content creation (modeling) with content manipulation (editing), and in many cases more manipulation than creation. With traditional history-based systems the editing process is completely based on the model creation process – and resulting structure. The only thing you can edit in a history-based system as far as geometry goes is the sketch, constraints, feature parameters and feature order. All of which are part of the “structure”. This is why most makers of history-based CAD are quickly adding direct geometry editing capabilities to their tool set.

Back to geometry creating. Another critical aspect of history-based modeling is that there are many ways to create 3D geometry, however many of the techniques may not be available in a history-based tool as there is no logical way to record the creation process into the history tree. As such the user may not even have access to some of the fastest methods for creating 3D geometry. Some simple examples of geometry creation methods that are not recommended for history-based modeling are basic Booleans. Uniting multiple parts into one, or subtracting or intersecting parts. These tools provide very powerful and fast ways to create geometry, but many history-based tools don’t even include them on the UI. Those that do, attempt to restrict them to multi-body parts, or seriously limit their use.

In general history-based modeling will always be more time consuming than direct modeling, but in many cases the extra effort provides great value.