Much pain, little gain

Again, we are reaching the point in some technical areas where gains are incremental, yet they consume significant engineering effort, time, and money.

We all know how complicated and sophisticated many consumer products have become today. Consider automobiles, those complex electromechanical systems with hundreds of processors, millions of lines of code, multiple internal networks, countless sensors and motors, and so much more. Though many industry commentators call them “computers on wheels,” I dislike that cliché because it degrades the many real-world mechanical issues they endure in design and use.

It’s not just cars, of course. What were previously basic and dull appliances such as clothes washers, dryers, or kitchen stoves and ranges are now enhanced, networked, and laden with features of dubious or no demonstrable value.

Along with this complexity and sophistication comes the inevitable spike in recalls, bugs, downloads to fix problems (assuming that’s even an option), and similar headaches when trying to get a product to fulfill its basic functions. We’ve all heard the stories of people who haven’t used the vast majority of features in their car, kitchen appliance, or similar, and instead use only a few of the many modes and simply ignore the rest of the available “benefits.”

But a more insidious problem has me concerned: products have become so advanced in their development and debug stages that even their designers can no longer prepare them for release and successful deployment. At the same time, the gain they hope to achieve, even if successful, is modest at best and marginal or trivial at worst.

While it would be easy to substantiate this concern by citing any one of the many electronic-centric products with which we are so familiar, I think it’s actually more instructive to step back and look at a system that has lots of electronics but is also much, much more: a modern passenger-jet aircraft that is in actual use — and with lots of challenges.

What prompted my interest was an article in The Wall Street Journal titled “It’s the Airplane of the Future. It’s Still Grounded” about the engine-related problems that Airbus has had with its A220 mid-size plane (it was originally called the CSeries from Bombardier, but when that company went bankrupt in 2017, Airbus bought and renamed the airplane family).

In addition to modest airframe issues, the Pratt & Whitney PW1500G geared turbofan (GTF) engines (Figure 1), which were supposed to last 20,000 flight cycles before overhaul, actually need to go to the shop at 5,000 cycles, and some are being sent in before 600 cycles. This has, of course, had a major impact on aircraft availability, utilization rates, and actual operating costs.

Figure 1. The highly advanced PW1500G geared turbofan engine has had major operation issues, thus negating its intended benefits. (Image: Pratt & Whitney/RTX Corp.)

What are the problems with the engine? There are several impurities in the parent material of the powdered metal used for the high-pressure turbine (HPT) disks, including impurities in the parent material of the powdered metal. These impurities could lead to premature cracks that could cause uncontained disk failures. There are also issues with the compressors and air seals, and the engines have trouble in hot and dusty environments. The grit in the air blocks cooling holes, degrades seals, and corrodes the metal, which happens far beyond the normal prototype-evaluation phase.

Recalling engines for inspections and replacement parts takes about 250 to 300 days per unit, resulting in about one-third of the units being out of service (Figure 2). This is costly in terms of having to lease replacement planes or cancel routes (and P&W has to pay severe performance penalties to its customers — unlike vendors of consumer products).

The irony is that these new engines, with their advanced technology, such as powdered metal for compressor disk blades, were developed at great expense to save fuel and reduce operating costs. By how much? Just a few percent, which may significantly affect the airline’s bottom line, seems modest in terms of actual gain.

It’s easy to ascribe this problem to the usual challenges many advanced designs have in their early stages. But I wonder if the actual issue is far deeper than that. We’re reaching asymptotic limits in many areas where the incremental gain is fairly small, but the associated development and start-up costs are huge. Think of IC fabs: as we slowly progress down to smaller nodes and eke out ever-smaller gains, the cost of a fab goes from a “mere” billion dollars or two to tens of billions, and that’s ignoring the up-front R&D costs to get there. It’s a textbook example of the law of diminishing returns.

On a closer-to-home level of pain versus gain, consider the stop-start function added in recent years to many internal combustion (ICE) cars to save on fuel and raise gas mileage statistics. The actual savings from this feature, which shuts the engine down when the vehicle is stopped at a light or similar and then restarts it when the driver steps on the accelerator, ranges from near-zero for those who are not stuck in traffic or at red lights to around 3% for those who do encounter those conditions. It was so disliked by car owners that soon after it was introduced, car manufacturers added a dedicated button that allows drivers to deactivate stop/start when they start the car on a trip.

However, that’s just the tip of the stop-start iceberg. It turns out that the invisible engineering-design effort and added manufacturing cost of this feature is quite large: it affects the wear on the starter motor, of course, the battery, the bearings, and many other power-train aspects of the car to ensure the vehicle meets its longer-term reliability and lifetime objectives (see Related Content). The ripple effect of this simple mandate to shut off the engine when stopped at light or heavy traffic and then quickly restart is a clear case of another law of unintended consequences.

There are many other examples of additional technical progress that have come at a severe cost in design effort, complexity, and the potential for serious problems. At the same time, the cost of achieving these and further gains is enormous.

Historically, the response to this gain/pain imbalance is a shift to a new technology using what academics refer to as a paradigm shift, as explored by Thomas S. Kuhn in his insightful 1962 book The Structure of Scientific Revolutions. If you want to know more, I suggest you read one of the many online summaries rather than the incredibly dense, academic-style book.

Perhaps we are approaching another such shift, where quantum computing or optical links and computing will allow us to jump past the computing and speed barriers we presently face, which are taking so much effort to overcome. Let’s check back in a decade to see what happened versus what the pundits and market researchers foresaw.