Back on 19th April 2025, we noted that the ‘AEMO commenced a consultation on the Constraint Formulation Guidelines’ – including with this note about Aspect #2:

Today we’ve already noted about ‘Release 1 of the ISF (Improving Security Frameworks) reform went live, on Tuesday 2nd December 2025’.

Minimum Left-Hand Side (LHS) factors have been increased

It’s also worth explicitly calling out the increase in minimum LHS factors (from 0.07 to 0.15) that also went live yesterday (Tuesday 2nd December 2025).   This was communicated at 17:25 the day earlier in MN131322 as follows:

‘——————————————————————-
MARKET NOTICE
——————————————————————-

 From :              AEMO
To   :              NEMITWEB1
Creation Date :     01/12/2025     17:25:32

 ——————————————————————-

 Notice ID               :         131322
Notice Type ID          :         CONSTRAINTS
Notice Type Description :         New/Modified Constraints
Issue Date              :         01/12/2025
External Reference      :         Change to constraint equation LHS thresholds – 2 December 2025

 ——————————————————————-

 Reason :

 AEMO MARKET NOTICE

 As a part of the 2025 Constraint Formulation Guidelines consultation the threshold for the constraint equation left-hand side (LHS) factors is changing from 0.07 to 0.15. This change becomes effective on 2nd December 2025. AEMO will implement the constraint equation changes in a staged process from 1000 to 1600 hrs on 2nd December.

Many of the constraint factor changes have already been applied to AEMO’s pre-productions systems.

For more information on the Constraint Formulation Guidelines changes see this link:

https://www.aemo.com.au/consultations/current-and-closed-consultations/isf-consultation-of-constraint-formulation-guidelines .

Ben Blake
AEMO Operations

——————————————————————-
END OF REPORT
——————————————————————-’

For those with a keenness to delve further, the ‘ISF and IPRR Consultation of Constraint Formulation Guidelines’ page linked above contained this ‘LHS Factor Analysis’ spreadsheet from August 2025 that might be of interest:

Why this is (might be) a useful change?

There have been a number of articles written here on WattClarity about the large impact that certain constraint equations (like the X5 constraint) have had on dispatch outcomes, by virtue of the small LHS factor.

For those who want an example of the gory details, Allan O’Neil helpfully provided an explanation way back on 8th December 2020 in his article ‘Constraint Case Study Followup – Early outcomes of the new “X5 constraint”’ … in particular with respect to the ‘maths alert’ in the lower half of the article:

PS1 on Thursday 4th December 2025

It’s been suggested that I add in the following to explain more clearly/directly what the situation was, prior to this change (and the motivation for the change):

The original X5 articles pointed out that when binding:

1)  these constraints would tend to disproportionately swing interconnector flows (which has flow-on effects to units in the adjoining region … even those not on the LHS of the constraint in question)

2)  in preference to curtailing down DUIDs directly on the LHS in question (even those with high constraint coefficients).

This is because of the lower Marginal Cost to dispatch of reducing the constraint LHS one unit by constraining the interconnector.

Here’s an example:

Let’s say the constraint has two terms on the LHS – a DUID and an interconnector.  Therefore, the constraint has two options for reducing the LHS by 1 unit:

Option 1 = the constraint could reduce the LHS by constraining the DUID by the required amount:

1a)  relative to its LHS Factor

1b)  with the Marginal Cost to dispatch calculated as:

(DUID region spot price – [-$1,000/MWh]) / coefficient

Option 2 = the constraint could reduce the LHS by constraining the Interconnector by the required amount:

2a)  relative to its separate LHS Factor within the constraint

2b)  with the Marginal Cost to dispatch calculated as:

(DUID region spot price – adjoining region spot price)/coefficient

Typically the adjoining region spot price will be much higher than -$1,000/MWh so even with a smaller interconnector coefficient it may well be the interconnector that gets affected.

Adding numbers  to the example:

For simplicity let’s say:

Spot prices are $100/MWh (DUID region), $50/MWh (adjoining region)

The coefficients 1.0 (DUID) and 0.07 (Interconnector)

The DUID is bidding at -$1,000/MWh at the RRN

The relative costs are then

$1100 / 1.0 = $1100 to constrain the DUID

$50 / 0.07 = $714 to swing the interconnector (reduce import or increase export from DUID region)

So the interconnector will be affected before the generator.

Furthermore, to reduce the constraint LHS by 1 unit would require changing the interconnector flow by 1/0.07 = 14.3 MW whereas the curtailment of the DUID would require only a 1 MW reduction to have the same impact on the LHS – again because of the coefficient relativities.

The impacts of this interconnector reduction (in this example) are that:

Outcome #1 = marginal generation in the adjoining region (not even part of the constraint) has to reduce 14.3 MW (because of the changed interconnector flow),

Outcome #2 = while higher cost generation in the receiving state has to increase 14.3 MW.

Outcome #3 = which supports region prices either side of the interconnector going in opposite directions (low being lower, high being higher).

So in the case of the X5 constraints, in practice we would see heavy restriction or counterprice flows on VIC1-NSW1, meaning that marginal Victorian generators were being wound back by the constraint by many more MW than would be the case if the highest-coefficient DUID in NSW had been curtailed instead.

Increasing the minimum coefficient threshold would take interconnectors with very low coefficients out of the LHS and avoid these outcomes.

So one general motivation for increasing the minimum threshold for LHS factors is to reduce* the extent large changes in interconnector target flow are preferenced (thereby curtailing more low cost generation on the other side of the interconnector).

* though it may not eliminate, because the new lower bound (0.15) is still relatively low, compared to 1.0

Time will tell what effect this change actually has … perhaps with some longitudinal study of flows on VIC1-NSW1 when constraints like (what) X5 (used to be**) are bound.

** see below for an example of the change

An example … Before, and After

Allan’s articles above referenced the earlier form of the X5 constraint’

… then affectionately known as the ‘N^^N_NIL_3’ constraint equation

Some readers will recall that the NEM had its own ‘Prince moment’ in 2023, with the constraint being renamed (and even split into two), which we documented:

• On 16th March 2023 in ‘Splitting the ‘X5 constraint’ into two … double trouble?’
• and then on 5th April 2023 in ‘By X5, Hi X5’

As noted in those articles, the constraint equation (under system normal conditions) has since been represented by the pair:

• the ‘N^^N_NIL_X5_BESH’ constraint equation; and
• the ‘N^^N_NIL_X5_BEKG’ constraint equation.

Utilising two copies of the ‘Constraint Dashboard’ widget in ez2view – each focused on one of these ‘new’ constraint equations, we can do this ‘before and after’ comparison:

There’s three things I’d like to highlight in this comparison:

• There are 9 LHS terms that were removed from the constraint, because their LHS factors were below the new/higher 0.015 threshold in absolute terms
• Plus two other changes:
  • Two new batteries have been added … coincidentally; and
  • Some LHS factors of remaining terms have been adjusted

That’s all for now…