Methodology

What goes in

• Transpower Interconnection Branch Capacity. The MVA ratings on every transmission corridor (monthly).
• Electricity Authority EMI nodal pricing. Three years of half-hourly settlement prices at every grid exit point (monthly).
• Transpower Envision Opportunities Data Table. The N-1 worst-case capacity at every substation (annual).
• Transpower Generation Connection Pipeline. Announced and anonymised generation projects, with stage and MW (monthly).
• Transpower Non-Generation Pipeline. Load and storage projects (monthly).
• Transpower public GIS. Substation and transmission-line geometry.

Sources and licensing

Beacon is built entirely on public data released by New Zealand agencies for re-use under open licences. Every input, its source and its licence is listed below. Beacon's value is the original compilation, calibration and commentary applied to these inputs, which remain Lumere's own work.

Beacon contains data sourced from Transpower New Zealand Limited, the Electricity Authority Te Mana Hiko, and Stats NZ Tatauranga Aotearoa, each licensed for re-use under the Creative Commons Attribution 4.0 International licence (CC BY 4.0). Lumere is not endorsed by, affiliated with, or acting on behalf of any of these organisations. Any analysis, classification, ranking or commentary is Lumere's own and is not attributable to the data providers.

How we weight the queue

Not every megawatt in the connection queue is equally real. A consented project under construction is close to certain. A speculative early-stage idea usually is not. Before we measure anything, Beacon scales every queued project by how likely it is to proceed, using its consent status and its delivery stage together. Early, speculative and on-hold projects are heavily discounted. Consented and delivery-stage projects count at or near their full size. The result, which we call probability-weighted MW, feeds every saturation and headroom figure on the platform. It means a location is never marked full on the strength of interest that may never build. The exact weightings are Lumere's own calibration.

Reading a node

A node is a grid connection point. The question at a node is simple. Is there room to connect here? We compare the node's probability-weighted generation queue against its secure capacity, which is Transpower's worst-case N-1 winter figure, the conservative limit they plan to. Non-generation load is shown alongside as context and is not folded into the number, because its true effect on headroom cannot be settled without power-flow modelling, and this is a screening tool. A node cannot honestly read its own headroom if the lines leaving it are full, so it also inherits the constraint of its worst evacuating corridor. Nodal price is shown as a ranking, not a gate. Three-year average prices across New Zealand sit inside a narrow band, so price separates otherwise-similar nodes rather than deciding the call.

• Opportunity. Clear room to connect. The weighted queue sits comfortably below the node's secure capacity, and its export corridors are not the binding limit.
• Consider. Filling up. Meaningfully committed but not yet full, or a read we have softened because we hold only part of the data. Worth a look, so test the specifics.
• Avoid. Effectively full. The queue, or the corridor that must carry the power away, leaves no meaningful room to build without reinforcement.
• Data Gap. We do not hold published capacity or pricing for this point. Usually a connection or local distribution asset that Transpower does not publish by Electricity Code design, or a new connection not yet in the annual capacity table. Shown for orientation, with no call attached.

Reading a corridor

A corridor is a transmission route between nodes. The question shifts from room to connect to room to export. Can this route carry more power? Corridor capacity is Transpower's published interconnection branch rating, the Winter Continuous MVA, converted to MW at a standard power factor. It is a transparent thermal screen, not a security-constrained transfer study. We load it with the same probability-weighted queue, then look for the binding span. Where a corridor runs as parallel circuits, the most-loaded circuit caps the whole route. On long multi-segment lines such as Benmore to Islington, the worst case along the chain sets the headline. We flag that bottleneck branch so you know which span would need engineering attention.

• Opportunity. The route has room. The weighted queue is well within capacity, including its most-constrained span.
• Consider. Tightening. Either overall load is climbing, or a single binding branch is the limit even though the average still looks comfortable.
• Avoid. Oversubscribed. Queued projects exceed what the binding span can carry, so exporting more would require reinforcement.
• Data Gap or display-only. Connection assets, tee-junctions and HVDC links carry no published interconnection capacity, so they appear for orientation only, with no call.

Bands and data quality

Every figure is shown as a green, amber or red traffic light. Price and spread are banded against the national distribution, so the bands do not drift as the market moves. Saturation uses Lumere's own capacity thresholds. Each call also carries a data-quality band of High, Medium or Low. It reflects how complete our inputs are for that location, not how attractive the opportunity is. At a node it rises with a matched pricing point, a usable price history and a capacity figure. At a corridor it reflects how cleanly the pipeline maps to branches, branch coverage, and how well the endpoint prices agree. A Low band can soften an Opportunity to Consider, but it never manufactures an Avoid. Data quality flags how far to trust a call. It does not invent a negative signal.

