What phase balance changes at peak demand
A transformer can have spare total capacity while one phase is already constrained.
Explore one illustrative evening peak: 48 homes, hot water and three EV chargers supplied by a 200 kVA transformer. Customer phase connections are initially unknown to the network operator. Move a home to another phase or manage its load to see how currents and voltages change at this same point in time.
The fullest phase runs out first
–
Phase data helps in two ways. It lets the EDB fix the imbalance that is already there by reassigning existing customer connections without replacing the transformer or feeder. And it lets new load be connected, or load be managed, where it counts. Reducing demand on another phase may provide much less relief at the constrained location. Phase and neutral coupling mean that the voltage effect also depends on where loads are connected.
Try System-level and observe which phases respond. Each press selects available loads at random, so the outcome varies. Reset, then try LV-level, phase-aware to target the phase carrying the highest current. Compare the actual kW removed as well as the current and voltage relief: whole EVs and cylinders give discrete steps. Balance phases reassigns whole customer connections at this snapshot; a real phase-swap plan needs time-series checks before implementation.
Keys: B balance · F LV-level, phase-aware managed load · G system-level managed load · E new EV · R reset
The same reinforcement, pushed further out
Illustrative values are conditional on the peak being a binding planning constraint and the intervention providing dependable relief. Each option is valued the same way: how long it delays the same reinforcement of this transformer, and what that delay is worth using an illustrative discount rate. The calculation projects this peak-load case to an assumed phase-current planning limit. Crossing that limit in a single snapshot does not by itself establish that reinforcement is due: duration, thermal history, voltage, reliability and operating options need assessment. A single growth rate describes LV load poorly, so every result is shown at 1%, 3% and 5% a year.
Effective peak loading over time, from the street as shown
| Option | Phases (kVA) | Effective peak | Time to assumed limit | Deferral value vs as shown | ||||
|---|---|---|---|---|---|---|---|---|
| 1% | 3% | 5% | 1% | 3% | 5% | |||
An illustrative national extrapolation
Deferral value only exists where a transformer is heading for its limit. This scenario extrapolation takes Vector’s share of 200 kVA units (2,630 of 22,358 records) and the share of those above 80% utilisation (about 14%), applies them to New Zealand’s 196,885 distribution transformers, and values each transformer by how close it is to its rating.
This extrapolation is not a measured national opportunity and does not follow the interactive street, so it uses a fixed reference case rather than the lab: transformers above 80% are spread evenly between 80% and 100% utilisation, and an assumed half carry a 40:30:30 imbalance, close to the street’s starting split of –. The rest are treated as balanced and benefit from managed load only.
| Step | Value |
|---|
| Utilisation | Due with 40:30:30 | Value each |
|---|
Rating and utilisation counts are supplied assumptions from a Vector GIS summary (22,358 records); the extract and utilisation definition have not been independently verified for this page. Applying this mix nationally assumes it is representative; the 50% imbalance share and uniform utilisation bands are scenario assumptions. RAB and capex from the Commerce Commission, Total electricity distribution — year to 31 March 2025 [1]. Values are net of phase-swap costs, before paying for managed load.
How much room is left depends on how you count it
A transformer’s nameplate is shared across three phases, but customers are connected one phase at a time. These indicators use different bases: the first two describe transformer-wide kVA; the third describes kVA on one limiting phase. They are not interchangeable measures of connection capacity. This page values deferral on the fullest-phase count, a balanced-equivalent index of the highest phase current, rather than a full thermal or hosting-capacity assessment.
Rating minus total customer load, across all three phases. Excludes network losses and ignores the distribution of load between phases.
Rating minus total load minus the neutral component, a simplified comparison inspired by the Dewesoft analysis, not a transformer thermal model. Imbalance costs capacity roughly in proportion to the neutral current.
Current margin on the heaviest phase, expressed as per-phase kVA at the nameplate voltage. The assumed 200 kVA / 415 V rating gives 278 A per phase. Multiplying this margin by three gives the balanced-equivalent index used in the valuation; it is not guaranteed usable connection capacity.
How the model works
- The street. 48 homes on 10 poles, each with a 1–2 kW base load and a 3 kW hot water cylinder; about half the cylinders are heating at the peak. Three 7 kW EV chargers run at full output at the transformer peak, as when a tariff synchronises charging. Loads are treated as kVA at unity power factor.
