- Home
- Research
- Core Topics
- Logistics and Supply Chain Management
Logistics and Supply Chain Management
Reinforcement learning framework for freight demand forecasting to support operational planning decisions
Year: 2022
Freight forecasting is essential for managing, planning operating and optimizing the use of resources. Multiple market factors contribute to the highly variable nature of freight flows, which calls for adaptive and responsive forecasting models. This paper presents a demand forecasting methodology that supports freight operation planning over short to long term horizons. The method combines time series models and machine learning algorithms in a Reinforcement Learning framework applied over a rolling horizon. The objective is to develop an efficient method that reduces the prediction error by taking full advantage of the traditional time series models and machine learning models. In a case study applied to container shipment data for a US intermodal company, the approach succeeded in reducing the forecast error margin. It also allowed predictions to closely follow recent trends and fluctuations in the market while minimizing the need for user intervention. The results indicate that the proposed approach is an effective method to predict freight demand. In addition to clustering and Reinforcement Learning, a method for converting monthly forecasts to long-term weekly forecasts was developed and tested. The results suggest that these monthly-to-weekly long-term forecasts outperform the direct long term forecasts generated through typical time series approaches.
Read the full publication
Resilience of U.S. Rail Intermodal Freight during the Covid-19 Pandemic
Year: 2022
The rapid onset of the COVID-19 pandemic in March 2020 marked a challenging time for the US and its freight industry. Manufacturing slowed, consumer purchasing patterns changed, and for many, shopping moved online. The freight industry suffered a sharp decline in shipments, followed by a surprisingly quick rebound. The industry had to adapt quickly to meet fast-changing demand and supply patterns upended by global supply chain disruptions. This paper uses U.S. intermodal activity data, supported by in-depth interviews with leaders of railroads, intermodal carriers, equipment manufacturers, car leasing companies, shippers, and e-commerce players to characterize and assess how the rail industry met the challenge of this demand whiplash and other performance impediments. What emerges is a rich picture of the multi-actor intermodal supply chain, the impacts of COVID-19 on it, the performance of the logistics system in general, and railroads in particular during the pandemic. Industry interviews revealed that a handful of choke points, many of which were outside the rail industry, complicated supply chain responses to COVID-19. The paper shows how the rail industry was an essential component of pandemic resilience, demonstrating a high level of adaptability to meet consumer and business demands. Through the use of depth interviews it reveals the complexity of the intermodal supply chain, and it accurately foretells the subsequent disruptions that continued to plague that supply chain long after the initial impacts of the pandemic.
Read the full publication
Train platforming and rescheduling with flexible interlocking mechanisms: An aggregate approach
Year: 2022
This paper proposes a route-based model for the Train Platforming and Rescheduling Problem (TPRP). Built on the concept of Degree of Conflict (DOC), the proposed model can accommodate various interlocking mechanisms with an aggregate railway yard representation. Thanks to the topology of a typical yard, such an aggregate representation promises to reduce the size of the optimization problems concerning yard operations. The TPRP model is formulated as a mixed integer linear program, and solved using both a commercial solver and two heuristic algorithms developed based on the idea of rolling horizon. The proposed model and algorithms are validated using several case studies constructed using data collected at a large high-speed railway station in China. We find the proposed TPRP model can produce, with reasonable computation resources, high quality platform/schedule decisions for real-world applications. In addition, the heuristic algorithms consistently offer high quality approximate solutions at a computational cost considerably lower than what is demanded by a benchmark commercial solver. The results from a simulation model show the differences between various interlocking mechanisms are well captured using a unified aggregate yard representation based on DOC. As expected, more flexible interlocking mechanisms can achieve greater operational efficiency at the expense of looser safety standards.
