Statistics for Environmental Engineers

Скачать в pdf «Statistics for Environmental Engineers»


Low Level Specimen Low Level Specimen


because of some weakness in the measurement protocol. Additional interpretation of the Youden plot is possible. Consult Youden and Steiner (1975), Woodside and Kocurek (1997), or Miller and Miller (1984) for details.

Ruggedness Testing


Before a test method is recommended as a standard for general use, it should be submitted to a ruggedness test. This test evaluates the measurement method when small changes are made in selected factors of the method protocol. For example, a method might involve the pH of an absorbing solution, contact time, temperature, age of solution, holding time-stored test specimens, concentration of suspected interferences, and so on. The number of factors (k) that might influence the variability and stability of a method can be quite large, so an efficient strategy is needed.


One widely used design for ruggedness tests allows a subset of k = 7 factors to be studied in N = 8 trials, with each factor being set at two levels (or versions). Table 9.2 shows that this so-called 27 4 fractional factorial experimental design can assess seven factors in eight runs (ASTM, 1990). The pluses


TABLE 9.2


A 27 4 Fractional Factorial Design for Ruggedness Testing


Factor_ Observed


Run


1


2


3


4


5


6


7


Response


1





+


+


+



У1


2


+





Скачать в pdf «Statistics for Environmental Engineers»