Why constraints exist

Wires have limits

Suppose a cluster of remote generators can produce 700 MW at $30 per MWh, a city 300 kilometres away needs 700 MW, and the one line between them is rated at 500 MW. NEMDE would love to dispatch all 700 MW of cheap power, but only 500 MW fits down the line. The remaining 200 MW must come from more expensive plant near the city.

 +----------------------+      +--------------+      +----------------+
 |  Remote generators   | ---> |   The line   | ---> |     City       |
 |  cheap, can do 700MW |      |  limit 500MW |      |  needs 700 MW  |
 +----------------------+      +--------------+      +-------^--------+
                                                      +------+---------+
                                                      | Local plant    |
                                                      | costly, nearby |
                                                      +----------------+

A constraint is the rule "flow on this line must stay at or below 500 MW", written in a form NEMDE can enforce. There were hundreds of thousands of constraint equations defined in AEMO's standing data as of August 2026, but only a few hundred matter in any given week, and every one has the same shape, which Chapter 3 dissects.

What kind of limit?

"The line can carry 500 MW" hides several different physical stories. AEMO classifies constraints by the phenomenon they guard against:

Thermal

Conductors and transformers heat up with current. Ratings depend on ambient temperature and wind, so many thermal limits are recalculated live. This is the most common family.

Voltage stability

Push too much power through a weak part of the network and the voltage can collapse after a fault. The limit is on total transfer, often across several lines at once.

Transient stability

After a large fault, synchronous generators can fall out of step with the grid. Limits keep enough margin that they hold together.

Oscillatory stability

Power swings between distant machines can grow instead of damping out. Rare but long-lived limits, usually on interconnectors.

System strength

Inverter-based wind and solar need enough synchronous machines online nearby to stay stable. These limits cap inverter output as a function of how many synchronous units are running.

Frequency control (FCAS)

Not a network limit at all: a requirement that enough fast reserve is held to catch the frequency after the largest credible loss. Same equation shape, different commodity (Chapter 8).

The next fault has not happened yet

Most network constraints are not about today's flow on a line. They are about the flow that line would carry if its neighbour tripped. The grid is operated to the N-1 standard: at any moment it must survive the loss of any single credible element (a line, a transformer, the largest generator) without exceeding any limit. So the typical constraint reads "keep the flow on line 39 low enough that if line 11 trips, line 39 is still within its rating". You will see that pattern in the names in Chapter 6: a monitored element and a contingency element, side by side.

This is why limits look conservative from the outside. A line rated at 800 MW may be held to 500 MW of pre-contingent flow because the trip of a parallel line would dump another 300 MW onto it within a cycle, long before any operator could act.

Beyond physics: the operational families

Not every constraint is a law of physics. Several families exist for operational reasons, and telling them apart matters because they behave differently:

• Outage constraints. When a line is out for maintenance the network topology changes and a different set of limits applies. AEMO keeps ready-made constraint sets for common outages and switches them on for the duration.
• Discretionary caps. Round-number limits an operator imposes on a named unit or group, for example capping a wind farm at half its capacity during a commissioning or system-security issue. The cap is often written into the name.
• Ramping constraints. Auto-generated each day per unit to enforce agreed ramp rates. They carry an ephemeral id beginning with # and are noise for any analysis of congestion.
• Commissioning and hold-point limits. A new plant is allowed up to a fixed output while it is tested.
• Quick constraints. Generated by the energy management system when an overload appears that no existing equation covers. They carry a CA_ prefix (constraint automation) and can appear within minutes.
• Negative residue constraints. When power flows from a high-price region into a low-price one, the settlement residue goes negative. AEMO can constrain the interconnector to stop the bleed.