[RR/LN – Rolls-Royce] Entry barriers, learning curves, and the hazards of complexity

Below the wings of a widebody commercial aircraft, you might find one of these:

That right there is a Rolls Royce Trent 1000 jet engine.  The engine forces air through several compressors, squeezing the air to 1/50th of its original volume before directing it into a combustion chamber, where it is mixed with fuel and passed through a turbine (the expanding butt-end of the engine, opposite the fan) that extracts energy out of the hot, compressed air to power the fan and compressors, before finally being shot out the back. 

The stress that this process imposes on the engine’s parts is unbelievable – 96 turbine blades spinning at 10k revolutions/minute, exposed to 1,700 Celsius degrees.  To withstand such extreme heat and spin, those turbines must be spun from a single crystal of a precisely honed alloy, ensuring there are no boundaries that might fracture under the strain.  Each blade is thoroughly examined by X-Ray, ensuring its dimensions are accurate to within 7 microns (that’s .007 mm).  The titanium blades that comprise the fan at the front of the engine, responsible for 75% of the engine’s thrust, endure a complex heating and shaping process to render them light but durable.  All 20 blades carry a slightly different mass and are attached to the disk in an order that ensures precise balance.       

The Trent is an engineering marvel, composed of 8 modules, each containing thousands of components.  Each bolt securing one module to another is adjusted to an exact torque.  Every step of the manufacturing process is repeatedly checked.  No part of it is entirely automated.  Welders, mechanics, engineers, and inspectors swarm the engine at various points as it’s being assembled¹.

With labor playing such a central role, learning curve effects become relevant.  The learning curve is a theory borne of the empirical observation that the number of hours required to manufacture an incremental unit declines along a predictable path, such that the labor costs of producing the second unit is 80% of the first, the third unit is 80% of the second, and so on, until improvement eventually plateaus².  This theory follows from the premise that the longer a person does something, the better she gets at doing it, and encompasses not just the muscle memory of technicians on the shop floor but technical and process improvements that organically emerge when groups of people persistently work to achieve something as efficiently as possible. 

As an engine manufacturer makes more engines, its marginal costs trek down.  Scale economies also take hold as fixed costs are diluted over more units.  Newer, more fuel-efficient models build on yesterday’s innovations.  A focused newcomer, even one with access to 10s of billions in capital, would find itself incessantly behind the learning curve, unable to manufacture the same engine at comparable cost.  Looking forward and reasoning back, the entrant would see that it could not steal the share necessary to justify the many billions of capital and years of R&D required to compete in the first place.  

Scale economies, R&D, and learning curve dynamic sum to formidable entry barriers that have given rise to entrenched oligopolies.  There are only a few companies in the world with the engineering and manufacturing chops to produce such an essential and complicated piece of equipment at the volumes that airlines demand.  In narrowbody aircraft³, CFM International⁴ and IAE⁵ account for 71% and 19% of the installed base, respectively; in widebody⁶, GE has got 50% and Rolls Royce 35% of outstanding engines⁷. 

The learning curve phenomenon, though widely observed, mustn’t be taken for granted.  After all, we’re dealing with human beings who operate according to incentives and cultural norms.  If engineers and technicians don’t proactively look for efficiencies or process improvements are hampered by the strictures of a repressive bureaucracy, the cost curve may not bend according to expectations.  And then, of course, supply chain hiccups and engine malfunctions can disrupt the assembly process, reversing learning curve gains and giving rise to significant, unanticipated cost overruns that together conspire to dilute an engine program’s returns.