Re-Design Intent

sentryThe CAD model at its very core, like the engineering drawing before it, is primarily a means of documentation and communication.  CAD technologies continue to ease the creation and manipulation of increasingly intricate 3D models across a dizzying array of platforms and formats.  The focus: reducing the time required to manifest a design or change from a spark in the mind’s eye to something that can be used as quickly and cost effectively as possible.  A noble goal, indeed.  But in the shadow of all this progress and innovation, there’s a silent war threatening the central purpose of engineering communication: design intent.  This summer, in a world where CAD technology reigns, who will defend the last human stronghold for design intent?

To put it simply, design intent is the why of a design.  Why is it made of titanium?  Why is there a notch on this flange?  Why does it look like a Space Waffle?  Frequently, focus remains on the what of the design, as readily presented in a drawing or in a CAD model.  But the rationale that created the what is critically important, now more than ever.  Matt of Dezignstuff lays it out perfectly:

“Design happens in your head. Design is developing the concept of how something looks or works. This is the kind of thing that cannot happen on a computer. It has to exist somewhere before it can be developed on the computer. The computer just records, organizes, or helps others visualize the design. The design was in your head before it was in the computer. If such a thing as “design intent” exists, it consists of a reason for geometric elements or selection of material or manufacturing process. The reasons would be the intent. And the design is the design”

In the past, it was more common for a single engineer to have ownership and responsibility.  If there was issue with the design, the source (the original engineer) was often readily available, often heads down in his/her office trying to solve the world’s problems and optionally chain smoking.  Chad Jackson of Lifecycle Insights said it best in a recent blog posting:

“An engineer’s prime responsibility has traditionally been to design products for the three Fs: form, fit and function. But that wasn’t a responsibility without consequences. It carried personal accountability. Engineers placed their signature on drawings to signify their personal approval of a design. If products failed, drawing signatures identified responsible engineers so they could justify their decisions. In summary, more so than any other group, engineering has had a strong culture of accountability.”

Nevertheless, relying on the original engineer to hang around isn’t really a long term solution (unless the engie happens to be a rather durable android).  Especially for designs likely to outlive their creators, accurately recording design intent is paramount.  Let’s venture thru the design intent saga.

In the drawing centric world of the past and that still persists to some degree, the most critical pieces of design intent was most often recorded in written form, sometimes intrinsic to the drawing, but often not.  When CAD technologies first emerged, they could do little to change this paradigm, even when they transitioned from curves on flat planes to solid bodies.  3D Solids were just that, and it was not possible to practically constrain how they were formed or changed.  One of the many reasons PDM and PLM first came to be was to try and get a hold of all the stuff lying around that had valuable design intent.

Then came along parametric modeling.  For the first time it was possible to imbed the computer model with important constraints – design intent – to control how the model would behave relative to changes.  While not perfect by any means, the technology achieved significant progress towards providing a reliable and intrinsic means to capture at least some forms of design intent.

But there was a bit of a problem.  Parametric models often became exceptionally complex, with lengthy hierarchies of dependent features.  Feature based modeling was based on the noble assumption that the designer would author the model to function intuitively.  The authoring flexibility inherent in CAD tools created an unintended side effect.  Often, the methodology used to build a particular design was not always obvious from one engineer to another.  Furthermore, unless the engineer happened to have +5 clairvoyance (and maximum charisma!) it was rarely possible, given time constraints, to author a model with resiliency across a variety of potentially unexpected changes.  So models that were changed would often break, and there was much grumbling, gnashing of teeth, and occasional throwing of chairs.  Sometimes remodeling was involved to resolve the feature failure – an expensive proposition for an engineering change.  Increased re-use of design assets thanks to PDM and PLM would only magnify the issue.

Compounding the problem, the world of the one responsible engineer has faded.  Much of today’s design whether due to complexity, schedule, or even resource allocation results in multiple engineers touching any one particular design or CAD model.  Change is the one constant in engineering, and often the engineer responsible for the change is not the original designer.  Parametric modeling, despite the obvious advantages was getting in the way.

Then came the resurgence of direct modeling (aka history free).  Direct modeling became the new hotness, while parametric modeling was deemed old-and-busted.  It’s a perception based on the pressures outlined above – too little time, too much complexity.  Changes were easy, simply pushing and pulling on geometry to make quick edits.  The speed of execution was the obvious advantage.   But as an authoring methodology history free modeling outright killed design intent.  Interestingly, this has all happened before and it will happen again.  Where the heck is Starbuck when you need her?

Aware of the design intent regression inherent with history free, a number of variation modeling technologies then appeared under various fabulous marketing names.  These technologies permitted direct modeling to coexist with parametric feature hierarchies or provided other means of defining constraint based design intent while retaining the sheer speed of direct modeling tools.  So the pendulum again began to swing back towards design intent Zen enlightenment.

The latest episode aims to swing the pendulum yet again.  And this time the problem is M-M-M-M-M–M-MultiCAD.   Sorry, couldn’t resist.  Today’s landscape is replete with CAD models passing from one kernel or platform to another again and again and again.  From contractor to supplier to OEM to Uncle Ted.  Depending on the translation, some of the direct editing tools suffer what amounts to death by math.  So remodeling scenarios arise once again.  Along comes technology like TransMagic, designed to handle the CAD translation such that direct edits are once again viable.  It’s an interesting trick, and one that allows engineers to continue pushing, pulling, and dragging away.  But few stop to think about what happened to all that design intent that might have existed before the translation?  Who cares, let’s get this change done and move on with life!

You see my point.  The popularity behind easing the pain of engineering change seems to trump the initiative to record and improve the fidelity of design intent.  Every time.  So the big question is do we intend to hate design intent?  And is that by design?