TB3 - Optimization and Applications in Energy Systems 2

Low Earth orbit (LEO) constellations offer a promising solution for extending broadband connectivity to remote regions. The rapid motion of LEO satellites introduces critical challenges, including frequent user–satellite handovers and power-inefficient operations. This paper addresses handover management within a quasi-Earth-fixed cell architecture by proposing an optimization framework designed to jointly minimize satellite consumed transmit power and handover frequency. We formulate a cluster-level model that jointly optimizes cluster–beam association, frequency, and power allocation while satisfying rate constraints. The formulation yields a mixedinteger nonlinear program, which is computationally challenging. To obtain a tractable solution, we first relax the binary association variables to continuous ones, and then address the nonconvex couplings using a successive convex approximation procedure. We further introduce a generalized handover-aware association preference (HAAP) score that integrates expected satellite visibility with the duration of the current cluster–satellite association. Accordingly, the weights at both the cluster and link levels are adaptively adjusted based on the association age and visibility duration. The efficacy of the proposed algorithm is demonstrated through simulations against conventional maximum visibility (CMV) and minimum power consumption (CMPC) baselines. The proposed scheme achieves a 62% reduction in power consumption compared to CMV, with only a 5% increase in handovers. It also reduces the number of handovers by approximately 75% relative to CMPC, at the cost of about 19% higher power consumption.

In this work, we present an optimization model for shipboard power systems that simultaneously determines the vessel’s trajectory, speed profile, and power management strategy. Our approach integrates solar photovoltaics, shore power, batteries, and diesel generators while accounting for power balance, propulsion dynamics, and itinerary. To optimize both path and speed while avoiding obstacles, e.g., land, shallow waters, or protected areas, the non-convex navigable space is partitioned into a set of convex, obstacle-free zones, which can then be embedded in the optimization problem using disjunctive constraints. The convex zones also serve to evaluate the impact of spatially varying meteorological conditions, such as irradiance, winds, currents and waves on energy efficiency. The resulting problem is formulated as a mixed-integer convex program (MICP) and can be solved to global optimality using off-the-shelf commercial solvers, while accounting for all operational and navigation decisions. Numerical results on a Saint Lawrence River case study using the S175 ship model are presented to illustrate the benefits of the model.

A scalable residential EMS framework integrating diverse building archetypes and stochastic occupant behaviors controls HVAC, hot water, batteries, and EV charging. Four strategies—RBC, MPC, standard RL, and lexicographic RL—are benchmarked. Multi-household scaling enables multi-agent optimization, evaluating centralized and decentralized architectures across comfort, cost, and grid trade-offs.

We present a two-stage stochastic programming model for scheduling hydropower in the Nordic power system, co-optimizing energy dispatch, reserve procurement, and the allocation of transmission capacity between day-ahead and balancing markets. The second stage explicitly captures reserve activation consequences in cascaded watercourses, bridging multi-area transmission models and detailed hydropower representations.