Excessive Modularity Hindered Development of the 787

TAGmclaren writes "The Harvard Business Review is running a fascinating article exploring the issues facing Boeing's Dreamliner. Rather than simply blaming outsourcing, as much of the commentary has been focused on, the article delves into the benefits of integration and how being integrated when developing a new product gives engineers more degrees of freedom. From the article: 'Historically, Boeing understood that, and had worked with its subcontractors on that basis. If it was going to rely on them, it would provide them with detailed blueprints of the parts that were required — after Boeing had already created them. That, in turn, meant that Boeing had to design all the relevant pieces of the puzzle itself, first. But with the 787, it appears that Boeing tried a very different approach: rather than having the puzzle solved and asking the suppliers to provide a defined puzzle piece, they asked suppliers to create their own blueprints for parts. The puzzle hadn't been properly solved when Boeing asked suppliers for the pieces. It should come as little surprise then, that as the components came back from far-flung suppliers, for the first plane ever made of composite materials... those parts didn't all fit together. Time and cost blew out accordingly. It's easy to blame the outsourcing. But, in this instance, it wasn't so much the outsourcing, as it was the decision to modularize a complicated problem too soon.'"

No specs?

[BVis]

So Boeing told the contractors what they needed to build, but didn't give them hard specifications? What the hell? Two things:

Boeing needs to have their collective asses kicked for doing it this way, and:
The subcontractors should never have agreed to the work without specs first.

The first one is probably the result of Boeing not wanting to spend the engineering dollars to develop the blueprints, and the second is due to the enormous amounts of money involved in making the parts.

Now that I know this, you'll never catch me on one of those abominations. What the hell was Boeing thinking?

Interesting problem

[astralagos]

Systems design in engineering basically involves drawing a box around a bunch of parts and saying "this is a system". The interfaces after that are hopefulyl clean -- good systems design does that, but implicit in the choice of a system breakdown is efficiency loss. I might not, for example, think about the fact that the giant engine at the heart of my car could also run heating. There's this long term conflict in engineering between the need to abstract, which enables all forms of delegation, including outsourcing, subcontracting and even building teams, and the loss of efficiency. Good engineers learn things at an almost inexpressible level,developing jargons for the systems under their purview -- in the case of Boeing, there was literally one guy who was their expert on cabling. If you wanted to submit a drawing change, he could envision the change in the cabling of the plane and whether the change was physically possible. That's always been the bane of system abstraction - you find these things that have to cross systems and, if you don't recognize them early enough, they come back to bite you in all sorts of creative ways. Kelly Johnson was a big believer in this. His rules for skunkworks explicitly required that engineers had to be within a specific number of feet of the shop floor -- that way they weren't too divorced from the reality of the products they were making. You see this in the design of a lot of the early computer systems as well, parts bolted together in weird ways before we started developing this high-level view of what systems actually made up a computer.

Re:No specs?

[bunratty]

No, the problem wasn't no specs. The problem was that the system was designed on paper first, without actually building it. Then the specs for the individual pieces were created, and those individual modules were built from the specs. The idea was that then the parts were completed, they would be integrated and work perfectly together. Of course, that never happens because when the pieces come together for the first time, unanticipated problems occur. This is why early integration [ibm.com] is a good idea and is part of the philosophy of release early, release often [wikipedia.org].

Engineering

[the eric conspiracy]

Boeing didn't want to hire all the engineers needed to design the 787. So when they outsourced these subsystems they also counted on their suppliers to do the engineering of these subsystems.

The problem is that engineers are not fungible. Boeing didn't appreciate this, any more than the software industry did when it started outsourcing.

An aerospace structural frame engineer is not the same thing as a marine structure engineer. There are huge differences in the body of experience despite the fact that they both use the same tools.

This was the primary cause of the delays Boeing had. It will continue to be a problem for anyone who tries this sort of outsourcing.

Re:No specs?

[vlm]

Having lived for over 5 years in Japan, I doubt the Japanese subcontractors would build anything without clear specifications.

The problem is the toilet seat bracket had to be made 1/10 mm thicker for supersized passengers, and that was properly annotated by the seat mfgr on blueprint revision #24352. Unfortunately the news never reached the design engineers for the landing gear who need to adjust blueprint revision #7652 foward by 2 mm

My extensive experience with electronic design is if the Chinese say they'll give you a container full of old fashioned thru hole 1K resistors at a tenth of a penny each or whatever they will in fact do so. Maybe they painted the resistor color code with lead paint and the assembly line workers are political prisoners, but the resistance and power dissipation specs will be more or less as per the data sheet. And you can talk the Indonesians into providing a container full of microwave medium power bipolar transistors with a Pd of one watt and a Ft of 25 GHz for two bucks each and they will in fact do it. But god help you if you tell both of them, "I'd like a class A biased driver amp assembly so you two kids cooperate mkay?" Now multiply that by one zillion subcontractors all operating more or less without adult oversight by design to save money as a new project management technique, and you've got a recipe for disaster.

"I've got an idea, lets improve the obvious metrics, then you little guys can work together to design and build it which will make me a bunch of money, mkay?" That stuff doesn't fly.

Re:No specs?

[DerekLyons]

Any manufacturer worth their salt (and Boeing is one) should be able to fully prototype their product prior to outsourcing bits of its production (except for things like batteries). But any parts that needs to fit together very precisely should be prototyped first.

Welcome to the 21st century, I hope your trip from 1950 was a comfortable one! Here in the 21st century computers have advanced to the point where we don't need to physically prototype things anymore - it can, and has been, done digitally for well over a decade now. In fact, one of the most complex things that man is currently building (a nuclear submarine, something else new to you but take my word on it) are now routinely and successfully designed and built without any physical prototype.
 
Seriously - you and a bunch of other commenters are utterly clueless as to the state-of-the-art of over a decade ago. Boeing has built (IIRC) three new aircraft now (plus major upgrades like the new 747) using completely digital design, visualization, and validation tools. While it's not entirely a mature technology, it's not new and very complex vehicles are and have been in service for years that were designed and built using it.
 
Prototyping persists with smaller items because the requisite systems and software are so expensive, and is enabled by the fact that the teams involved are relatively small and simple, physically located in one place, and the prototypes are relatively cheap and new ones can be turned around (at worst) in a few weeks. On the other hand, a mockup/prototype of something like a nuclear submarine or a major aircraft can cost tens of millions of dollars (or more) and take a year or more to assemble. To assume that the latter must prototype because the former do is... ludicrous at best.
 
The problem here isn't lack of prototyping, it appears that they tried to extend the process too far and the management systems weren't weren't set up properly to handle the new process.