Five-minute dispatch and price formation

Take three independent, lossless tranches: 100 MW at A$20/MWh, 150 MW at A$50/MWh and 100 MW at A$90/MWh. Demand of 200 MW uses 100 MW of the first and 100 MW of the second. Offer cost is A$7,000 per hour, or A$583.33 for a constant five-minute interval. The marginal price is A$50/MWh while the second tranche has spare capacity.

At 280 MW, the first two tranches supply 250 MW and the third supplies 30 MW. Offer cost is A$12,200 per hour and the marginal price becomes A$90/MWh. Do not average offers to infer the marginal price. Energy settlement at a uniform price is a different quantity from total offer cost.

If the second tranche can reach only 120 MW because of availability or ramping, recalculate before pricing: at 280 MW the third must provide 60 MW. A unit initially at 40 MW with a 3 MW/min upward ramp can reach 55 MW after five minutes, even if its nameplate is 100 MW. An end-of-interval target is not necessarily the average output used for MWh.

Use two units in a single lossless region. A has 100 MW capacity, energy offer A$20/MWh and raise-reserve offer A$5/MW/h. B has 150 MW capacity and offers energy at A$80/MWh. Only A can provide the required reserve in this deliberately simplified model.

With demand 160 MW and no reserve, A=100 MW and B=60 MW; total hourly offer cost is A$6,800. Requiring 20 MW reserve imposes A energy + A reserve ≤100 MW. The least-cost feasible result is A energy=80 MW, B energy=80 MW and A reserve=20 MW. Hourly offer cost is 80×20 + 80×80 + 20×5 = A$8,100; the five-minute cost is A$675.

Adding 1 MW reserve displaces 1 MW of A energy with B energy, so the incremental hourly cost is 80−20+5=A$65, not A$5. Adding 1 MW demand costs A$80 per hour here. Actual FCAS co-optimisation has multiple services, eligible providers, response requirements and trapezium constraints; do not treat this exercise as the production solver.

A physical teaching delivery ratio of 0.95 means 100 MW generated by A delivers 95 MW at the load. An A$20/MWh generator offer then costs 20/0.95=A$21.05 per delivered MWh in this constant-loss model. This is not an AEMO marginal loss factor or a settlement instruction.

Suppose a transfer limit prevents a cheap generator from increasing its delivered output. A higher-cost local unit may then supply the next MW even while cheap capacity remains available elsewhere. Regional price separation can be consistent with binding transfers and losses, but a spread alone does not identify a particular binding constraint.

A constraint A + 0.5B ≤140 is tight at A=100, B=80. Its marginal value measures the objective change from relaxing its right-hand side under the model's conventions. Tightness alone does not establish a positive marginal value, especially with redundant constraints or degeneracy. Read coefficients, direction and run status before assigning causation.

For 1 July 2026 to 30 June 2027, the AEMC publishes a market price cap of A$23,200/MWh and cumulative price threshold of A$2,225,900/MWh. The threshold is a cumulative price statistic under the rules, not a participant's revenue, a bill or an amount of energy. Attach the applicable year to both values.

The administered price cap is A$600/MWh under the relevant staged market-price-settings rule. Administered pricing responds to sustained price stress; it is not equivalent to applying the market price cap to a single extreme interval. The timing, calculation, associated floors and service treatment must come from the current rules and AEMO process.

A model must distinguish the unconstrained result of a teaching optimisation from a published price after applicable pricing arrangements. Do not silently clip an infeasible model to the price cap and call it a feasible dispatch. A spreadsheet recreating the cumulative trigger needs a complete, correctly classified rolling history and the effective rule version.

System operations may require intervention when normal market outcomes cannot meet the relevant security or reliability need. Directions and reserve arrangements have their own triggers and procedures. An instruction to change physical operation is not, by itself, a rule for replacing the settlement price.

Where intervention pricing applies, the methodology uses a pricing calculation intended to represent conditions without the intervention. The physical dispatch run and the pricing run can therefore answer different questions. Not every direction automatically leads to intervention pricing; check the methodology and the actual event.

In an event pack, keep run identifiers, intervention flags, relevant notices and final price records together. Explain which run produced the physical target and which produced the price used in your calculation. RERT is a reserve-procurement/intervention mechanism, not just another ordinary energy offer in the toy stack.

A price-setter report can contain multiple contributing bands and adjustment terms. The historical NEMDE Queue guide explains marginal band cost contributions; it is not a promise that every interval has one identifiable price-setting generator. Losses, constraints and co-optimisation can link products and regions.

First match the dispatch interval and the appropriate run. Then inspect the contribution fields, units and sign conventions against the schema for that release. Reconcile the reported contributions to the corresponding regional/service price and investigate residuals rather than forcing a single generator explanation.

Participant solver-file access and public historical archives are different access routes. The local CSV lab below reads published regional prices, not price-setter XML. That first dataset cannot establish who set the price. Request matching offers, constraint outputs, interconnector records and solver evidence before making that stronger claim.

Download a public AEMO DispatchIS report, retain its filename and original URL, and import its extracted CSV in Apply it. MMS data contains control, header and data records. Identify DISPATCH/PRICE by table name and map columns from its I-header rather than memorising column numbers.

Select one matching interval and preserve the intervention flag. Compare VIC1 and TAS1 if both are available. A non-zero spread is a useful question about losses, transfers and constraints, not a conclusion about Basslink. Missing regions or duplicate interval-region-run keys must be resolved before aggregation.

For a hypothetical constant 10 MW metered load, energy in five minutes is 10/12 MWh. Multiply that illustrative quantity by the imported RRP and label the boundary. The price is from your selected file; the load is invented. It is neither an actual customer's bill nor a generator's realised revenue.

The AEMC's Calculating the cumulative price decision was made on 23 April 2026, but its commencement is 1 November 2028. At the 18 October 2026 source check, that future methodology must not be used as the current trigger calculation. Store both the decision date and commencement date.

When a related briefing changes, ask which lesson claim, worked example, simulator assumption, question and source record it affects. Update the teaching material and its test together; a news link alone is not a course revision. Price limits need their own annual review independently of the future methodology.

Review this module through five observable capabilities: construct a feasible dispatch, explain energy–reserve opportunity cost, distinguish price controls, interpret run/price-setter evidence, and reproduce a regional-price calculation. A completed quiz tests selected knowledge; the written evidence pack still requires a competent reviewer.