Data on supply chains is often sparse due to reluctance among actors to share their data, making supply chain simulation modeling difficult. As a result, supply chain simulation models suffer from parametric and structural uncertainties, and there is a large variety of plausible simulation models that would align with the sparse observations about the real-world supply chain. Constructing a diverse set of models that fit sparse data is not an easy task. A relatively unknown approach to generating this diverse set of plausible models is the Quality Diversity (QD) algorithm. This study evaluates the feasibility of using QD to generate a diverse ensemble of supply chain simulation models for a varying degree of data sparseness. The results show that QD is able to generate a diverse ensemble of supply chain models, including the ground truth. As expected, QD successfully identifies the structure of the ground truth most frequently for a low level of data sparseness. When the sparseness of the data increases, QD is prone to overfitting, identifying supply chain structures that are more complex than the ground truth. Further research should focus on reviewing the calibration metric for sparse data, to reduce the overfitting of complex network structures.

Illicit supply chains for products like counterfeit Personal Protective Equipment (PPE) are characterized by sparse data and great uncertainty about the operational and logistical structure, making criminal activities largely invisible to law enforcement and challenging to intervene in. Simulation is a way to get insight into the behavior of complex systems, using calibration to tune model parameters to match its real-world counterpart. Calibration methods for simulation models of illicit supply chains should work with sparse data, while also tuning the structure of the simulation model. Thus, this study addresses the question: “To what extent can various model calibration techniques reconstruct the underlying structure of an illicit supply chain when varying the degree of data sparseness?” We evaluate the quality-of-fit of a reference technique, Powell's Method, and three model calibration techniques that have shown promise for sparse data: Approximate Bayesian Computing, Bayesian Optimization, and Genetic Algorithms. For this, we use a simulation model of a stylized counterfeit PPE supply chain as ground truth. We extract data from this ground truth and systematically vary its sparseness. We parameterize structural uncertainty using System Entity Structure. The results demonstrate that Bayesian Optimization and Genetic Algorithms are suitable for reconstructing the underlying structure of an illicit supply chain for a varying degree of data sparseness. Both techniques identify a diverse set of optimal solutions that fit with the sparse data. For a comprehensive understanding of illicit supply chain structures, we propose to combine the results of the two techniques. Future research should focus on developing a combined algorithm and incorporating solution diversity.

Supply chain visibility concerns the ability to track parts, components, or products in transit from supplier to customer. The data that organizations can obtain to establish or improve supply chain visibility is often sparse. This paper presents a classification of the dimensions of data sparseness and quantitatively explores the impact of these dimensions on supply chain visibility. Based on a review of supply chain visibility and data quality literature, this study proposes to characterize data sparseness as a lack of data quality across the entire supply chain, where data sparseness can be classified into three dimensions: noise, bias, and missing values. The quantitative analysis relies on a stylized simulation model of a moderately complex illicit supply chain. Scenarios are used to evaluate the combined effect of the individual dimensions from actors with different perspectives in the supply chain, either supply or demand-oriented. Results show that when a data sparseness of 90% is applied, supply chain visibility reduces to 52% for noise, to 65% for bias, and to 32% for missing values. The scenarios also show that companies with a supply-oriented view typically have a higher supply chain visibility than those with a demand-oriented view. The classification and assessment offer valuable insights for improving data quality and for enhancing supply chain visibility.

Schedule design in the transportation and logistics sector is a widely studied problem. Transport service providers, such as the train industry and aviation, aim for schedules to be on-time according to the planning (i.e., on-time performance or OTP) in order to increase the service level by ensuring that passengers actually make their connections and to reduce costs. Transportation services also aim for schedules that serve a high variety of destinations and frequency of connections (i.e., connectivity). OTP and connectivity are both highly dependent on buffer time: more lucrative connections can often be offered by reducing the buffer time in the schedule, while more delay can be absorbed by more buffer time. Given strict constraints on the minimum turnaround time of aircraft and minimum (and maximum acceptable) transfer times of passengers, assigning buffer time in an already tightly planned schedule to optimize OTP and connectivity simultaneously is a big challenge. This research presents a novel multi-objective formulation of a daily flight schedule where buffer scheduling is used to ensure the optimal balance between OTP of the schedule and the passenger connections as connectivity, given the tight restrictions. This problem formulation is solved using a simulation–optimization framework. Specifically, we use the Multi-Objective Evolutionary Algorithm (MOEA) BORG. As a proof of concept, a daily European flight schedule of a large international airline is optimized on both OTP and connectivity. The results demonstrate that the presented multi-objective formulation and associated solving through simulation–optimization can result in candidate schedules with both better on-time performance and a higher connectivity.

The explosive growth of e-commerce creates a need for increasingly responsive omni-channel fulfillment capabilities, which raises new challenges in inventory management and order fulfillment for retailers. In response to these challenges, many retailers attempt to establish so-called ship-from-store concepts, which leverage their physical store networks to fulfill online orders. In this study, we analyze the optimal setup of these in-store fulfillment processes of online orders for an omni-channel retailer. We use a simulation-based approach combined with exploratory modeling to prescribe optimal fulfillment policies under a variety of sources of uncertainty. We apply our proposed model to a case study informed by real data from a leading sports fashion retailer in New York City in order to illustrate the practical applicability and value of our approach. Our results determine (i) the optimal amount of time to allow for batching of online orders prior to starting the in-store picking process; (ii) the optimal amount of time to allow for readily picked orders prior to starting the delivery process; (iii) the optimal number of pickers; and (iv) the optimal number of packers, and the related performance measures. Finally, we build on our analysis results to derive a set of managerial implications applicable to many omni-channel problems.