Read the full publication
Potential for a logistics island to circumvent container port congestion in a constrained environment
Year: 2020
This paper examines a new hybrid intercontinental freight transport alternative (IFTA) that combines a logistics island (i.e. an offshore container port), vertical take-off and landing (VTOL) aircraft, and ocean vessel transport. The hybrid IFTA offers a ‘midway’ alternative for intercontinental shippers that is cheaper (and slower) than conventional air freight, but slightly faster and significantly more reliable (albeit, more expensive) than conventional ocean vessel transport calling at busy ports that may be subject to disruptive delays. To compare the proposed hybrid IFTA with the conventional air- and ocean-based IFTAs, this paper employs a utility maximization framework. A mathematical model determines the shipper value of time (VOT) range, normalized by payload, for which the proposed hybrid IFTA would be preferable to the conventional air- and ocean-based IFTAs. Preliminary results suggest that this range of shipper VOT, normalized by payload, is between $0.6/ton-hour and $22.9/ton-hour, indicating a potential market for the proposed hybrid IFTA might exist. Additionally, sensitivity analyses reveal several interesting insights; most notably, the attractiveness of the proposed hybrid IFTA hinges on decreasing the distance VTOL aircraft transport freight between the logistics island and an onshore warehouse or transloading facility.
Read the full publication
Shipment status prediction in online crowd-sourced shipping platforms
Year: 2020
This paper empirically studies the matching and delivery process in a major crowd-sourced delivery platform. The aim is to develop models to understand and predict crowd-shipping delivery performance and using the findings to design incentives to improve user experiences as well as system performance. We apply the random forest machine learning algorithm to predict the shipment status of 14,858 crowd-shipping requests recorded between January 2015 and December 2016 throughout the U.S. The models are used to predict three phases of the crowd-shipping performance, namely bidding, acceptance, and delivery, using shipping request, built-environment, and socioeconomic features as explanatory variables. The results demonstrate that the context of the shipment provides strong predictive performance even when shipping request and package information is unknown. Calculating the sensitivity of bid probability, we show that offering a higher reward and posting a shipping request in the morning has the largest effect on the probability to secure a bid. We also find that larger shipments, out-of-state destinations, and peer-to-peer shipments lead to higher sensitivity, likely reflecting the higher perceived risks of such transactions. In practice, the models presented in this study show promise in their ability to effectively predict shipment status in real time. We illustrate a valuable application of the sensitivity analysis derived from the random forest models to develop customer-tailored crowd-shipping smartphone applications. Based on the data mined from past deliveries, customers are given empirically based delivery forecasts for their specific package request and can modify delivery requests to increase their odds of delivery. We find that pricing is the variable with the highest potential to increase delivery probability followed by the timing of the request.
Read the full publication
Supply, demand, operations, and management of crowd-shipping services: A review and empirical evidence
Year: 2019
Crowd-shipping promises social, economic, and environmental benefits covering a range of stakeholders. Yet, at the same time, many crowd-shipping initiatives face multiple barriers, such as network effects, and concerns over trust, safety, and security. This paper reviews current practice, academic research, and empirical case studies from three pillars of supply, demand, and operations and management. Drawing on the observed gaps in practice and scientific research, we provide several avenues for promising areas of applications, operations and management, as well as improving behavioral and societal impacts to create and enable a crowd-shipping system that is complex, yet, integrated, dynamic and sustainable.
Read the full publication
Performance analysis of crowd-shipping in urban and suburban areas
Year: 2019
Crowd logistics is a novel shipping concept where delivery operations are carried out by using existing resources, namely vehicle capacity and drivers from the crowd, thereby offering potential for economic, social, and environmental benefits. Despite the promise of this new logistics model, little is known about its actual functioning, performance, and impact. This paper presents a pioneering study of the performance of a real crowd-shipping system in the U.S. using empirical data from 2 years of operations. We contribute to the literature by: (1) defining performance metrics and developing models that account for the specificity of crowd-shipping systems by distinguishing the essential stages from bidding to acceptance and delivery of shipments, (2) identifying the significant covariates, including shipment features, built environment, and socio-demographic factors giving rise to different delivery performance outcomes, and (3) deriving sensitivity analysis to study the performance and implications of crowd-shipping in urban and suburban areas. The analysis is formalized as two-level nested logit models with nests representing bidding and delivery outcomes. The results show that not only does the delivery outcome performance vary significantly between urban and suburban areas, but the explanatory factors also vary significantly for the two contexts. Additionally, several factors have ambiguous impacts depending on the stage. Larger shipment size (versus strict deadlines) leads to increasing (decreasing) the likelihood of bids being placed, while having the opposite effect when it comes to the delivery phase. The findings highlight the need for developing different strategies to foster and improve the performance of this novel system depending on both the urban–suburban shipping context and the stage of delivery.
