Rolling Horizon Implementation of the Replenishment at Sea Planner
Abstract
To sustain ships on station and create a forward presence overseas, the United States Navy relies on replenishment at sea (RAS) operations. Using these events, shuttle ships provide fuel and materiel to deployed combatants while underway. With a global commitment and dynamic schedules, the Navy faces a complex, operational logistics problem. A mixed integer linear program called the Replenishment at Sea Planner (RASP) provides the optimal schedule; however, schedulers use a heuristic to develop less-than-optimal RAS plans due to RASP's computation time. Ill-planned schedules limit the number of combatants in theater, reduce time-on-station for deployers, and increase costs due to shuttles consuming more fuel. The purpose of this research is to implement a rolling horizon algorithm with future relaxation into the current version of RASP. A rolling horizon partitions the time horizon into subproblems then optimizes each subproblem sequentially using a portion of the previous solution. The analysis examines both computation times and objective values across four scenarios, comparing these results to the monolith model. We then apply the rolling horizon as a heuristic for warm starting the monolith model. Compared to the monoliths performance with a two-hour computation limit, the rolling horizon implementation can provide operational quality replenishment plans, improving theaterwide sustainment and readiness at a cost savings of $15-22 million over a sixty-day period.
Document Details
- Document Type
- Technical Report
- Publication Date
- Sep 01, 2023
- Accession Number
- AD1224377
Entities
People
- Daniel S. Wright
Organizations
- Naval Postgraduate School