During the development of electromechanical products there is a natural transit from unconstrained to highly constrained design. The initial ideation phase of the development is often very fluid and not only can the shape of the product change markedly and rapidly but so too can the architecture itself. This is not a reflection of the design team spinning its wheels but rather the team rapidly absorbing new information and incorporating beneficial changes into the design. At some point in the development the CAD can benefit from a spring cleaning or complete re-architecting.
Arguably, there are two main types of CAD modelling, top-down and bottom-up. Bottom-up can be thought of as building from the inside out. Generally the designers have only a few well defined components or constraints and the structure and architecture grows around these components. Top-down approaches are more formal and rigid, often employing a skeleton or master modeling technique. The two techniques have advantages and dis-advantages that are sometimes hotly debated. My experience has shown that both techniques should be used during the development of a product.
The ideation/conceptualization phase being a necessary and beneficial reality of early product design is ideally suited to the nimble and flexible modeling afforded by bottom up design. Bottom-up modelling leads to faster turns of the iteration crank and does not tie a modeller to the rigor of top-down techniques. Ideation in the top-down architecture can stifle a designer from exploring a radically different concept simply because implementing it in the structure of this stiff architecture may be daunting. Conversely taking a design through to production with a bottom-up developed design can lead to issues and frustration since often high level changes do not ripple through the model in an orderly fashion. High level changes made to an assembly constructed with a thoughtful top-down design, especially one employing a skeleton or master model, are typically better behaved.
The difficulty that flows from this is that at some point the CAD needs to be essentially turned on its head. Transforming a reasonably complex design from bottom-up to top-down is not trivial. My experience is that the effort required to essentially scrap the initial concept CAD and move to a top-down design is worth the time and expense because it leads to savings in the latter stages of development. I advocate making the transition when the team has met a big development goal, like completion of the first functional prototype. Generally at this stage, the team has been focused on large changes to achieve things like increased functionality and creatively minimized part counts, and less on CAD. Once a functional prototype has been demonstrated the focus shifts to refinements and improvements, not requiring radical changes and is ideally suited to top down design. An example of this may be for the base of a cart that may or may not have wheels – bottom up could be used to lay out a few options and top down would be best used to change the number and location of wheels. Many of the constraints and assumptions no longer even exist and yet the artifacts of these exist in the CAD. It is therefore, at this point, a reasonable time to throw away the CAD, implement a top-down approach and rebuild the assembly to be robust and hospitable to high level changes.
Of course, the entire CAD model is not useless since the part files are likely salvageable; however, by all accounts, a great deal of the constraint based modelling and CAD architecture requires scrapping. Nonetheless, by investing in this strategic change in the CAD, the design team will be far more likely to quickly adjust to a significant constraint change or overall geometry change with minimal effort. Short term thinking can make this investment difficult to swallow, but the alternative can be much more effort down the road.