Engineers routinely make design changes in order to cut costs. But, in this case, they should have looked before they leaped.
A young man took a friend for a spin on the back of his motorcycle. Moments later they skidded through a stop sign and smashed into an oncoming car, smashing themselves up pretty good in the process. The rear brake pedal had failed during the attempted stop, which roughly doubled the stopping distance. The motorcycle was of recent make by a famous manufacturer. I was called in to analyze the metallurgical aspects of the failure.
The pedal was of a common aluminum alloy containing about 6% (by weight) magnesium added for strength. Aluminum alloys are readily melted and may be cast into a variety of shapes. Cast objects may be comparable in strength to their counterparts produced by rolling, forging, or another competing process, and they are usually cheaper to produce.
The may in the preceding sentence is a big one. Any of a number of errors in mold design, melting practice, or melt chemistry may result in a product with unacceptable properties. In addition, casting is a net-shape process, so there is no chance to ameliorate the effects of heterogeneities. In contrast, rolling or forging may remove a bubble by welding the sides together or elongate a globular nonmetallic inclusion into a harmless thread.
A vital step in analyzing a failure is study of the fracture surfaces. Tools for analysis range from the naked eye and optical microscope, to the high-power scanning electron microscope (SEM). The SEM has great depth of field to go with high magnification, so that the fracture surface may be examined in great detail. Close examination of the SEM image of the rear brake pedal gleaned some interesting clues about the failure.
I noticed rounded features, which were dendrites that had grown from the melt into gas bubbles or voids. I also observed rough regions, where the dendrites had been broken away from one another in the fracture. Only a small part of the surface was apparently ever bonded, so that the strength was far below the value for sound, low-porosity material. Dendrites are supposed to grow to impingement to give a non-porous solid.
The problem started with the presence of water, probably from humid air. When the molten aluminum contacted water vapor, the aluminum combined with the oxygen atoms to give aluminum oxide, and the hydrogen went in to solution in the melt. Hydrogen is less soluble in solid aluminum than in molten, hence is segregated to the melt during solidification. The high hydrogen content in the melt produced bubbles, ultimately leading to the fracture mode seen in the SEM image.
Hydrogen is easily removed from molten aluminum by bubbling chlorine gas through the liquid. The chlorine bubbles capture the hydrogen and carry it off. So why wasn't the melt de-gassed? Apparently the foundry did not customarily de-gas and no one told them they should. Also the foundry had never before cast a structural component. Earlier production had been limited to ornamental objects, such as porch light brackets and the like where fracture is both unlikely and benign. There was no record of an engineer of any kind ever having set foot in the foundry.
The motorcycle company had changed from a forged and welded steel pedal to the aluminum casting to save about a dollar per pedal. The company issued a recall two years after the cast pedal was introduced, but that was too late for the youths on the subject motorcycle.
Was it worth the change? There were only a few thousand of the pedals made by the foundry in question, so that the overall dollar value of the savings over the old steel pedal was small. I know of several accidents that were directly caused by pedal failures, each of which resulted in a settlement in the million dollar range.
Readers may draw their own conclusions about the wisdom of the design change.