How it works
A settlement block divides the day into equal intervals. Fifteen-minute blocks give 96 in a day, and five-minute intervals give 288. Every schedule is written block by block, as a power level in MW held through the block. The meter records the energy delivered in each block, in MWh or kWh. The settlement system then compares that energy with the scheduled energy for the same block.
Energy in a block is power multiplied by the block's length, so 40 MW held for fifteen minutes is 10 MWh. Prices, deviations and charges are worked out separately for each block and then summed over the day or the billing period. A surplus in one block does not cancel a shortfall in the next. Each block has its own price, and the grid had to be balanced at that moment. Shorter blocks follow the physics of the grid more closely, but they need finer forecasting, metering and bidding.
Why it matters
The block length sets the resolution at which a project must be modelled and run. Solar and wind output, battery cycles and deviation charges all need to be simulated per block, because hourly averages hide the swings that cost money. A forecast that is right for the hour can still miss in every quarter of it. For storage, shorter blocks create more price spreads to trade, which changes what a battery is worth.
A worked example
A wind farm is scheduled at 60 MW for the hour from 14:00, in four fifteen-minute blocks. Actual output is 66, 58, 52 and 64 MW. The deviations are +6, −2, −8 and +4 MW, or +1.5, −0.5, −2.0 and +1.0 MWh at 0.25 h per block. Over the hour they net to zero. Settled per block, the farm still has 2.5 MWh of surplus and 2.5 MWh of shortfall, each priced in its own block.
Illustrative figures.
Across markets
India schedules, trades and settles deviations on fifteen-minute blocks. Europe has moved imbalance settlement, and its day-ahead and intraday markets, to fifteen minutes in most countries. US markets dispatch and price in real time on five-minute intervals, while day-ahead markets are usually priced by the hour. A model built for one region's block length has to be resampled before it can be used in another.