event

PhD Defense by Idil Arsik

Primary tabs

You are cordially invited to attend my thesis defense.

 

Thesis Title: Investigations into effectively moving people and goods

Advisor: Dr. Martin Savelsbergh

 

Committee members:

Dr. Alan Erera

Dr. Natashia Boland

Dr. Alejandro Toriello

Dr. Mauricio G. C. Resende (Amazon.com)

 

Date and Time:  5:00-6:30 pm EST, Thursday, May 14th, 2020

 

Meeting URL

https://bluejeans.com/489424138

 

Meeting ID

489 424 138

Abstract:

 

In this thesis, we investigate practical methods to move people and goods effectively on the road network. In the first part of this thesis, we focus on route guidance for the movement of people  while in the second part we focus on the movement of goods by investigating the two main aspects of service network design:  flow and resource planning.

 

Chapter 2 introduces a centralized proactive route guidance approach motivated by the anticipated introduction of autonomous vehicles which, with full adoption, can create an environment  in which a specific origin-destination path can be assigned to each self-driving vehicle and in which that vehicle will follow the assigned path. Our approach integrates a system perspective, i.e., minimizing congestion, and a user perspective, i.e., minimizing inconvenience. As a design choice, we only solve linear program which is more likely to scale well and be of practical use. The linear structure of our models allows us to derive theoretical properties.  In particular, we show that for the problem of minimizing maximum arc utilization,which is used as a measure of congestion in a road network, results analogous to those well-known for the maximum flow problem, e.g., the max flow-min cut theorem, can be derived.

 

In Chapter 3, we focus on cost-effective routing of commodities on a line-haul network from its origin to its destination while meeting tight service requirements, satisfying operational constraints and minimizing transportation costs. We introduce a marginal-cost path based greedy heuristic that works with a partially time-expanded network to solve large scale real-life instances found in practice. Our approach involves two consolidation improvement heuristics and novel use of iterative refinement within the greedy heuristic to obtain a continuous-time feasible service network design.

 

In Chapter 4, we analyze the value of outsourcing transportation for different negotiated prices with contractors and, while doing so, explicitly account for driver considerations.  We introduce a depth-first search algorithm to generate a set of time-feasible cycles of chosen length in terms of the number of dispatches that covers a set of planned dispatches in a given load plan, where dispatches can be connected by empty travel. When generating cycles, we respect the company specific rules and  hours-of-service regulations that ensure road safety and prevent fatigue related accidents. We solve an integer programming model that identifies a subset of company cycles (and contractor cycles and one-way moves if the outsourcing option is available)  that maximizes the cost savings over the (unrealistic) scenario in which company drivers perform a one-way move and return empty. When a company only performs out-and-back cycles, as is current practice at SF Express, one of the largest Chinese small parcel carriers, we efficiently choose the set of cycles by solving bipartite matching problem for each out-and-back lane pair separately.

 

Status

  • Workflow Status:Published
  • Created By:Tatianna Richardson
  • Created:05/07/2020
  • Modified By:Tatianna Richardson
  • Modified:05/07/2020

Categories

Keywords