Optimization Algorithms For Truck Scheduling Problems In Automotive Inbound And Outbound Logistics

Automotive industry plays a significant role in national economic development and has been the pioneer of many advanced ideas and techniques on operations and management.The inbound logistics of auto parts and the outbound logistics of finished automobiles serve the production and sales activities of the automotive manufacturers,respectively,and are the key processes affecting their operational efficiency.Based on the real problems in current operations and the review on the existing literature,this dissertation studies the key truck scheduling problems in the automotive inbound and outbound logistics operations.The main contents are as follows:First,this dissertation considers a truck scheduling problem of delivering different types of auto parts from a supplier park to an automotive assembly line according to time-varying real demands.Traditional logistics methods of auto parts may lead to many problems,such as high inventory costs,dead stock and unexpected shortage.To avoid these problems,major automotive manufacturers build up supplier parks located next to an assembly plant.A group of suppliers are gathered in the park and provide a number of auto parts directly to the assembly line.There is a lack of research in the existing literature on auto parts logistics problems taking time-varying real demands and the characteristics of supplier parks into account.Thus,based on the emerging applications of supplier parks,this dissertation studies the problem of making auto parts dispatching decisions according to the real demands of the parts on a finite time horizon.The goal is to minimize the total working time of required shipments and ensure that the inventory level of auto parts of each type in its buffer beside the assembly line meets the upper and lower limits at any moment.According to the characteristics of the supplier park,the working time of each shipment is determined by the round-trip transportation time between the park and the assembly line and the waiting and operating time at each visited supplier.Moreover,auto parts of each type is delivered in predetermined packages provided by the supplier so that the quantity delivered by each shipment can only be an integral multiple of the number of parts in each package.For this problem,an integer programming model is formulated and the intractability and feasibility conditions are analyzed.A column generation based algorithm is developed to solve the problem.Experimental results based on randomly generated data and a case study based on real data show that the proposed algorithm is capable of generating high-quality solutions for the problem efficiently,and is practical to be used to reduce costs in daily operations of supplier parks.Second,from the view of the basic distribution network in automobile outbound logistics,this dissertation studies an automobile dispatching problem arising in the urban area.The problem is to deliver a set of finished cars from a departure warehouse to a set of dealers.The related dealers are located within the same city with the warehouse or in an area close to the city.Based on the characteristics of this problem,a value is defined for each car to be dispatched and a carpooling cost is defined for each auto-carrier.The larger the value,the higher priority to be delivered the car has.If cars to different dealers are assigned to one auto-carrier,its carpooling cost is defined based on the total distance among the dealers.The cost represents the number and concentration of the dealers that need to be visited by each auto-carrier.Moreover,the cares assigned to each auto-carrier have to satisfy a loading constraint defined by a downward compatible loading structure proposed in this dissertation as follows.Based on the classification of cars,each auto-carrier consists of a given number of slots of different types.Each slot can only be assigned with one car of the same type with the slot or one car of a smaller type.The proposed loading structure fully considers the combinatorial nature of auto-carrier loading.Differing from those based on the weights or lengths of automobiles in the related literature,the loading constraint in this dissertation ensures the practicability of the solutions generated for the automobile dispatching problem.The above problem is formulated as an integer programming model.Based on reformulation,a column generation based algorithm with a branch-and-bound algorithm solving the pricing subproblems to optimality is developed for the problem.According to the structure of the proposed algorithm,two programing structure are considered based on sequential and parallel programing,respectively.Computational results and a case study show that the proposed algorithm can generate optimized solutions efficiently for the automobile dispatching problem in the urban area.The algorithm has a natural structure to take advantage of parallel programing which can reduced the computational time.Third,based on real operations of the largest automotive third-party logistics(3PL)company in China,this dissertation considers a pickup and delivery problem of automobiles on a multi-layer network.The network consists of multiple warehouses and dealers located in different cities.The objective of the problem is to balance the service level of the 3PL company and the total transportation cost.The service level of the 3PL company is determined by the total value of orders handled during each day.The value of each order is defined by how many days it has been waiting to be assigned to an auto-carrier after release and whether it is urgently required by a dealer.The total transportation cost of each auto-carrier on the multi-layer network consists of the traveling cost among different cities determined by the traveling route,and the visiting cost determined by the number of warehouses or dealers visited in each city.This pickup and delivery problem also considers the downward compatible loading structure and the full-load constraint proposed earlier to ensure the loading feasibility of each auto-carrier.Moreover,this problem has to consider a compatible constraint between each auto-carrier and a set of destination cities and other constraints on generating feasible pickup and delivery routes.A column generation based algorithm is developed to solve the problem where the pricing subproblems are relaxed and solved by a dynamic programming algorithm.Moreover,this dissertation proves that,based on the optimal objective value of the relaxed subproblems,an upper bound for the optimal objective value of the problem can be generated during each iteration of the column generation.Based on the upper bounds,an early termination strategy,a local search algorithm for improving the integer solution and a cycling solving strategy,along with parallel programing structure,are developed to improve the implementing efficiency of the proposed algorithm.Computational results based on randomly generated data and a case study show that the proposed algorithm with the strategies is capable of generating near-optimality solutions for the pickup and delivery problem efficiently.The proposed algorithm is the key for the 3PL company to make optimized decisions and establish intelligent operation systems.To sum up,this dissertation focuses on the essential truck scheduling problems arising in the automotive inbound and outbound logistics operations.Based on the characteristics of the problems,efficient algorithms are designed for the problems which can help the automotive industry make optimized decisions.The proposed algorithms are also the foundations of building intelligent operations systems for automotive logistics and achieving collaborative operations on the whole supply chain in the future.Moreover,this dissertation studies the design,improving,testing and evaluating of algorithms based on mathematical programming and column generation for truck scheduling problems considering different optimization objectives and loading and routing constraints.As a result,this dissertation enriches the related research fields on combinatorial optimization problems and algorithm design.