Read the full publication
Adaptive orienteering problem with stochastic travel times
Year: 2018
In this paper, we evaluate the extent to which one can increase the likelihood of collecting greater reward in an orienteering problem with stochastic travel times by adapting paths between reward nodes as travel times are revealed. We evaluate whether this adaptivity impacts the choices of reward nodes to visit in a setting where the agent must commit to reward nodes before commencing operations. We explore the computational challenges of adding adaptive consideration in the selection of reward nodes to visit and examine the extent to which one can capture some of the benefits of adaptivity with a simpler model.
Read the full publication
Optimal Flexibility Configurations in Newsvendor Networks: Going beyond Chaining and Pairing
Researcher(s): Achal Bassamboo, Ramandeep S. Randhawa,
Jan A. Van Mieghem
Year: 2010
We study the classical problem of capacity and flexible technology selection with a newsvendor network model of resource portfolio investment. The resources differ by their level of flexibility, where “level-k flexibility” refers to the ability to process k different product types. We present an exact set-theoretic methodology to analyze newsvendor networks with multiple products and parallel resources. This simple approach is sufficiently powerful to prove that (i) flexibility exhibits decreasing returns and (ii) the optimal portfolio will invest in at most two, adjacent levels of flexibility in symmetric systems, and to characterize (iii) the optimal flexibility configuration for asymmetric systems as well. The optimal flexibility configuration can serve as a theoretical performance benchmark for other configurations suggested in the literature. For example, while chaining is not optimal in our setting, the gap is small and the inclusion of scale economies quickly favors chaining over pairing. We also demonstrate how this methodology can be applied to other settings such as product substitution and queuing systems with parameter uncertainty.
Intra Market Optimization for Express Package Carriers with Station to Station Travel and Proportional Sorting
Researcher(s): Luke Schenk, Diego Klabjan
Year: 2010
The flow of packages and documents, called splits, of an express package carrier consists of picking up the splits at costumers' locations by a courier and bringing them to a local station for sorting. Next the splits are transported to a major regional sorting facility called the ramp. At the ramp splits can be sorted again and then they depart to a hub. From this hub they are moved to the destination ramp, where the entire process repeats in the reverse order. In this work we focus on the afternoon and evening operations concerning with stations and the ramp. We deal with the sorting and transportation decisions among these locations. If splits are sorted in a particular sequence at the stations, then resorting at the ramp is not required, which can bring cost savings. The most important decisions are: (1) which splits to aggregate at the stations, and (2) what is the most efficient transportation among locations. In this work we enhance the existing model by considering several options for modeling the sorting process at stations and the ramp, as well as the possibility of vehicles traveling from one station to another station to consolidate volume before proceeding to the ramp. We model these processes by means of a dynamic program, where time periods represent time slices in the afternoon and evening. The overall model is solved by approximate dynamic programming, where the value function is approximated by a linear function. Further strategies are developed to speed up the algorithm and decrease the time needed to find feasible solutions. The methodology is tested on several instances from an international express package carrier. Our solutions are substantially better than the current best practice and the best solutions obtained from an integer programming formulation of the problem.
Industrial Strength COMPASS: A Comprehensive Algorithm and Software for Optimization via Simulation
Researcher(s): Jie Xu, Barry L. Nelson, L. Jeff Hong
Year: 2010
Industrial Strength COMPASS (ISC) is a particular implementation of a general framework for optimizing the expected value of a performance measure of a stochastic simulation with respect to integer-ordered decision variables in a finite (but typically large) feasible region defined by linear-integer constraints. The framework consists of a global-search phase, followed by a localsearch phase, and ending with a “clean-up” (selection of the best) phase. Each phase provides a probability 1 convergence guarantee as the simulation effort increases without bound: Convergence to a globally optimal solution in the global-search phase; convergence to a locally optimal solution in the local-search phase; and convergence to the best of a small number of good solutions in the clean-up phase. In practice, ISC stops short of such convergence by applying an improvement-based transition rule from the global phase to the local phase; a statistical test of convergence from the local phase to the clean-up phase; and a ranking-and-selection procedure to terminate the clean-up phase. Small-sample validity of the statistical test and ranking-and-selection procedure is proven for normally distributed data. ISC is compared to the commercial optimization via simulation package OptQuest on five test problems that range from 2 to 20 decision variables and on the order of 104 to 1020 feasible solutions. These test cases represent response-surface models with known properties and realistic system simulation problems.
