Comparing airport apron layout designs using computer simulation and the cross-entropy method

[摘要] ENGLISH ABSTRACT: The demand in air travel is continuously increasing. In order to handlethis increase in demand, airports need to physically expand or themanagement of the airports needs to improve. When the demand atOR Tambo International Airport gets too high, more passengers willneed to travel to Lanseria International Airport, which will thereforeneed to be expanded. The study was done in collaboration with VirtualConsulting Engineers, who decided that the concept of AtlantaInternational Airport in Georgia, USA, which is ranked the busiestairport in the world, will be used in this expansion. The aim of thestudy was to minimise passenger walking distances and waiting timesat Lanseria International Airport. This was done by comparing di erentairport apron layouts, using simulation, and improving the aircraftgate assignment, using the cross-entropy method.Four di erent designs of airport layouts, all based on that of AtlantaInternational Airport, were compared in the study. A model of eachwas developed using simulation. The performance measures used tocompare the designs included 1) the average walking distance of arrivingand departing passengers at the airport, 2) the average time spentat the airport by arriving and departing passengers, 3) the averagedistance travelled by aircraft at the airport, 4) the average time bywhich each aircraft is delayed and 5) the average number of aircraftpresent at the airport.The walking distance of arriving and departing passengers was largelya ected by the way in which ights were assigned to gates. The gatesat the airport are of three di erent sizes: small, medium and large.Small aircraft can park at any of the gates, while medium aircraft canonly park at medium or large gates and large aircraft can only park at large gates. Three rules for the ight-to-gate assignment processwere developed. In therst two rules an arriving ight was assignedto the available, suitable gate closest to the terminal building. Theconstraint that small aircraft cannot be assigned to medium or largegates if there are small gates available and that medium aircraft cannotbe assigned to large gates if there are medium gates available,was used in Rule 1 and not in Rule 2. In the third rule, metaheuristicoptimisation was used to determine a ight-to-gate assignment schedulewith the objective of minimising the passenger walking distances.This metaheuristic optimisation was performed in real-time and wasthus repeated every time a delay occurred at the airport.The background of airports, simulation, metaheuristics and relevantcase studies were investigated in the literature review. The simulationand metaheuristic optimisation models were then developed. Theresults identi ed the best of the four designs that were compared. Itwas also concluded that the use of metaheuristic optimisation, usingthe cross-entropy method, results in a reduction in passenger walkingdistances at the airport.

[发布机构] Stellenbosch University