Enforcing CAD standards without resorting to a book-to-the-head

In the constant battle to improve CAD data quality, a vigorous CAD modeling standard is an effective remedy and the bane of bad modelers everywhere. The right modeling standard drives both data reusability and downstream manufacturing efficiency. However, for a modeling standard to be effective, its mere existence is insufficient.

Screen Shot 2015-04-22 at 10.45.34 AM

Some, whether driven by unknowing carelessness or perhaps even premeditated mutiny, are going to ignore your finely crafted standard. "But think of the standards!" you might decry in knowing disapproval. Identifying, collecting, and documenting knowledge for the right standard is hard work. What to do when a casual interloper considers your well-crafted tome mere words on a page? How do you get people to follow such standards, anyway?

You could print out your CAD standard in a menacing font with CAP-HEAVY UNDERLINES on bomb shelter grade card stock, if only to enhance its inertial mass as you forcibly accelerate the wisdom into the noncompliant party's cranial lobe. Twitter-Bird

But CrossFit-style AMRAP book-to-the-head isn't going to win you any allies. We're going to have to be more practical and less violent. We're engineers, not barbarians.

Below is a set of underlying tenets to help drive the creation (and subsequent enforcement) of your standard. Perhaps in true meta fashion, this could be a CAD standard standard of sorts. Don't make me throw this at you either.

Here we go:

Do ensure standards add value

This tip might seem like a Captain Obvious PSA, but you’d be surprised at just how many modeling standards go out of their way to excruciatingly catalogue every last entity color or provide a litany of random attributes that ultimately make little difference. Everything that is in your standard needs traceability to real value. If you want your standard to be widely respected and not seen as an obstacle, you're going to have to prove that value. Remember, consistency / quality is often a tradeoff with time-to-market. Meeting market challenges requires a carefully calibrated approach. So don't mess it up by going OCD.

Don't arbitrarily inherit

No two CAD standards are alike, but often the creation or modification of a standard takes cues from someone else's standard. After all, it's easier to start from something than nothing. But what was good for giant corporate conglomerate A is not necessarily good for agile engineering contractor B. Now, I do realize you sometimes don't have a choice (thanks, contracts), but you'd be surprised what can be negotiated when unnecessary costs come into play. Bottom line: question every rule you inherit to ensure that it makes sense for your organization and your products.

Do educate by example

You might feed a man for a lifetime by teaching him to fish, but he won't invent the fishing boat until he understands why he has to do all that fishing to begin with. Engineers need to know why things are in a modeling standard. Effects and value are often downstream, so a little forensic analysis goes a long way. Any reasonable engineer will immediately appreciate causal relationships and will be motivated to do something about it. Reasonable engineers will also pressure unreasonable engineers – community is a powerful tool if it’s on your side. If you can't find direct evidence that part of your standard is making a difference, then you need to rethink it.

Do treat standards as living documents

Technology changes and so too must modeling standards. Letting a standard fall behind the introduction of new CAD features and techniques threatens its relevancy.

Don't unnecessarily limit options

There’s a fine line between eliminating bad practices that result in questionable geometry and dictating precisely how to model. The former avoids problems that will cost your company money, while the latter generally just makes engineers slower and grumpier. Once again, justify specific practices.

Do provide aids

Sometimes fine details are necessary in a modeling standard, be it layer conventions, coordinate systems, or assembly structure. Remove the drudgery of compliance wherever possible. That means templates, macros, add-ons - anything to reduce the cost of compliance.

Do consider automation

You’re going to need some automated checking at some point, because no amount of due diligence can keep up with every noncompliance, especially with the volume of engineering you’re likely to create.  The price of a modeling standard is eternal vigilance. Twitter-Bird

Do provide early and continuous enforcement

For all practicality check early and often. Too often models get rejected only as part of a final workflow approval, after all work has been long completed and schedule pressures are at their most critical. At that point the struggle between cost, quality, and schedule reaches a crucible and can result in hasty exceptions that erode your standard, and create longer-term cost implications. Automation can be handy here as well, give people tools to check not at designated intervals, but on-demand.

Keep in mind engineering natures is just like water or electricity, naturally seeking the path of least resistance. It's not laziness, it's about optimization. When creating and enforcing a modeling standard capitalize on this fundamental philosophy.

And, after all this, if that one last engineer in your group absolutely refuses to play nice with the standard? Book-to-the-head. Twitter-Bird