What is covered, and what is not

Beacon covers Transpower's published interconnection network, the high-voltage backbone with published capacity, and every node and branch on it carries a call. It does not cover connection assets, the radial feeds from the backbone to a local grid exit point, for which Transpower does not publish capacity by Electricity Code design. Those lines are drawn for orientation only, with no investment signal. HVDC links and infrastructure deviations are shown the same way. Where you see Data Gap or a display-only line, it reflects what is genuinely unpublished, not work we have yet to do.

The BESS read (Tier 2)

Tier 2 reads each node as a battery site, from the market data, with no forecasts. There are four steps.

• The spread. From 36 months of half-hourly EMI prices we measure the daily arbitrage spread at each node, the gap between the cheapest and dearest hours of a typical day, matched to a battery run-time of 2 hours and 4 hours. This is what a battery earns cycling once a day, not a theoretical annual range.
• The node's own signal. Spreads rise and fall nationally, so we subtract each island's median to isolate what is specific to this node, whether it is a stronger or weaker arbitrage spot than its neighbours.
• The trend. Using a robust three-year trend on that node-specific signal, we show whether the edge is widening, steady or narrowing. This is the durable part, with the national market swing stripped out.
• Siting. Arbitrage is not the whole case, so we surface grid strength (the short-circuit ratio, or how hard and costly a battery is to connect), storage competition (battery capacity already operating or queued at the node), and reserves (the FIR and SIR markets, a second island-priced revenue stream that is expected to compress as more batteries connect).
• First mover. Where a standalone battery is already operating at a node we step the band down by one (for example strong to moderate), because the running battery holds the best connection and the early reserves position, so a new project is a second mover competing for what is left. We apply this when the operating standalone battery is 50 MW or more, and we name it on the card. Batteries co-located to firm a solar farm do not trigger this, since they are not competing for standalone arbitrage. This is a screening signal, not a bankability verdict.

The green, amber or red call is set by the daily arbitrage spread ranked across all NZ nodes, then stepped one band where this node's premium is widening or eroding faster than its region, so the colour reflects both how big the spread is and whether the node's edge is durable. A thin-data node never shows green, and a node with a standalone battery already operating steps down. A strong BESS node has a wide daily spread, a durable or widening node-specific trend, workable grid strength and connection room, and no first mover already operating. We rank locations on what three years of market data actually show. We do not publish forward projections. The signal is the evidence, not a forecast.

Tier 3, Curtailment lens (solar & wind)

Measured, backward-looking screening of what solar and wind output has actually earned at each node (capture factor), whether that value is eroding, and where grid curtailment risk is accumulating. Not a forecast or a bankability tool.

Tier 3 answers one question: will solar or wind output actually be worth building here? Two different things erode a renewable project's value, and we call both 'curtailment':

• Economic curtailment - your output is worth LESS. Wind and solar tend to generate at the same time as everything else of their kind (it is windy or sunny across a wide area at once), which pushes the price down in exactly those hours, so a wind or solar MWh can be worth less than the average MWh. We measure this as the capture factor: the average price the technology actually earns divided by the average price across all hours. 100% means it earns the average; below 100% is a discount (cannibalisation); the trend shows whether that discount is deepening.
• Physical curtailment - your output cannot get OUT. If the local grid is full, the system operator instructs the farm to back off and the energy is lost entirely. We read this from Tier 1: how full the connection already is (saturation and headroom) and how much capacity is queued ahead of you.

We use real data, not a model, wherever we can. Wind capture is measured from operating wind farms' actual metered output multiplied by the price at that node - today NZ wind earns about 84% of the average price and is slowly falling. Solar still earns a small premium (about 102%) but it is eroding as more solar connects; because only a few solar farms operate today, solar leans on the national measured level plus each node's modelled daily shape, and shifts to fully measured as the fleet grows. We have also checked the effect directly across operating farms: wind capture clearly falls as more wind connects nearby, confirming the erosion is real and not just assumed; solar points the same way more weakly while its fleet is still small.

We ignore noise on purpose. A new farm's first months (commissioning) generate little and erratically, so we drop them and measure only steady operation - otherwise a ramp-up would look like a sudden value crash that is not real.

The call you see combines both mechanisms: today's capture level sets the band, while the erosion trend and the next-12-month grid pressure show where it is heading. Each node is flagged as measured (an operating farm here) or analogue (estimated from the nearest fleet) so you always know how direct the evidence is.

Each node is measured against its own island, because the HVDC link makes the North and South Islands distinct price regions; this isolates how a node performs versus its island peers rather than mixing two different markets.

This is a measured, early-warning screen - not a revenue forecast or a bankability study. Use it to decide where a deeper forward study is worth doing.

Screening tool, not a bankability or forecasting product. All figures observed and backward-looking; leading indicators are derived from announced or measured facts.

What Beacon is not