Tactical Decisions for Contested Logistics: Performance, Robustness, and Resilience
Abstract
Approved for Public Release The proposed work will develop a tactical logistics planning tool that quantifies through stress-testing the robustness and resilience of current and candidate plans for operating a last-mile distribution system in support of an expeditionary force. The tool also recommends optimized plans that balance robustness and resilience against factors such as economy and performance. The tool will provide analysts with the capability of exploring and assessing plans for complex, joint operations with relative ease. It will answer questions such as: How much disruption can a last-mile logistical plan absorb without delaying ongoing and future operations? How will different concepts for the hand-off from an intra-theater distribution system to the last-mile distribution system affect efficiency, robustness, and resilience of the overall logistical system? The tool will produce improved logistical plans offering commanders added flexibility, especially in scenarios with thin margins. We envision the tool being used continuously, generating and assessing tactical plans for a rolling horizon of seven days. The tool seeks to construct suitable capability sets through optimized allocation of connector vehicles with the goal of aligning capabilities and aspirations. We will develop the mathematical models and algorithms underpinning the tool as well as a prototype programmed in Python using Pyomo and other packages,and distributed per ONR guidelines via GitHub. The technical and methodological challenges center on the difficulty of representinguncertainty within a large-scale decision model in a dynamic environment. The proposed work will rely on innovations in topologicaldata analysis as well as variational analysis to overcome technical and methodological challenges. In particular, we will leverage subdifferential calculus to estimate the effect of changes in parameter values on minimum and maximum objective function values. We will consider reformulations based on an event-driven timeline, surrogate models, as well as computational procedures stemming from state-of-the art topological arguments. We will also develop dual methods for quickly solving large-scale, nonlinear optimization problems while retaining the benefits of the Gurobi solver.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Apr 11, 2024
- Source ID
- N000142412277
Entities
People
- John Gunnar Carlsson
Organizations
- Office of Naval Research
- United States Navy
- University of Southern California