Energy Operations
Decision horizons · system layers · eleven models
Energy operations research asks how a grid operator, utility, market participant, or energy planner can keep the lights on — reliably, affordably, and increasingly, without carbon — across horizons that stretch from a multi-decade capacity-investment plan to a sub-second frequency-regulation response. This section presents eleven canonical OR problems, each as a live interactive solver grounded in a real power-systems or electricity-market decision, organized along the decision-horizon × energy-system-layer taxonomy of Wood, Wollenberg & Sheblé (2013) with modern extensions for renewables, storage, and sector coupling.
Why energy OR matters
Scale of the problem · three anchor statistics
Decision framework
Four lenses on the same eleven applications
The canonical taxonomy of Wood, Wollenberg & Sheblé (2013) organizes power-systems decisions along two axes: the decision horizon (long-term planning, medium-term, short-term operations, real-time) and the energy-system layer (supply, transmission, distribution, demand, storage, markets). Every application in this section occupies one cell. Dashed cells are honest gaps — decisions that exist in practice but are not yet modelled here.
The OR-research grouping reorganizes the same eleven applications by the mathematical problem family they belong to. Grid operators see this as the table of contents of Wood & Wollenberg or Conejo et al. (2016): unit commitment, economic dispatch, and optimal power flow form the operations triad; expansion planning and storage cover long-horizon investment; markets and demand-side close the loop.
Modern energy OR is defined by how it handles uncertainty: wind and solar forecast errors, demand variability, equipment failures, fuel-price swings. This lens cross-tabulates the same applications by decision horizon and uncertainty-treatment method, following Morales, Conejo, Madsen, Pinson & Zugno (2014) and the stochastic-UC review of Zheng, Wang & Liu (2015). Most cells are sparse today — the field is actively moving from deterministic toward stochastic, robust, and data-driven formulations.
The energy transition — decarbonizing electricity, integrating variable renewables, coupling sectors (electricity + heat + gas + hydrogen), and evolving market design — is the defining story of twenty-first-century energy OR. This lens groups the same eleven applications by which transition pillar they advance, following the framings of Pfenninger, Hawkes & Keirstead (2014) and Mancarella (2014) on multi-energy systems.
Application catalog
All seventeen pages · click a card to open the interactive solver
Decision timescales
The multi-scale view · complementary to the matrix above
The same eleven applications, arrayed on a logarithmic time axis from multi-decade capacity planning on the left to sub-second frequency response on the right. Power-systems operations span nine orders of magnitude in time; the same concept (commit-and-dispatch) appears in different forms at different scales. Color by energy-system layer.
Current research frontiers
Where energy OR is actively evolving
Uncertainty at scale for renewable-heavy grids
Stochastic, robust, and distributionally-robust formulations for UC and OPF that scale to real system sizes, paired with scenario-reduction and decomposition methods. Still an open problem for continental interconnections at minute resolution under high wind and solar penetration.
Convex relaxations and tight formulations for AC-OPF
Second-order-cone and semidefinite relaxations (Low 2014, Molzahn & Hiskens 2019) can give provably optimal solutions on a growing class of networks; tighter MILP formulations for UC (Morales-España et al. 2013) have collapsed branch-and-bound times by orders of magnitude.
Sector coupling and multi-energy optimization
Joint optimization of electricity, heat, gas, hydrogen, and transport networks (Mancarella 2014, Lund et al. smart-energy-systems programme). Hydrogen supply-chain design is opening as a full sub-field, with 2020+ reviews building on Almansoori & Shah (2006).
Machine learning and RL for grid operations
Data-driven surrogates for AC-OPF and UC, safe reinforcement learning for real-time dispatch, and physics-informed neural networks for contingency screening. An active but still operationally immature frontier — safety guarantees and interpretability remain core open questions.
Key references
Cited above · DOIs & permanent URLs