Determining tactical operations policies for an auto carrier using discrete-event simulation

[摘要] Passenger cars are either imported to or assembled in South Africa, and thendistributed to the customer. An important part of the supply chain is formedby the auto carrier companies who do this distribution. The basis of this thesisis a study that was executed in collaboration with a South African auto carriercompany, and the objective was to improve the long-distance auto carrier fleetmanagement through improved tactical operational policies. These policiesfocus on application of the fleet by assigning transportation vehicles to routes,as well as the business rules that must be followed at pick-up and drop-offlocations.Several rules were developed during this study, which, together with specifictransportation vehicle (carriers) assignments, form operational scenarios.The quality of each scenario was evaluated using discrete event simulation overa six month time-span, and considering four decision parameters simultaneously.These parameters are 1) useful kilometres travelled by the long-distancecarriers, 2) empty kilometres travelled by the same long-distance carriers,3) the expected number of cars waiting to be transported and 4) the expectedtime it takes to deliver a car to its destination.A high level of uncertainty prevails in these transportation operations,while fluctuating demand calls for the dynamic allocation and management ofcarriers in order to sustain an acceptable service level in a cost-effective manner.The best tactical policies should maximize the number of cars deliveredon time at the lowest cost. Major constraints considered are staff- and maintenanceschedules.While searching for the best of several scenarios, multiple, conflicting criteriahad to be evaluated, as mentioned above. Two multi-criteria decisionanalysis (MCDA) methods were used namely SAW and TOPSIS, while theMahalanobis distance method was also applied as an evaluation technique.These methods were used to rank scenarios. Additionally, the application ofPortfolio theory and the efficient frontier was investigated for applicability tothe problem studied. An analogy to the efficient frontier providing an additionalmeans for scenario selection and evaluation was developed. The result of this study provides the decision maker of the auto carrier companywith a tactical decision aid, consisting of the MCDA and Mahalanobisscenario rankings, a cost-benefit graph and a fleet portfolio efficient frontier,to aid long-distance carrier management. Additionally, a sensitivity analysiswas also done for strategic planning concerned with the sufficient long-distancecarrier fleet size.The first part of this thesis comprises a study of literature in which freightoperations, auto carrier studies and the auto carrier context in South Africaare investigated. The problem is formulated and a suitable formulation andsolution tool identified. Multi-criteria decision analysis is also investigated inorder to enable scenario evaluation.The solution development phase consisted of the simulation model conceptdevelopment, acquisition of input data, model verification and validation,scenario construction, simulation execution, and analysis of results.

[发布机构] Stellenbosch University