Aerospace Production Engineering

Or, why things take so long.

It would be forgivable to believe that as a product’s design nears completion the engineer’s job is pretty much done. If they did it well, there is a complete documentation package that production can use to build 1 or 10 or 100 of the item that was designed. But it is surprising how much engineering effort is required between design completion and hardware delivery. Production engineering is an often overlooked, but critical bridge between the drawings and the hardware. I have found production engineering to be of vital importance even in the early stages, before the first serial number has been completed.

Assuming the designers have created the perfect spacecraft on paper, it is inevitable that undesirable situations in manufacturing arise, especially with complex systems. Machined parts may have a hole in the wrong place, or it is discovered that a plating process couldn’t accurately control build-up, or an incorrect material was used. Whatever the case, this question is always asked: is this good enough, not the way it was designed, but the way it is? When schedules are tight that question is asked all the more quickly. If it can’t, we start to ask if the part can be repaired, or if other parts can be modified to fit, or if assembly steps can be changed be to correct for the error. But the first question is always “can we ignore it?”

It is in this moment that the value of documenting both analysis and the engineering *thought process* is realized. Repeating efforts by analyzing tolerance stackups or verifying stress is not trivial. In most cases the engineer tasked with finding out if the parts are ‘good enough’ isn’t the one who designed it or analyzed it in the first place. Even with a low rate of employee turnover, the slow nature of aerospace development means that the documentation can be quite old. Interpreting stale powerpoint slides and reviewing drawings can take a lot of time, and may be one of the most frequent activates of a production engineer. This process of digging through ancient records to ascertain margins, tolerance stackups or design intent has been referred to by one of my colleagues as “Engineering Archeology”. Engineering Archeology is a common practice at many aerospace firms, and they employ many ‘Archeologists of Engineering’, commonly abbreviated as AE. With some products, like those licensed from another firm, all that may be documented is what is required for manufacturing, leaving intent or margins a complete mystery. We know it works, just not why.

If lucky, our Archaeologist has uncovered previously lost insights into the nature of the product. It is to be hoped that they will immediately update documentation with the new analysis, evidence of why it is acceptable, or change the design to reflect new information about process limitations. If not done soon enough the parts will re-ordered the same way with the same problems.

And then there are the occasional design errors. With surprising regularity, our Archeologist will scratch their head and wonder: “of the dozens of products that have already shipped, by what blind luck did any of them fit together?” As it turns out many manufacturing processes are much better than what is required, and getting lucky is common. In the case where a design or interface error is discovered which affects performance or fit, the pressure to use parts ‘as-is’ switches to use 'almost-as-is'. Adding a dead-bug mod on a board or filing a part seem like quick solutions, but too-often the root-cause is never addressed. When quick-fixes work, they tend to immediately become ingrained and unmovable.

Problem products which need extra inspection, discussion, research, repairs and waivers are awful for schedule and bad for cost. Having seen it first-hand, it is easy to see the truth behind Augustine's Law XII: “It costs a lot to build bad products.” Dealing with bad products is just one of the things that keeps engineers busy after they are “done” with the spacecraft. But mostly, it just so happens when you build something the first time there is a lot that can go wrong, even with parts that are simple by aerospace standards.​