A Systematic Approach to Planning for a Designed Industrial

Measurement precision and setting error are not

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: ting error are not always comparable, however. For example, a, < 1~~1 not unusual is for a continuous-batch mixing process. Suppose that the concentration of constituent A is at 10% and is continuously reduced towards a target of 5%. Batches might be produced with concentrations of lO%, 7%) and 4%. In this case, perhaps 1~~1 l%, but a spec= trograph may measure with a,,, = .l%. Another example is a thermostat, which often provides u,,, < l&sl, especially if it has a “dead zone” in its logic. Alternatively, one may find a, > 1~~1. For example, physical laws may make it possible to accurately set gas pressure in a sealed cavity by setting gas temperature, but there may be no precise way to directly measure pressure. Through In the design of experiments classes he teaches at Alcoa, J. S. Hunter gives a rule of thumb for experiments on existing processes. For each control TECHNOMETRICS, 0, .015 inches Internal, *The x, y, and z axes are used to refer to the part and the CNC machine. Knowledge Sought Experimentation Predicted effects (for various responses) external (guess) Feed rate (% of nominal) 4.3 Proposed settings, based on predicted effects -1% (indicator on control panel) 85-115% Fixture height C. MONTGOMERY FEBRUARY 1993, VOL. 35, NO. 1 The a axis refers only to the machine. variable, low/high settings should be selected to cause a predicted effect (main effect) for the “key” response variable equal to one standard deviation of its variation in ordinary use, C~ (if there is “ordinary use”). This is a large enough change in response to have practical consequence and also large enough to likely be detected if measurement error is negligible and the experiment has enough runs. If the rule of thumb is followed, every control variable has “equal opportunity” to affect the response variable. Naturally, it is harder to suggest such a rule for immature processes. Moreover, other issues and constraints must be taken into account when settings are selected-safety, discreteness of settings, process constraints, ease of changing a setting, and so forth. These are s...
View Full Document

This document was uploaded on 02/12/2014.

Ask a homework question - tutors are online