Job shop heuristics and statistical inference in a taxicab simulation

TLDR: In this article, a simulation of the taxi service system of the University of California Lawrence Radiation Laboratory (UCLRL) was undertaken to investigate (without disrupting normal operations) ways of modifying the operations of the existing system which would result in substantially better service at no great additional cost.

Abstract: A simulation of the taxi service system of the University of California Lawrence Radiation Laboratory (UCLRL) was undertaken to investigate (without disrupting normal operations) ways of modifying the operations of the existing system which would result in substantially better service at no great additional cost. The successful achievement of the above objective resulted from the recognition of the close relationship between job shop scheduling and the decision-making problem of UCLRL's taxi service system; and the applicability, therefore, of some of the rules and heuristics employed in job-shop scheduling. GPSS III was used to model the proposed system, rules and heuristics because it readily permitted the formulating of complex models. Although GPSS flexibility allowed accurate modeling of the majority of the system, GPSS approximations were still necessary to represent the distribution of arrivals. To defend the independent Poisson distribution a statistical inference study was used to establish that the model's distribution was robust with respect to the effects of the actual distribution of arrivals.A brief description of UCLRL is necessary prior to any further discussion of how the simulation study was particularly applied.