Skip to content

What is Energy Operations Research?

Foundation

Operations research (OR) is the discipline of making better decisions with mathematics. It takes a messy operational question ("which turbine should we repair first?"), writes down precisely what can be decided, what we want, and what we are not allowed to do, and then finds the best permitted answer.

Energy operations research applies that discipline to the daily decisions of a renewable-energy business:

  • How much of each farm's output should we offer, at which price bands?
  • When should the battery charge, hold, discharge or hold FCAS capacity?
  • Which week, and which hours, should a crane-dependent gearbox exchange happen?
  • Which turbine gets the one spare gearbox and the only specialist crew?
  • What is another megawatt of network capacity worth, right now?

None of these is a purely mathematical question. Each sits on top of physics (power, energy, wind, heat, wear), markets (dispatch, prices, constraints), contracts (PPAs, LGCs, hedges), and people and equipment (crews, cranes, spares, lead times). The value of OR is that it forces all of these into one explicit, inspectable model.

The loop this book teaches

flowchart LR
    A[Real asset] --> B[Operational problem]
    B --> C[Data]
    C --> D[Mathematical model]
    D --> E[Optimisation]
    E --> F[Visual explanation]
    F --> G[Interpretation in MW, MWh, $]
    G --> H[Backtest]
    H --> I[More realistic constraints]
    I --> J[Production implementation]
    J -.-> B

Every chapter walks round this loop at least once. Early chapters stop at interpretation; later chapters close it with backtests and running services.

Forecast, optimise, backtest

Three questions recur throughout the book, and keeping them separate avoids a lot of confused modelling:

Question Discipline Example
What might happen? Forecasting Price, wind, constraint binding, failure
Given that, what should we do? Optimisation Bid, dispatch, maintain, allocate
Would that decision have helped? Backtesting Actual vs optimised, with only information available at the time

A forecast with a better error metric does not automatically produce better decisions. Measuring decision value as well as forecast accuracy is one of the threads that runs through the whole book.

From one asset to a portfolio

flowchart LR
    A[Asset] --> S[Site] --> M[Multi-asset] --> F[Fleet] --> P[Portfolio orchestration]

The book starts with a single decision for a single interval and grows, chapter by chapter, towards the capstone question: given the next several days of prices, weather, network constraints, equipment condition, spares, crews, cranes, battery state and contract positions, which permissible combination of trading, dispatch, maintenance and logistics decisions best meets the portfolio's objectives? The point is not to solve that in one giant model. It is to understand how it decomposes into smaller problems you can solve, test and trust.

Boundaries

Optimisation in this book always happens inside the permitted operating envelope. No model may recommend violating engineering or OEM limits, safety requirements, generator performance standards, market rules or contractual obligations. Where a model approximates an AEMO process, it says so explicitly. Nothing here claims to reproduce NEMDE.

Next: How this book works.