NEM Weather Relationships

There goes an argument that because wind and solar is correlated in the NEM that transmission between states is of low value.

A single correlation factor is not adequate to describe the relationship between regional weather patterns. The relationships are driven by weather systems large enough to affect multiple regions at once. These systems create correlation between some areas, and anti-correlation with other areas simultaneously. A system creating wind in one location must be depriving another of wind - it cannot be windy or calm everywhere at once. The weather system will also move which changes which areas are correlated or anti-correlated over time.

Let's track the path of one weather system over time.

A high weather system sits over western Victoria, creating calm conditions at its centre. Southern NSW is also calm, but wind speeds are enhanced in SA, TAS, QLD, and northern NSW.

The high has moved north-east, weakening wind in NSW and TAS. Wind in western VIC is still quite weak but about to pick up. All three regions are correlated in terms of potential wind power production.

The high continues east and night-time winds pick up. Windy everywhere except NSW. Correlation between QLD and TAS even though they are far from each other.

The high moves further north, calming QLD. Good winds in all other regions.

Across a day and a half we've had correlation and anti-correlation between all regions for wind.

When a cold front rolls in from the southern ocean you see initial anti-correlation as it hits SA, growing correlation as it sweeps east across VIC and southern NSW, and then anti-correlation again as it disperses.

These systems are constantly changing in strength and location, and a single modestly positive correlation factor between two points does not tell us anything about the other points or the other anti-correlations.

Macro observations:

• QLD solar performs far better in winter than other regions, both because of latitude and climate (dry winter).
• TAS often has different weather at one corner, it's own internal anti-correlation.
• Offshore wind at the tip of southern NSW performs well when other southern wind resources are weak.
• Wind in northern Australia is more affected by diurnal heat than in the south. This leads to anti-correlation with solar production. This is true as far south as New England and Geraldton.
• Rooftop solar is well correlated with cooling demand, but poorly correlated with heating demand on cold cloudy days.
• Adelaide & Melbourne commonly share heatwaves, but besides that the capital cities have mismatched maximum loads.

What are the implications?

If we want to minimise system costs and maintain reliability inter-regional transfers are a potent tool. I believe the states are undervaluing the benefits of trading with their neighbours and are too internally focused. At the same time we don't have an adequate assessment process for transmission investment that properly measures the impact on customer bills.

I've run models in the past to minimise system costs at high levels of renewable energy and what I find is that productivity is the most important factor in choosing renewable resources. That means solar in the north and wind in the south will be maximised first in order to reduce generation costs. Diversity is the next most important factor, meaning solar and wind is needed in all regions to reduce transmission and storage requirements. Diversity at the cost of productivity will not be selected, e.g. solar in TAS, or second tier wind farms in QLD. Because cloud patterns vary so much over the continent it is preferable to have solar resources spread out and not concentrated, even though the diurnal patterns are extremely correlated. Such a model will always try to minimise storage requirements given the high cost.

If we treat each of the regions as islands without sharing connections there will be an increased need for storage, as well as a large scale up in renewable generation to be able to charge the storage at a time when the local renewable resource is poor in that region.

The public process of determining transmission spending inflates people's impression of the relative cost of transmission. They do not see the spending on generation and distribution because it is not subject to public scrutiny. Transmission will be the smallest capital item in the transition, dwarfed by generation (10x) and distribution (3x), but from the negativity you'd think it was the largest cost.