Sort Planning for Express Parcel Delivery Systems

Thumbnail Image
Khir, Reem
Erera, Alan L.
Toriello, Alejandro
Associated Organization(s)
Supplementary to
Parcel logistics services play a vital and growing role in economies worldwide, with customers demanding faster delivery of nearly everything to their homes. To move larger volumes more cost effectively, express carriers use sort technologies to consolidate parcels that share similar geographic and service characteristics for reduced per-unit handling and transportation costs. This thesis focuses on an operational planning problem that arises in sort systems operating within parcel transportation networks. In Chapter 2, we develop optimization-based models to generate cost-effective and time-feasible sort plans for two-stage sort systems. A sort plan, in this setting, involves determining first a high-level grouping of parcels into piles which are dispatched over time to secondary sorters; there, each pile's parcels are segregated based on their loading destinations and service requirements for final sort and packing. We explicitly model time deadlines for sorting that enable the parcel carrier to meet on-time delivery service guarantees commonly offered in practice. For tractability, we propose an integer programming formulation for solving sort plan optimization problems that separates first-stage sort decisions from second-stage decisions but, importantly, ensures that the first-stage decision model preserves feasibility for the second-stage operations. This formulation allows a detailed time-space model to be replaced by a much simpler model that can be readily solved exactly for large-scale instances found in practice. We illustrate the proposed modeling approach and its effectiveness using real-world instances obtained from an international express service provider. In Chapter 3, we extend our modeling from Chapter 2 to explicitly incorporate various sources of demand uncertainty commonly faced by parcel carriers. Using practitioner insights and industry data, we propose different uncertainty models that take into account changes in arrival quantities and/or arrival times. We exploit certain problem structures to generate computationally-tractable robust counterparts that solve realistic-sized instances. We demonstrate the computational viability of the proposed models based on industry data and show that high-quality solutions can be obtained in relatively short computation times. We show the value of the proposed robust models in providing hub managers with sort plan alternatives that quantify trade-offs between operational costs and different levels of robustness. In Chapter 4, we study a flexible assignment balancing problem for minimizing workload imbalance across resources in sort systems. The idea is to enable the use of simple and practical recourse strategies that allow sort equipment to be reconfigured once information about the actual demand is revealed. We introduce the stochastic k-adaptable assignment balancing problem that generates k resource configurations apriori with the objective of minimizing the maximum workload allocated to any resource; a critical objective for improving utilization and reducing congestion to meet deadlines. The goal is to enable decision makers to tap into the availability of real-time data and adapt their operations to plans that work best under the realized demand while maintaining a good level of consistency and stability desired in practice. We compare exact and heuristic solution approaches and test them on real data obtained from a large parcel carrier. We show that by allowing up to six configurations, sort systems can achieve around 6% improvement on average over traditional fixed plans, which accounts for an average of 90% of the benefits obtained when using fully dynamic settings; illustrating the benefits of limited adaptability in the context of sort operations.
Date Issued
Resource Type
Resource Subtype
Rights Statement
Rights URI