A common industrial problem is to identify parts that cause assembled products to malfunction. When the cause of failure is not obvious from physical inspection of the parts or the malfunction only appears occasionally, it can be helpful to search systematically for the cause using a factorial experiment. Tippett presented the first known example of troubleshooting an assembly with a factorial experiment. We discussed his example in a previous column (see Ref. [2]). In this column, we will review another fine example of troubleshooting this time originally due to Taguchi (see Ref. [5, p. 75]). Taguchi's assembly, a Diesel injector, consisted of two types of parts, a body, factor A, and a needle, factor B. A general sketch of a Diesel injector is shown in Fig. 1. The purpose of the experiment was to determine if it was the body or the needle that sometimes caused the injectors to malfunction. To find out, Taguchi randomly selected six bodies and six needles from production. His experimental design consisted of assembling the six bodies and six needles in all the possible 6×6=36 factorial combinations.
展开▼
