Source
https://www.eeca.govt.nz/insights/eeca-insights/understanding-the-value-of-residential-solar-pv-and-storage-in-new-zealand/ — original source (opens in a new tab; the file is not redistributed)
EECA (2025) — Understanding the Value of Residential Solar PV & Storage in NZ (+ Appendix Six, load profiles)
Resolves RT_224 — the authoritative NZ residential DEMAND diurnal shape (NI engine E1/E2)
The official EECA / EPECentre (University of Canterbury) study and its Appendix Six representative-load-profile dataset (~49,385 ICPs, 4 main centres, half-hourly, gross pre-solar, 8 K-means archetypes). This is the underlying document behind the model’s demand-shape input — cited directly (data_quality high), with the 3-slice fractions recorded as a transparent derivation from its load profiles, not from the AI synthesis. Companion to OT_061 (Appendix One of the same publication).
Summary
EECA’s Understanding the Value of Residential Solar PV and Storage in New Zealand (2025, EPECentre/University of Canterbury authors) and its Appendix Six — Selection of Representative Load Profiles are the most authoritative, openly-retrievable NZ residential half-hourly demand dataset: ~49,385 ICPs across four main centres, gross (pre-solar) demand, clustered by K-means into 8 archetypes. From the documented cluster ratios (Appendix Six, Table 8) the model’s three-slice demand shape is derived as solar_day 0.25 / evening_peak 0.20 / night_base 0.55, which independently converges with the GREEN Grid household study (~0.23/0.26/0.51). This is the primary source behind the NI demand-distribution cell — replacing the model’s earlier triangulated estimates with a sourced figure. It supersedes the original 2016 EPECentre energywise PV Solar Calculator dataset (~18,000 ICPs / ~32 archetypes), whose database is not separately published in open form.
Key claims
- claim: "EECA (2025) 'Understanding the Value of Residential Solar PV & Storage in NZ' + Appendix Six: a NZ residential half-hourly demand dataset of ~49,385 ICPs across 4 main centres, GROSS (pre-solar) demand, clustered by K-means into 8 representative archetypes. Authoritative, openly retrievable, government/UC primary. Supersedes the original 2016 EPECentre 'energywise PV Solar Calculator' dataset (~18,000 ICPs / ~32 archetypes), which is not separately published in open form. Companion Appendix One = OT_061."
source_location: "EECA 2025 main report + Appendix Six (Selection of representative load profiles)"
- claim: "NZ residential DEMAND diurnal shape (for the model's 3 slices) DERIVED from Appendix Six load profiles ≈ solar_day 0.25 / evening_peak 0.20 / night_base 0.55 (gross, pre-solar; night-heavy ~55%; dual-peak). It is a derivation from the documented cluster ratios / figure curves (EECA does not publish a 3-slice energy table in the model's exact windows), so it carries ±0.02–0.03/slice uncertainty (evening_peak defensible range 0.20–0.26); it independently converges with the GREEN Grid household study (~0.23/0.26/0.51). Per-dwelling (not per-person); 4 cities (diurnal SHAPE is far less region-sensitive than total consumption, so national use is reasonable)."
source_location: "Appendix Six (cluster ratios, Table 8); cross-check: GREEN Grid EECA Part B (Jack et al. 2019)"Neobiome Intelligence relevance
Sources the model’s demand-distribution cell (E1/E2) — moving it from triangulated estimate to an authoritative document. The NI engine’s default demand shape (A_stripped-style 0.25/0.20/0.55) is confirmed and now sourced to EECA’s ~49k-ICP dataset (supersedes the triangulated CR_019/CR_020). No model value change — the figure matches what’s in the engine. With the PV generation shape (CR_035) this completes a fully-sourced temporal engine (gross demand × generation, dispatched in the 3 slices).
⚠ One construct caveat: this is the gross/total demand shape (includes space- and water-heating load); the model’s A_stripped variant is conceptually heating-stripped (the heat-pump load is modelled separately in the heat slice). The two are numerically close here but are different constructs — keep the distinction when the heat slice feeds back into electricity demand.
Research targets
Research gaps
- The original 2016 EPECentre PV Solar Calculator ~18k-ICP / ~32-archetype database (RT_224’s named target) is not openly published as a dataset; it sits in the UC Research Repository, and would have to be obtained from there if the original archetype detail is ever needed. The 2025 ~49k-ICP dataset (this source) is the better-documented successor and is what the model rests on.
- A seasonal demand split (winter evening peak is sharper) and a heating-stripped 3-slice shape (to pair cleanly with the heat slice) — future refinements if the engine seasonalises.
Connections
Links to
Referenced by
Sources (1): LIT_073
EDT domains (1): D01: Renewable Energy & Storage Systems