Business Intelligence for Gang Scheduling
Researcher(s): Diego Klabjan
Year: 2010
Railway tracks wear down and this need to be constantly maintained. Groups of maintenance workers, called gangs, are responsible for such maintenance tasks. Throughout a year, a gang works for a few days in a particular track section and then reallocates to another section. Railways incur significant expenses related to gangs. They range from the direct costs such as salary and travel allowance to indirect costs consisting primarily of the impact to operational disruptions. It is thus of vital importance to railways to schedule the gangs as efficiently as possible.
In conjunction with the Norfolk Southern Corporation (herein called NS), we have started developing a gang scheduling information system based on business intelligence and state-of-the-art analytics. At the core of the system there will be a sophisticated optimization algorithm considering the multi-objective nature of the problem and all of the underlying complexities. The algorithm will be composed of initially constructing a schedule and then iteratively refining the schedule based on state-of-the-art mathematical programming techniques combined with very large neighborhood local search strategies. After initially rolling out the system as NS, we plan to repackage the software around the Software-as-a-Service business model and offer it to other railways.
Global Dual Sourcing: Tailored Base-Surge Allocation to Near- and Offshore Production
Researcher(s): Gad Allon,
Jan A. Van Mieghem
Year: 2009
When designing a sourcing strategy in practice, a key task is to determine the average order rates placed to each source because that affects cost and supplier management. We consider a firm that has access to a responsive nearshore source (e.g., Mexico) and a low-cost offshore source (e.g., China). The firm must determine an inventory sourcing policy to satisfy random demand over time. Unfortunately, the optimal policy is too complex to allow a direct answer to our key question. Therefore, we analyze a tailored base-surge (TBS) sourcing policy that is simple, used in practice, and captures the classic trade-off between cost and responsiveness.
TBS policy combines push and pull controls by replenishing at a constant rate from the offshore source and producing at the nearshore plant only when inventory is below a target. The constant base allocation allows the offshore facility to focus on cost efficiency, whereas the nearshore facility’s quick response capability is utilized only dynamically to guarantee high service. The research goals are to (i) determine the allocation of random demand into base and surge capacity, (ii) estimate corresponding working capital requirements, and (iii) identify and value the key drivers of dual sourcing. We present performance bounds on the optimal cost and prove that economic optimization brings the system into heavy traffic. We analyze the sourcing policy that is asymptotically optimal for high-volume systems and present a simple “square-root” formula that is insightful to answer our questions and sufficiently accurate for practice, as is demonstrated with a validation study.
Analyses of Advanced Iterated Tour Partitioning Heuristics for Generalized Vehicle Routing Problems
Researcher(s): Anupam Seth, Diego Klabjan, Placid M. Ferreira
Year: 2009
Theoretical analyses of a set of iterated-tour partitioning vehicle routing algorithms applicable to a wide variety of commonly-used vehicle routing problem variants are presented. We analyze the worst- case performance of the algorithms and establish tightness of the derived bounds. Among other variants we capture the cases of pick-up and delivery, and multiple depots. We also introduce brand new concepts such as mobile depots, partitioning of customer nodes into groups, and potential opportunistic under-utilization of vehicle capacity by only partially loading the vehicle, among others, which arise from a printed circuit board application. The problems studied are of critical importance in many practical applications.
Robust Airline Scheduling Under Block Time Uncertainty
Researcher(s): Milind Sohoni,
Yu-Ching Lee, Diego Klabjan
Year: 2008
Airline schedule development continues to remain one of the most challenging planning activity for any airline. An airline schedule comprises of a list of flights and specifies the origin, destination, scheduled departure, and arrival time of each flight in the airline's network. A critical component of the schedule development activity is the estimation of flight block-times, which depend on several factors. Many airlines estimate these block-times simply by using limited historical data, however, such techniques have not resulted in significantly improved on-time performance of the schedule during operations. Thus, from a passenger's perspective, the service level guarantee of an airline's network continues to be low. We first define two service level metrics for an airline schedule. The first one is similar to the on-time performance measure of the U.S. Department of Transportation and we define it as the flight service level. The second metric, called the network service level, is geared towards completion of passenger itineraries. We then develop a stochastic integer programming formulation that optimally perturbs a given schedule to maximize expected profit while ensuring the two service levels. We also develop a variant of this model that maximizes service levels while achieving desired network profitability. To solve these models we develop an efficient algorithm that guarantees optimality. Through extensive computational experiments, using real-world data, we demonstrate that our models and algorithms are efficient and achieve the desired trade-off between service level and profitability.
