The Decline of Manufacturability

Or "DEFM":  Designing (Engineers) for Manufacturing".

Design and engineering of products has advanced dramatically over the past decades and centuries. It's hard to remember that knowledge of static vs. dynamic loading on bridges is less than 200 years old ("hammerblow" from locomotives on the Dee Bridge, and thank you to Howard Tayler for pointing out the series), and knowledge of metal fatigue barely 100 years old. We have moved from paper-and-pencil drawings with a few rough manual calculations to modern computer-aided design (CAD) software suites that can pinpoint stress concentrations and conduct various stress analyses before the first metal is cut.

That CAD software now means that machine operators must still be proficient in properly setting up the part on a machine (no small task!), but the actual speed and path of the cutting head shaping a part is completely computer-controlled, ideally by that same CAD model. This has enabled incredible feats of weight-reduction, but has also distanced design engineers from the finished product.

Aircraft manufacturing then...

 

And now

Increasing Distance

Up to 50-75 years ago (maybe even more recent), engineers tended to start their jobs on the production floor. Although they might aspire to designing "clean-sheet" products, they might spend their early days

• Designing manufacturing tooling ("Prevent the part from bending or shifting during machining by clamping it in these two locations"),
• Developing manufacturing plans ("Turn the part on a lathe to make this section a cylinder with radius 8 inches, then change setups to grind that curve on a second section, and then finally manually blend everything with hand tools to avoid stress concentrations"), or
• Making small changes to existing parts to improve durability or ease of manufacture ("If we can replace these slots with round holes, then we can save a half-hour per part").

These experiences gave engineers a deep appreciation for how parts turned from raw material into a finished product. In the back of their minds would always be the question "how would I make this?"

But incoming engineers didn't want to do this work when they could jump directly into designs (who would?), and so gradually this core knowledge of the production floor faded away from the design engineering team. If a part could be envisioned in CAD, then the manufacturing team would find a way to make it. Tolerances would tighten because they made the design more elegant and weight-efficient. Parts could be so thin that they would distort in post-machining surface treatments like shot peen. When made to blueprint, they were beautiful, but they were painful to manufacture and often had high repair and scrap rates in the factory.

Countermeasures

I've seen companies try to combat this. Lead engineers often have that production knowledge and try to impart it to their teams, but if that lead engineer is trying to make a part more manufacturable after the low-level engineer already drafted it, some of the "sins" are already baked into the design.

"Manufacturability reviews" with the manufacturing-engineering team have even more risk: they might catch blantant errors and contradictions in the design ("You defined this dimension differently in two places!"), but being brought in at the end and needing to review a sheaf of parts on a tight deadline ("We need these parts released to manufacturing now!") means that only the most egregious issues are stopped and fixed.

Technically, these issues aren't "errors" and the parts are still "manufacturable": with sufficient attention, precise machinery, and sometimes luck those blueprints could be realized in metal. The problem is that the designs are so difficult to manufacture that they slow the process and create many opportunities for scrap, rework-to-print, and rework-not-to-print.

Side note: as a layman before joining a manufacturing firm, I wasn't aware that last category of rework not-to-print existed. Manufacturing firms maintain some of their sharpest engineers in a team that can review why a part doesn't meet blueprint (like "this section is too thin" or "this hole is too big") and propose a repair that doesn't actually get the part back to blueprint, but so it will perform as well as a "correct" part in service. The skill and creativity of your engineering team dictates how often a manufacturing "oops" is scrapped versus allowed into production. THESE PARTS ARE SAFE, but they suck up engineering and manufacturing resources to analyze and perform the rework.

Suppliers and the company's factories are the final source of input on manufacturability, but theirs are similar to the other inputs above: only the worst parts are worth requesting design changes, and often for them the first question from the (overworked) design-engineering team is "Can you make the part today?"

Case Studies

Fighter jets

My company designed and manufactured the landing gear for two fighter jets: a late-Cold-War design and its replacement. The replacement was heavier and more complex, so unsurprisingly it was more expensive. What is surprising is that it was roughly an order of magnitude more expensive. The older design made extensive use of long tubular structures which are fast and easy to manufacture (lathe or turning operations) while the newer one used multi-axis machining centers for all structural parts which required more machining time and therefore more cost. In addition, manufacturability issues plagued the program during my entire tenure at that company: the design always had more tweaks needed to avoid quality problems and rework. I know several of the people whose names grace the blueprints on both designs and the lead engineers are all very sharp...but the older design was paper-and-ink and the new one was CAD.

Developmental aircraft

I had the honor of being involved in a new aircraft program from concept through first-flight. The customer's design was unusually mature and they had a precise weight target for the landing gear in mind. Normally, weight reductions are valuable (and so we always design to minimize weight), but in this case when we told the customer, "You asked for a landing gear weighing X pounds and we can give you one weighing 80% of X!", the customer responded, "That's great, deliver us a gear that weighs X."

This was alien to the design team. We eventually hit the target weight, but we had many components that were needlessly difficult to manufacture because the team was so used to pulling every ounce out of the system.

As one of my favorite examples of "didn't think through manufacturing" decisions, we had an anti-wear coating planned for a lug halfway up the length of a very long part. The team designed the part and lug and marked it "use this coating" on the lug and a few other high-wear points. Great. Except this coating is applied via spray, and there was no way to apply the coating without spinning the entire part unbalanced around the short axis and with no good clamping points: that's expensive and very-specialized tooling. We lucked out that we could use a different coating applied via dip (much easier), but I have to imagine that we could have designed the part differently (screw in the lug as a separate feature?) if we'd thought far enough ahead.

When visiting suppliers during that program, I tried to bring along one or two engineers from the design team; not necessarily for them to help the suppliers interpret the drawings, but for them to see their drawings become real physical parts. In addition, I wanted them to interact directly with the manufacturing teams to help them understand the results of their work. Management told me we didn't have the travel budget for any of the engineers.

The Way Forward?

As much as I'd like to say "engineers should grow up on the production floor!", companies are too hungry for them and people graduating from school are unlikely to want to have a long apprenticeship that way. But we could put more effort into continuous learning. As soon as an inexperienced engineer proposes a concept for a part, ask them "How would you manufacture that?" And if they can't answer, educate them and/or redesign the part. We should include "field trips" for the engineers to suppliers and factories so that the engineering team can learn which of their parts are manufacturable. We need those communication loops between manufacturing and engineering so that both groups can create parts faster and at higher quality.

Any favorite manufacturing stories? If you're an engineer, do you feel seen or attacked? What do you think? Let me know in the comments below or at blog@saprobst.com.