The left plot in Fig. 3 displays the minimized expected total cost, as a percentage of the expected total cost before the incoming inspection, at any given amount of inspection resource (i.e., the inspectors’ time). The plot shows that the minimized expected total cost declines as the inspection resource increases; however, the marginal reduction diminishes and will approach zero, finally.
The right plot in Fig. 3 shows the maximized expected rate of return (i.e., the maximized reduction of the expected NC cost divided by the investment in incoming inspections) at different levels of inspection resource investment (i.e., the wage paid to inspectors). The maximized return on investment declines when the investment continues to increase and the reason for it is the decreasing margin of investment. The optimization decision model ensures a non-increasing total cost, thus rendering an expected rate of return greater or equal to 1.
This paper further uses a simpler example with only three parts to illustrate the impact of the resource constraint on sample size decisions. Parameters of the three parts are listed in Fig. 4. This figure displays a series of bubbles in the space of decision variables. Each bubble corresponds to a case at a unique level of inspection resource: the location of each bubble is the optimal decisions, and the size of the bubble measures the minimized expected total cost (as a percentage of the expected total cost before inspection).
Fig. 4 , again, shows that the minimized expected total cost declines if the inspection resource increases. When the inspection resource is sufficient, no longer posing a constraint on sample size decisions, the expected total cost can be reduced to only 3.6 percent of the expected before-inspection cost. More importantly, Fig. 4demonstrates that the three parts with heterogeneous quality features are treated differently by the incoming inspection. Part #1 is seen as the least important by the incoming inspection because both the NC cost and ci/ti ratio are the smallest among the three parts. Although parts #2 and #3 are similar in their NC rate and ci/ti ratio, part #2 is more important than part #3 in that the lot size of the former is 50 percent larger than that of the later. As a result, increments of the inspection resource initially are used to increase only the sample sizes of parts #2 and #3, and the sample size of part #2 increases faster than that of part #3. After the inspection resource exceeds a certain level, the sample size of part #1 starts to increase.