- Effective loading. Three times the fullest phase, taken from the load flow as the phase current leaving the transformer, expressed as kVA at 415/√3 V. It differs from three times the largest customer phase load because the heavy phase runs at a lower voltage and draws more current for the same power. For a 40:30:30 split this is about 1.2 × the total load; the Dewesoft neutral add-on would give 1.1 ×, so this page sits at the cautious end. The fleet estimate uses phase kVA.
- Balancing. The same whole-customer greedy balancing algorithm is used by the action button and valuation, moving customers from the fullest phase to the lightest until no move lowers peak phase load further, at NZ$500 per phase swap. The count comes from the actual homes on the street, not an assumed load per ICP.
- Managed load. An illustrative additional 7 kW per phase (21 kW total) is removed from the balanced peak and the load flow is recalculated. Future uncontrolled effective load grows, while this calculated relief is held fixed. The same fixed-relief assumption applies to loads switched off in the lab. This requires repeatable availability at every relevant future peak; delivery costs, duration and rebound are not valued. The lab buttons shed about 9–10 kW at a time: LV-level, phase-aware managed load pauses EVs on the fullest phase first and tops up with hot water; system-level managed load picks any controllable loads, as a signal sent from upstream without phase visibility would.
- Your changes. The reinforcement present value for the street as it started, minus the present value for the street as shown, minus NZ$500 for every home now on a different phase. Adding EVs uses up value; managed load is valued before paying for it.
- Growth and discounting. The pre-control effective load grows at 1%, 3% or 5% a year, with fixed effective-load relief subtracted afterwards; headline figures are at 3%. The discount rate is the DPP4 midpoint vanilla WACC, 6.68% [2]. Reinforcement of a 200 kVA transformer is taken as NZ$150k.
- Fleet. NZ 200 kVA units = 196,885 × 2,630 / 22,358. Units above 80% utilisation use Vector’s 380 of 2,630, spread evenly across 80–100% (band midpoints 82.5–97.5%). Half carry a 40:30:30 imbalance and are valued for balancing (net of swaps at 3 kVA per ICP) and managed load; the rest for managed load only.
- Out of scope. Voltage-driven reinforcement timing (voltage is shown for the snapshot but does not set the valuation horizon), peak duration and frequency, thermal history, rebound, solar hosting and export constraints, upstream benefits, cyclic ratings above nameplate and consumer benefits. These factors may increase, reduce or eliminate the deferral value shown.
Fixed assumptions
- Transformer
- 200 kVA, 11 kV / 415 V
- Reinforcement cost
- NZ$150,000
- Discount rate (DPP4 WACC)
- 6.68% [2]
- Load growth fan
- 1% · 3% · 5%
- Phase swap cost
- NZ$500 / ICP
- Managed load
- 7 kW per phase
- NZ distribution transformers
- 196,885 [1]
- Vector 200 kVA units
- 2,630 of 22,358
- Vector 200 kVA above 80%
- 380 (14%)
- Fleet share with 40:30:30
- 50%
- Distribution transformer RAB
- NZ$2,181.5m [1]
- Total EDB capex, one year
- NZ$1,686.6m [1]
- Commerce Commission, Total electricity distribution — year to 31 March 2025.
- Commerce Commission, DPP4 cost of capital determination, [2024] NZCC 21 (midpoint vanilla WACC, used here as a discount-rate assumption).
Dots on each conductor are drawn in proportion to the current in that span, so they thin out at every pole as customers take their share. Load flow: a four-wire unbalanced load flow runs in the page; independent validation is pending publication of the referenced OpenDSS test cases. 200 kVA transformer represented by a three-phase source and series impedance, 1.2% R and 3.8% X, 415 V line-to-line / 240 V line-to-neutral at no load (assumed source voltage), LV neutral earthed at the transformer. 400 m overhead line in ten 40 m spans, four 185 mm² aluminium conductors (0.164 Ω/km) on a flat crossarm at 8 m, with self and mutual impedances from Carson’s equations as computed by OpenDSS. MEN earth of 10 Ω at each pole. Constant-power loads at unity power factor, service lines ignored. The displayed voltage screening band is an assumed 207–253 V; it is not a site-specific compliance assessment. The neutral add-on method follows Dewesoft, Extending Grid Capacity.