What's more, he says, composites allow engineers to make custom shapes, but these custom shapes compound the already difficult modeling problem. "You have many more design options, which can be both a strength and a weakness. There are many more things I can do with composite materials--add strength in specific places, take it away--but then you have combinations of both the geometry and the particular layup of the composite materials" that are unique.

Boeing's mechanical stress tests start with representative pieces (known as coupons), then move on to progressively larger parts of the structure, and finally to the full structure. Boeing puts the structural parts into huge hydraulic machines that bend and twist them to mimic stresses that go far beyond worst-expected conditions in real flights. It was during such tests that problems emerged with structural spars in the wing box.

Shanahan said in last week's conference call that Boeing has traced the problem back to an error in earlier modeling analysis, but he did not explain the details. "We discovered it. We'll go back and correct it," he said. Shanahan added that Boeing has not lost faith in its decisions to more widely use composites; 95 percent of thousands of tests have yielded as-good or better-than-expected results. On one such test--of the composite-made fuselage "barrel"--engineers had to stop the test for fear of breaking the test equipment, he bragged.

David Roylance, a composites expert and associate professor in materials engineering at MIT, says that Boeing's experience with the 787 shows that the industry is still on a learning curve in using composites more widely in commercial planes. "There are a whole variety of things with composites that are engineerable, but are different than metals," he says. "So it takes time for people to feel comfortable with it."