Operating within YKK’s integrated manufacturing vision, its Machinery and Engineering Group supplies internally manufactured equipment to factories throughout the world. Despite making much use of purchased parts, in the past the group lacked a strategy for improving the procurement process; most procurement changes were motivated purely by the cost of the parts. Under recent conditions of unanticipated delays in the supply of parts, reviewing procured materials on the basis of cost alone appeared unwise, but the procurement personnel lacked knowledge of the functional evaluation strategy found in quality engineering. To introduce them to functional evaluation, a case study was carried out on the procurement of general-purpose bearings, issues and effectiveness were noted, and this information was shared with the personnel. In particular, the process of choosing among proposed purchases was studied, and a summary of the points that should be considered by the procurement department was prepared. This had the effect of raising the level of awareness of functional evaluation throughout the Machinery and Engineering Group, which began using functional evaluation to confirm the quality of all procured parts. The procurement department also undertook a restudy of the procurement process flow.

抄録全体を表示

Toner charge is the key to good image quality in an electrophotographic system. Many discussions of this key factor are based on the well-known ratio q/m of the toner charge q and mass m. Here we discuss parameter design using an L18 orthogonal array in which the incremental blow-off output data of the developer in the development module of an electrophotographic system are treated as an (m, q)vector series instead of a q/m scalar series. In the analysis, the nonadditive ratios q/m were converted to an additive series of two-dimensional vectors (m, q)．The reproducibility of the standard S/N ratio analysis applied to the vector series was quite good. In contrast, the reproducibility of the standard S/N ratio analysis applied to the q/m distribution generated artificially from the same measurement results was extremely bad. The method we propose provides a simple path toward evaluation of the robustness of multidimensional characteristics.

抄録全体を表示