Abstract When discussing flow-control matters with production managers , it is noteworthy that whereas there is normally great emphasis on the reduction of variance of specific transformation processes (quality control) as well as on level variability in general, as for instance represented by the Bullwhip effect, the possible disturbing interference, which potentially could arise due to autocorrelation (variability dependence over time) in otherwise level-stable demand streams of events in the production flows, does not seem to attract much attention, actually almost no attention at all. From the literature it is well known that the impact of certain types of autocorrelation in queuing systems (Livny, Melamed and Tsiolis, 1993) can lead to a fairly dramatic deterioration of the system performance compared to an event-independence case. Also, a study on the effect of autocorrelated demand in JIT-production systems (Takahashi and Nakamura, 1998) establishes that autocorrelation plays definitely a non-negligible role in relation to the dimensioning as well as functioning of Kanban-controlled production flow lines . This must logically either imply that production managers are missing an important aspect in their production planning reasoning, or that the 'realistic' autocorrelation patterns , inherent in actual production setups , are not like those so far considered in the literature. In this paper, an attempt to characterise relevant and 'realistic' types of autocorrelation schemes as well as their levels will be pursued. Previous work on the subject of the impact of autocorrelation in queuing system is thereby put into perspective.
Main Research Area:
18th International Conference on Production Research, 2006