How the auction works
The big picture
Every morning, Israel's System Operator (Noga) runs a day-ahead auction. Generators submit their offers — how much electricity they can produce and at what price — and the system determines the cheapest way to meet tomorrow's demand.
The result is a dispatch schedule: which generators run during each half-hour interval and at what output level. The price that emerges is the Market Clearing Price (MCP).
This is a uniform pricing system. Every dispatched generator receives the same MCP, regardless of their individual bid. The MCP equals the cost of the most expensive generator needed to meet demand — the marginal unit.
The MCP is not set by any single generator. It emerges from the interaction of all bids, demand levels and system constraints. Change any one input and the entire dispatch, and the price, can shift.
The optimization problem
The auction is solved as a Mixed-Integer Linear Program. "Mixed-integer" means the solver works with two kinds of decision at once:
- Continuous — how much power each generator produces.
- Binary — whether a generator is on or off.
The objective is to minimise total generation cost while satisfying every system constraint, across all 48 half-hour intervals of the next day.
A simple merit-order stack cannot capture the physics of power plants that take hours to start, have minimum output levels, or face fuel delivery limits.
What shapes the price
| Power balance | supply must equal demand at every interval |
| Reserve requirements | margin held for unexpected changes |
| Ramp rates | generators cannot change output instantly |
| Min up / down times | thermal units need time to start and stop |
| Fuel supply | pipeline constraints limit total gas generation |
| Storage dynamics | batteries couple decisions across intervals |
How the price is set
Once the solver decides which generators run, the MCP is set by asking one question: if we needed one more megawatt right now, who would supply it and at what cost? That generator — the marginal unit — sets the price for everyone.
This works cleanly when all generators compete on cost alone. Israel's regulator chose to put additional values into the system. Environmental policy requires coal to dispatch last regardless of its low cost. Energy security means that when gas pipelines fail or hit capacity, the market does not charge consumers diesel-level prices for a problem they did not cause.
So the basic pricing rule needs exceptions. Seven scenarios govern how the MCP is actually set:
| # | Condition | MCP rule |
|---|---|---|
| 1 | Normal operation | marginal gas unit price — the cheapest unit that can respond |
| 2 | Coal unit is marginal | highest gas price already purchased; coal is excluded from price setting |
| 3 | Shortage, no demand management | energy: highest purchased offer · deviations: VOLL |
| 4 | Renewable curtailment | MCP = 0 — the cheapest fix is to restore the free renewables that were cut |
| 5 | Gas capacity shortage, units unavailable | if coal set the price → highest gas purchased; otherwise → diesel unit price |
| 6 | Gas supply shortage, supplier unavailable | highest gas price purchased |
| 7 | Gas transmission failure | highest gas price purchased |
Behind the rules is a balancing act: prices that are economically honest, that do not reward pollution, and that do not penalise consumers for infrastructure limits.