Here is a quick white paper-esque introduction to some new stuff I’ve been learning, presented in the least boring way I could think of impromptu. This will be the first post in a series of posts dedicated to introducing manufacturing in a (hopefully) new perspective.
Manufacturing, in the context of process control, is an active process where material flows through a certain manufacturing unit process. The goals of manufacturing are:
- Quality–the plant must produce what is specified at all times.
- Rate–production rate must meet demand.
- Cost–production must be at a viable, competitive cost.
- Flexibility–rapid change to what is produced must be possible without significant cost or quality penalites to match demand.
However, a level of variation pertaining to a certain property of the workpiece (extrinsic or intrinsic, discussed later) always exists between each individual part made by a unit process–one product is not identical to another. Control refers to the control of variation in the aspect of a workpiece being changed (a prime example being a physical property such as geometry) in order to identify sources of variation, characterize them, and subsequently minimize variation.
Variability can have a very distinct effect on the final product. For example, car body panels can have variable gaps when assembled, high density electrical connectors can have highly variable contact spacing which can cause circuit failure, and in areas such as microelectronics seemingly small variations (such as uneven flow channels in a DNA diagnostic chip) can result in serious performance anomalies. Therefore, we can conclude that high levels of variation negatively affect any manufacturing process, and that in general: variation is the bane of good manufacturing.
Processes are inherently variable; this is a core concept of the Shewhart hypothesis (to be discussed later) and thus statistical quality control. This means that variation can never be truly eliminated from a process–however, it is possible to minimize variation of both deterministic and random origins. Tomorrow, I will start to clarify how to do so via first discussing variation within a general process model framework, so that there is a solid basis from which to build upon.