What went wrong with the Boeing 737 NG Pickle Fork?

As you may have read, Boeing has a major fatigue issue with the 737 NG Pickle Fork. Information about the nature of the problems can be found in numerous articles on the web, like:

Boeing 737 NG aircraft have a design service goal (DSG) of something like 75,000 flights and are certified for that DSG. Cracks occur before 30,000 flights, so way before reaching the DSG. The 737 NG is certified as derivative from older 737’s. Apparently this implies that the pickle fork is still a safe life item instead of a damage tolerance item. What does that mean? In simple words, safe life analysis means that the fatigue life of the item must be shown. This is done by analysis using S-N curves and should be supported by full scale fatigue testing. Since fatigue shows a lot of scatter, a safety factor on life (called scatter factor) must be applied. For analysis only, this factor is usually 5 or 6, for justification based on full scale testing this factor is usually 2. Since the aircraft is certified as derivative, probably no full scale fatigue certification test has been performed. So analysis is done based on analysis of earlier versions of the 737. Such an analysis must have shown a fatigue life of at least 400,000 to 500,000 flights, yet cracks were found before 30,000 flights.

As mentioned, the 737 NG is certified as a derivative. The first 737 entered service in the late 1960’s. Certification basis then was quite different from certification basis now. Was the 737 an entirely new aircraft? Not quite, the fuselage is a modified version of the Boeing 707 fuselage, developed in the 1950’s. I would not be surprised if the pickle fork is one of the 707 items that appeared in the 737. I would even not be surprised if the first fatigue analysis on the pickle fork was performed in the 1950’s and that since then only changes in load spectrum have been considered (in other words: get an old stress report, multiply the loads with some factor and see how the fatigue life changes).

Now what may have caused the large discrepancy between analysis and actual fatigue life. I will mention a few:

  • Incorrect assumptions in the first analysis.
  • Changes in production methods, surface treatments, etc. over the last decades.
  • Changes in material properties. Although probably the same material specification applies as decades ago, changes within the allowable production and specification window may affect fatigue properties to a large extend. Note that usually only minimum static properties are specified.
  • Developments in analysis methods, e.g. FEA. Decades ago, methods were much less accurate.
  • Assuming linear dependency between (increasing) loads and stresses. This is not necessarely true for joints.

Would damage tolerance analysis have helped? Not if the same errors are made as with the safe life analysis. In that case the inspection intervals would be too long. However, a crack growth analysis differs alot from a S-N approach analysis, so good chances that an analysis would have been made from scratch giving realistic inspection intervals. Obviously (though often forgotten!), the damage tolerance analysis is only valid as long as safe life of the item has not been reached (and that is less than 30,000 flights for the pickle fork).

A lot of people might blame the pickle fork issue to the fact that it is certified on safe life basis. To my opinion, that is only a part of the problem. A much bigger problem is that authorities like FAA and EASA allow aircraft manufacturers to certify completely modified aircraft as derivaties. In my view, the 737 NG, but also aircraft like the 737 MAX and A320 NEO must be certified as new aircraft, or at least according to the latest CS and FAR requirements.

More about mistakes in fatigue analysis

We have compiled an overview of common mistakes made in fatigue analysis. This overview is available in a free ebook. Use the form below to get this ebook.