December 08, 2023, ©. Leeham News: We are discussing the different phases of an airliner development program. After covering Conceptual, Preliminary, and Detailed design, the manufacturing of prototypes, and their roles in Flight Tests, we now look at Production.
Last week, we explained why aircraft projects often fail 100 to 200 aircraft into production. What’s not well understood is the effects of production learning on product cost.
A new airliner consists of millions of parts, and at least a few hundred thousand of those are bespoke designs adapted to the aircraft’s configuration. The rest are standard parts like rivets, fasteners, brackets, pipes, ducts, wires, connectors, and system parts that could be reused from other aircraft projects, often with modifications.
Anyone who has worked with design and production knows that the initial fitting together of manufactured parts creates challenges. Mating areas don’t match exactly; fits are too tight or too wide as tolerances of manufacture cause the non-fit of items.
The way the designer and production planner have foreseen how things shall be brought together might not be optimal. When the production mechanic has put things together several times, it becomes obvious there is a better way to do things.
Modern 3D design and simulation tools have improved the situation, whereas, in the days of manual drawings or 2D CAD (basically an electronic Drawing Board), the initial fit of parts was troublesome.
The reason is that it’s difficult for a designer to care for all the effects of 3D fits, including tolerances of manufacture. For the initial assembly of parts, the production planner and designer had to be hands-on in production to understand what problems their design and instructions caused during the assembly of a system or structural parts.
Modern 3D tools force a designer to insert bolts into a thread that he’s specified and physically mate part objects into an assembly on the screen to check parts fit, Figure 2. During the assembly of parts the CAD tool will signal if the tolerance specification of two parts creates a fit problem during assembly.
It means the 3D design tools have improved the situation, but at the same time, the demands of tight, efficient designs have increased, balancing the improvement in fit to a degree.
The effects of the nonperfect design or manufacture of parts and the learning and understanding of how to fit parts together cause an increased time used for the production of systems, structures, and aircraft sections until things get changed, and the operations are part of brain and muscle memory.
A longer time used for the assembly of something is the same as a cost increase. If you study the production of aircraft, you conclude that the cost of production is, to a very high degree, the cost of work hours. The parts that go into a manufacturing operation are called parts in a “Bill of Material” or BOM. But the truth is what’s most of the time labeled as the material is not material; it’s material or parts that have a lot of work hours in them.
Hence, early produced parts of an aircraft are more expensive to produce, simply because more workhours are spent in fabricating them.
We showed the typical cost curve for the combined learning curves of an aircraft last week, Figure 3.
In the example, the extra cost of the first parts to mature parts was 250% to 300%. This is when a project runs well. Most aircraft projects contain design changes between prototypes, first serial units, and serial units a bit into production.
Such changes cause new fit problems and changes in how things shall be put together. The consequence is an increased use of work hours and increased cost.
A recent example is the Boeing 787, where Boeing publicized the learning curve data in their quarterly reports, Figure 4. The prototype aircraft cost more than 10 times the mature costs to build, and the early series cost more than five times the cost of mature units to build.
As the production was troubled by a bad fit of parts and redesigns (a big change from Aircraft 21 forward), the learning curve essentially restarted for certain areas of the aircraft when the design changed during serial production.
It can be seen that the learning curve for the 787-8 didn’t settle down to the typical shape until 100+ aircraft had been produced. Had Boeing not been a financially strong company during the 787 cost overruns (development and production), the project could have brought the company into severe financial difficulties.
Another example of a learning curve that has been hard to master has been the CSeries/A220 curve. The first delivery of the CS100 was in July 2016. Airbus says the cost of production of the A220 will be equal to the net sales price by 2025 when it should pass 400 aircraft produced. It means the CSeries/A220 has cost money to produce and sell during the first nine years of production, proving what we have discussed about the production learning curve effects.
The lowering of the production cost with higher serial numbers is not only due to the better fit and learning how to do things, but it’s also because as production progresses, more rational methods of manufacturing are employed. With higher production numbers, the motivation to invest in production automation or better jigs and tooling comes when the amortization of the investments gets assured with the progress of aircraft sales.
The cost of initial production is a very tough load of an aircraft project and, if not properly assessed and planned for, can challenge the existence of the OEM.
Given the very large financial impact of the initial production costs for an aircraft program, a number of method improvements have been developed to try and lower the cost for early production of an aircraft. We look at such developments in the next Corner.