Home/Method notes/The modelled revenue stack

Method note · scenario

The modelled revenue stack

What it measures

What a one-megawatt battery could earn per month from day-ahead arbitrage and from holding capacity in reserve, at modelled reserve prices.

Inputs

  • The daily two-hour spread (TB2) for Germany and Luxembourg on the canonical hourly grid, from the day-ahead price
  • A simulated monthly price for each reserve product (FCR, aFRR upward, aFRR downward), generated from the fixed parameters in pipeline/pilot_reserve_sim.json
  • The shares of the battery's megawatt given to each use, from pipeline/revenue_stack.json

Method and weighting

In every time slice the modelled battery splits its one megawatt in fixed shares between day-ahead arbitrage, FCR (primary frequency reserve) and aFRR (automatic frequency reserve). Each share is the weight of a use divided by the sum of the weights, so the shares add up to the whole megawatt. The weights are in pipeline/revenue_stack.json and the view states the resulting shares from the data file.

The arbitrage layer for a month is the arbitrage share times the sum over its days of the daily two-hour spread times the battery's duration in hours, per megawatt of battery power. It uses one cycle a day and no round-trip losses, so it is the upper bound described in the perfect-foresight arbitrage bound note.

A reserve layer for a month is the product's share times its simulated price per megawatt and hour of availability times the hours of that month in Berlin time, so a month with a clock change carries one hour more or fewer. FCR is symmetric and earns one price. The megawatt held for aFRR earns the upward and the downward price, kept as two separate layers.

The reserve prices are simulated and drawn from no market data. Each product's monthly price is a base level, plus a seasonal swing over the calendar year, plus a bounded random term from a fixed seed, floored at a minimum; the parameters are in pipeline/pilot_reserve_sim.json. Every reserve offer is assumed accepted at the month's mean price. A month enters the chart only when every calendar day has a spread and every reserve price exists for it.

Window

Every complete month from the first month of the series to the last complete month, refreshed with the data; the headline figure covers the last twelve of them, or fewer when fewer exist. The current window runs from to . Data last refreshed on .

Known limits

  • The reserve prices come from a simulation with fixed parameters and from no market data, so every reserve figure is modelled.
  • The arbitrage layer is an upper bound: it assumes perfect foresight, no round-trip losses and no effect of the battery on the price.
  • Payment for energy delivered when reserve is called is not modelled; only the payment for holding capacity is.
  • The configured shares are a neutral placeholder for an operating strategy. The model covers Germany and Luxembourg only.

Used on

Sources: ENTSO-E. Each is described on the Data sources page.