More details to come...
Figure 1. Flow of multi-step solution procedure
Figure 2. Computer application of proposed solution procedure
Considered "aircraft routing and scheduling problem" in a collaboration with a local division of a major express carrier.
This business problem refers to a service network design problem on a hierarchical hub-and-spoke network in the literature.
The aim is to generate daily schedules for feeders operated in the Caribbean.
A multi-phase solution procedure is proposed for the problem including a route generation and elimination algorithm, path-based MIP formulation, an algorithm to resolve departure-time conflicts between routes, and extensive data processing parts (see Fig. 1).
All components of the proposed procedure are integrated in a computer application (decision support system) along with additional features (see Figure 2).
The results show that the carrier can save from 0.91% to 13.45% in total operating cost as well as from 36 minutes to 9.5 hours in total flight time, and the company can decrease the number of feeders in the fleet as many as three aircraft under a variety of scenarios.
Ongoing part:
The problem is extended through incorporating allocation and location decisions for some layers of the service network.
Illustration of "2-weeks anomaly" in commodity flows.
Considered a generic service network design problem (SNDP) for freight carriers with an assumption that has not been addresses well in the literature.
Proposed a set of actions for a freight carrier to scale its capacity.
Developed an arc-based MIP formulation as well as valid inequalities to enhance the capability of the formulations.
Observed "2-weeks anomaly" in solutions obtained by the formulation (extended based on the formulations presented in the literature) and resolved through adding new set of constraints (see the figure above).
Achieved to save 50% and 12% of computational time for small and medium size instances by adding valid cuts.
Proposed a custom multi-phase dedicate-and-mix algorithm (DAMA) to solve practical instances of the problem.
Submitted to a journal.
Considered a hub location and hub network design problem (HLP) in a multi-period setting as an extension of classical hub location problems.
Discussed both the single allocation and the multiple allocation versions and proposed a MIP formulation for each case.
Proposed a set of valid inequalities to enhance the corresponding models.
The value of the multi-period modeling is quantified by comparing with a static counterpart.
Published in Annals of Operations Research in 2016. Access to read-only version of this paper here or regular pdf version with academic access here.
A rapid increase in solid waste volumes combined with variations in handling and constituents by municipality have increased the challenges associated with Solid Waste Management (SWM) due to potentially higher risks to human health and safety. To identify and evaluate the direct and indirect hazards among SWW, potential safety concerns need to be investigated. This study presents a classification and regression tree (CART) model to evaluate the occupational health risks among the municipal SWW in Florida. CART is a powerful analytic tool used to assess the significance of multiple predictors/causes in a specific dataset on a target/effect. In this study, the waste workers’ compensation (WWC) dataset collected by multiple insurance companies between 1993 and 2013 in Florida with more than 10,000 data points was used to implement the CART model. We first selected the applicable Standard Industrial Code (SICs) of major SWM methods such as refuse collection, recycling, incineration, landfilling, and composting. Then, we identified the major predictors in the WWC such as season, date and time, injury type, affected body part, county, gender of injured person, and business type. Next, we categorized the predictors in numerical and categorical classes and determined the main sub-classes for each class. Lastly, the developed CART model was used to establish the relationship between these categorized predictors and target variables. The CART model provides a powerful tool for waste management stakeholders to identify the most relevant predictors and forecast injury type and affected body part among solid waste workers based on different scenarios.