Abstract
This study addresses freight revenue management in the container shipping industry, a domain characterized by unique modeling challenges arising from its complex operational structure and multiple sources of uncertainty. Building on recent advances in subgradient optimization, we develop a scalable stochastic model that captures both long-term and short-term decision elements, including dynamic pricing, overbooking, container allocation, and empty container repositioning. Our approach extends the Lagrangian relaxation framework previously applied to air cargo to sea freight, incorporating interspersed period types, hub and weight constraints, and scalability via convex non-smooth optimization. In addition, we design two large-scale algorithms that optimize pricing and container management in both spot and allotment market stages. Extensive numerical experiments demonstrate the practical potential of our approach: algorithms recommend repositioning empty containers with volumes nearly double those traded in the spot market, revealing substantial untapped revenue opportunities through integrated optimization.