The role and benefits of using statistical design in experiments for the development and optimization of glass and ceramic products are examined. It is shown that considerable benefits can be derived by following an appropriate design in which a number of factors (processing parameters) are varied simultaneously while the system response such modulus of rupture, toughness, optical loss, etc is monitored.
If it is assumed that ccnplex factor interactions (e.g., having greater than four factors) can be ignored, it is further shown that essentially the same conclusions can be drawn from an examination of only a fraction of the whole factorial matrix. As a consequence, significantly less trials are required to determine (screen and rank) the various single- and multi-factor effects (having up to four factors) on the measured property. Examples drawn from recent work in materials development are presented.
|