Study On Modeling And Optimization Algorithms For Parallel Operating Equipment Scheduling In Bulk Terminals

A bulk terminal is a basic node in the transportation network, where bulkcargoes such as iron ores are transshipped from one mode of transportation toanother. Its typical operations are as follow:(1) unloading iron ores fromvessels,(2) loading iron ores onto barges,(3) storing iron ores temporarily instorage yard to account for the differences in arrival times of the vessels andbarges. Due to bulk cargo handling technology, structural constraints on bulkhandling equipments and the dynamic production requirements, productionscheduling in bulk terminals is a multi-constraint issue, and then there existdifficulties with modeling and solving it. Focusing on the typical operations inbulk terminals, this paper discusses the theory and methodology of parallelbulk handling equipment scheduling to improve the resource utilization. Themain research contents are as follow:(1) Acommon problem of parallel bulk handling equipment schedulingProduction scheduling in bulk terminals should be the application anddevelopment of scheduling theory in port production. Parallel bulk handlingequipment scheduling consists of stacker-reclaimer, ship loader and shipunloader scheduling, which have both the common character and itsindividual characters. The common character is then abstracted into identicalparallel machine scheduling problem with sequence-dependent setup times tominimize makespan. A mathematical modeling method for it is proposed todescribe quantitative aspects of the internal nature. Based on analyzingproperties and weaknesses of existing chromosome representations designedfor parallel machine scheduling, single-queue chromosome representation isintroduced. Both theoretical and computational comparisons are conducted toexamine the effectiveness of single-queue chromosome representation in genetic algorithm application.(2) Stacker-reclaimer scheduling with two-side and parallel productionOwing to dynamic and uncertain production environments, heavy bulkcargoes need to be handled in storage yard and then stacker-reclaimers are themost frequently used operating equipments in bulk terminals. The Individualcharacter of stacker-reclaimer scheduling can be described as two-side andparallel production with the special eligibility constraint, and then amathematical programming model is developed. To improve the local searchability of genetic algorithms, a hybrid genetic algorithm is then developedbased on single-queue chromosome representation, in which neighborhooddefinitions and the elite search strategy are discussed. According to theproposed mathematical programming model, a lower bound for thestacker-reclaimer scheduling problem is introduced to evaluate the quality ofnear optimal solutions.(3) Ship loader scheduling with two-side and serial productionDue to the increasing bulk cargo throughput and transhipment volumesin bulk terminals are all increasing, barge delay often occurs. The Individualcharacter of ship loader scheduling can be described as two-side and serialproduction with work space and non-interference constraints, and then amathematical programming model is developed. To achieve optimal dynamicbalance between the quality of near optimal solutions and solving efficiencyof hybrid genetic algorithms, an adaptive hybrid genetic algorithm is thendeveloped based on single-queue chromosome representation, in which theadaptive local search procedure is designed based on the presentation ofevolvability characteristics and niche strategy.(4) Ship unloader scheduling with one-side and serial productionUnloading operation should be the beginning point of running a bulkterminal, and unloading efficiency is an important indicator of performanceevaluation for bulk terminals. The Individual character of ship unloaderscheduling can be described as one-side and serial production with thenon-interference constraint, and then a mathematical programming model is developed. According to the biological immune mechanism, an immunealgorithm is developed, which also employs the single-queue encodingmethod that has been applied to the genetic algorithms. Based on theproposed mathematical programming model, lower bounds for the shipunloader scheduling problem are introduced to evaluate the quality of nearoptimal solutions.