Further analysis was conducted to assess the validity
of (10) by comparing the calculated trend with that of a
simulated system with parts subject to rework. A simple
discrete event simulation model was constructed using the
software Arena. In the model a specified percentage of
parts, corresponding to p, were rerouted back to the
process queue as a means of capturing the effects of
rework. Fig. 1 demonstrates the output of the relationship
shown in (10), where all inputs aside from p, where p
varies from .1 to .9, were held fixed. The lower curve
shows the relationship given the same input, but evaluated
through the simulation model. The mean and standard
deviation of processing time for both curves corresponds
to product A. Five replications were run in the simulation
for each value of p, and up until 80% rework the queuing
theory result falls within the half width of the variation
among these replications. While the pattern of variation
appears to be consistent, there is a discrepancy in the two
curves which becomes more exaggerated as rework rate
increases. Further study is required to understand the
gradually increasing discrepancy; this could be due to
dynamic or interactive effects that the proposed queuing
model does not account for.
Further analysis was conducted to assess the validity
of (10) by comparing the calculated trend with that of a
simulated system with parts subject to rework. A simple
discrete event simulation model was constructed using the
software Arena. In the model a specified percentage of
parts, corresponding to p, were rerouted back to the
process queue as a means of capturing the effects of
rework. Fig. 1 demonstrates the output of the relationship
shown in (10), where all inputs aside from p, where p
varies from .1 to .9, were held fixed. The lower curve
shows the relationship given the same input, but evaluated
through the simulation model. The mean and standard
deviation of processing time for both curves corresponds
to product A. Five replications were run in the simulation
for each value of p, and up until 80% rework the queuing
theory result falls within the half width of the variation
among these replications. While the pattern of variation
appears to be consistent, there is a discrepancy in the two
curves which becomes more exaggerated as rework rate
increases. Further study is required to understand the
gradually increasing discrepancy; this could be due to
dynamic or interactive effects that the proposed queuing
model does not account for.
การแปล กรุณารอสักครู่..