Fully Sequential Selection Procedures with Control Variates
Researcher(s): Shing Chih Tsai, Barry L. Nelson
Year: 2008
Fully sequential selection procedures have been developed in the field of stochastic simulation to find the simulated system with the best expected performance when the number of alternatives is finite. Kim and Nelson proposed the KN procedure to allow for unknown and unequal variances and the use of common random numbers. KN approximates the raw sum of differences between observations from two systems as a Brownian motion process with drift and uses a triangular continuation region to decide the stopping time of the selection process. In this paper we derive new fully sequential selection procedures that employ a more effective sum of differences which we called a controlled sum. Two provably valid procedures and an approximate procedure are described. Empirical results and a realistic illustration are provided to compare the efficiency of these procedures with other procedures that solve the same problem.
The Importance of Decoupling Recurrent and Disruption Risks in a Supply Chain
Researcher(s): Sunil Chopra,
Gilles Reinhardt, Usha Mohan
Year: 2007
This paper focuses on the importance of decoupling recurrent supply risk and disruption risk when planning appropriate mitigation strategies. We show that bundling the two uncertainties leads a manager to underutilize a reliable source while over utilizing a cheaper but less reliable supplier. As in Dada et al. (working paper, University of Illinois, Champaign, IL, 2003), we show that increasing quantity from a cheaper but less reliable source is an effective risk mitigation strategy if most of the supply risk growth comes from an increase in recurrent uncertainty. In contrast, we show that a firm should order more from a reliable source and less from a cheaper but less reliable source if most of the supply risk growth comes from an increase in disruption probability.
Spiffed-Up Channels: The Role of Spiffs in Hierarchical Selling Organizations
Researcher(s): Fabio Caldieraro, Anne Coughlan
Year: 2007
We study a channel relationship in which manufacturer(s) use independent sales representatives (rep firms), which employ salespeople to do the actual selling. We show that commission-only payments by manufacturers to rep firms lead to suboptimal outcomes for the manufacturer, relative to those obtained under a vertically integrated channel. From the manufacturer's standpoint, these inefficiencies can be ameliorated through the use of sales incentives given to the rep firm's salespeople directly by the manufacturer (called "spiffs"). In a monopolistic environment, spiffs are shown to improve the manufacturer's profits in the face of contractual restrictions on the channel members' ability to set separate commission rates by product. For certain types of restrictions, spiffs may generate manufacturer outcomes close to the fully coordinated ones achieved under vertical integration even when compensating the rep firm through commission-only contracts. In a competitive environment, spiffs are shown to be used by a powerful manufacturer that shares a rep firm's sales efforts with the product of a weaker manufacturer (i.e., in the case of "common agency"). In this case, spiffs are used as a strategy to deter the weaker manufacturer from challenging the stronger manufacturer for the sales force's valuable selling effort.
Flexibility and Complexity in Periodic Distribution Problems
Researcher(s): Peter Francis,
Karen Smilowitz, Michal Tzur
Year: 2006
In this paper, we explore trade-offs between operational flexibility and operational complexity in periodic distribution problems. We consider the gains from operational flexibility in terms of vehicle routing costs and customer service benefits, and the costs of operational complexity in terms of modeling, solution methods and implementation challenges for drivers and customers. The period vehicle routing problem (PVRP) is a variation of the classic vehicle routing problem in which delivery routes are constructed for a period of time; the PVRP with service choice (PVRP-SC) extends the PVRP to allow service (visit) frequency to become a decision of the model. For the periodic distribution problems represented by PVRP and PVRP-SC, we introduce operational flexibility levers and a set of quantitative measures to evaluate the trade-offs between flexibility and complexity. We develop a Tabu Search heuristic to incorporate a range of operational flexibility options. We analyze the potential value and the increased operational complexity of the flexibility levers.
The Period Vehicle Routing Problem with Service Choice
Researcher(s): Michal Tzur,
Peter Francis, Karen Smilowitz, Maciek Nowak, Tingting Jiang
Year: 2006
The period vehicle routing problem (PVRP) is a variation of the classic vehicle routing problem in which delivery routes are constructed for a period of time (for example, multiple days). In this paper, we consider a variation of the PVRP in which service frequency is a decision of the model. We refer to this problem as the PVRP with service choice (PVRP-SC). We explore modeling issues that arise when service choice is introduced, and suggest efficient solution methods. Contributions are made both in modeling this new variation of the PVRP and in introducing an exact solution method for the PVRP-SC. In addition, we propose a heuristic variation of the exact method to be used for larger problem instances. Computational tests show that adding service choice can improve system efficiency and customer service. We also present general insights on the impact of node distribution on the value of service choice.
An Efficient and Robust Design for Transshipment Networks
Researcher(s): Robert W. Lien, Seyed M.R. Iravani, Karen Smilowitz, Michal Tzur
Year: 2005
Transshipment, the sharing of inventory among parties at the same echelon level of a supply chain, can be used to reduce costs. The effectiveness of transshipment is in part determined by the configuration of the transshipment network. We introduce chain configurations in transshipment settings, where every party is linked in one connected loop. Under simplifying assumptions we show analytically that the chain configuration is superior to configurations suggested in the literature. In addition, we demonstrate the efficiency and robustness of chain configurations for more general scenarios and provide managerial insights regarding preferred configurations for different problem parameters.
Cost Modeling and Design Techniques for Integrated Package Distribution Systems
Researcher(s): Karen R. Smilowitz, Carlos F. Daganzo
Year: 2005
Complex package distribution systems are designed using idealizations of network geometries, operating costs, demand and customer distributions, and routing patterns. The goal is to find simple, yet realistic, guidelines to design and operate a network integrated both by transportation mode and service level; i.e., overnight (express) and longer (deferred) deadlines. The decision variables and parameters that define the problem are presented along with the models to approximate total operating cost. The design problem is then reduced to a series of optimization subproblems that can be solved easily. The proposed approach provides valuable insight for the design and operation of integrated package distribution systems. Qualitative conclusions suggest that benefits of integration are greater when deferred demand exceeds express demand. This insight helps to explain the different business strategies of package delivery firms today.
Managing Risk to Avoid Supply Chain Breakdown
Researcher(s): Sunil Chopra, ManMohan S. Sodhi
Year: 2004
By understanding the variety and interconnectedness of supply-chain risks, managers can tailor balanced, effective risk-reduction strategies for their companies.
Supply Chain Design with Emission Considerations
Researcher(s): Diego Klabjan
The Arctic Science Division of NSF manages scientific research activities in the Arctic region. Summit in Greenland is one of the most active sites under the jurisdiction of NSF. Due to its remoteness, the operations at the site are challenging and costly. On top of the actual operations, the logistics services pose unique challenges including limited transportation options and frequency, high fuel costs, and time constrained access. Moving forward, in operating the site NSF faces several uncertainties such as the unpredictable span of future research activities, logistics costs, and environmental consequences. In collaboration with NSF and service providers a holistic operations model will be developed. The model will rigorously capture the peculiar aspects of operating within the Arctic Circle and the inherent uncertainties. The model solutions will provide a strategic plan to NSF to assist in future handling of Summit. The solutions will specify logistic operations, and acceptable energy options – renewable energy deployments – and their emission implications. The decision makers at NSF will be equipped with a tool capable of estimating the total future costs of operating the site and the underlying operations strategies. Key performance indicators will also by analyzed and aligned with the model’s objectives. The solutions will explicitly take into account the uncertainties thus providing robust plans.
Designing the Distribution Network in a Supply Chain
Researcher(s): Sunil Chopra
Year: 2003
This paper describes a framework for designing the distribution network in a supply chain. Various factors influencing the choice of distribution network are described. We then discuss different choices of distribution networks and their relative strengths and weaknesses. The paper concludes by identifying distribution networks that are best suited for a variety of customer and product characteristics.