NEOBIOME

LIT_031: Mohseni, Khalid & Brent (2022) — Metaheuristic isolated microgrid sizing with EVs, Aotea-Great Barrier Island

Properties3
typesource
source_typeliterature
topicsenergy_generation, energy_storage_grid, mobility_transport, methodology_frameworks, case_study, nz_specific

9 min read

LIT_031: Mohseni, Khalid & Brent (2022) — Metaheuristic isolated microgrid sizing with EVs, Aotea-Great Barrier Island

Source

doi:10.1016/j.egyr.2022.08.254 — original publication (opens in a new tab; the file is not redistributed)

Source type: literature (peer-reviewed journal paper).

Mohseni, S., Khalid, R., & Brent, A.C. (2022). “Metaheuristic-based isolated microgrid sizing and uncertainty quantification considering EVs as shiftable loads.” Energy Reports 8: 11288–11308. doi:10.1016/j.egyr.2022.08.254. Open access (CC BY-NC-ND 4.0). Affiliations: Sustainable Energy Systems, Victoria University of Wellington (NZ) and Stellenbosch University (South Africa).

Summary

This open-access journal paper develops a metaheuristic-based capacity-planning optimisation model for fully off-grid (stand-alone) microgrids that treats electric-vehicle charging as a shiftable (load-addressable) demand inside the optimisation rather than a fixed load. The model minimises whole-life net present cost (TNPC) subject to operational and planning constraints, using a cycle-charging rule-based dispatch strategy and the artificial hummingbird algorithm (AHA) as the optimiser, benchmarked against particle swarm optimisation (PSO) and a genetic algorithm (GA). It is applied to three real micro-communities on Aotea–Great Barrier Island, Aotearoa New Zealand — Medlands (MG 1, solar PV + wind + battery), Tryphena (MG 2a, solar-only + battery), and Mulberry Grove (MG 2b, solar-only + battery) — using 15 years (2007–2021) of hourly NIWA solar and wind data. The paper reports component sizing, levelised cost of electricity, capital-budgeting metrics, energy-flow breakdowns, and extensive univariate sensitivity analyses on meteorology, load, and EV-charging timing.

This is a sibling study to the wider NZ microgrid work already in the wiki: it shares an author (Brent) and methodology family with LIT_002 (community microgrids review) and the Totarabank (CR_007) and Rakiura/Stewart Island (RT_008) case studies. It is the most relevant published NZ off-grid case for EV-as-flexible-load coordination and for component-level techno-economic anchors in 2021 NZD.

Key claims

- claim: "Three off-grid microgrids were modelled for three Aotea-Great Barrier Island micro-communities — Medlands (MG 1: solar PV + wind + battery), Tryphena (MG 2a: solar PV + battery), and Mulberry Grove (MG 2b: solar PV + battery) — using 15 years (2007-2021) of hourly NIWA solar irradiance (SolarView) and wind speed (CliFlo) data; the island had a usually-resident population of 936 (2018 census)."
  source_location: "Section 4 (Case study), p.11296-11297; Appendix A.2, p.11303; Table A.1, p.11305"
- claim: "Optimal AHA-derived component mix and total net present cost: MG 1 = 209 PV panels + 9 batteries + 2 wind turbines, TNPC NZD 415,838; MG 2a = 796 PV + 6 batteries, TNPC NZD 411,073; MG 2b = 536 PV + 3 batteries, TNPC NZD 246,999."
  source_location: "Table 4, p.11298; reiterated p.11301"
- claim: "Levelised cost of electricity of the optimised systems: MG 1 = NZD 0.09/kWh, MG 2a = NZD 0.10/kWh, MG 2b = NZD 0.09/kWh, versus a current NZ average retail price of domestic electricity of NZD 0.19/kWh (MBIE 2020) — up to ~69% lower than retail."
  source_location: "Table 10 and Section 5.6, p.11301"
- claim: "Capital-budgeting metrics for the three microgrid investment proposals: profitability index (PI) 2.91% / 2.41% / 2.06%; discounted payback period (DPP) 7.1 / 7.5 / 8.4 years; internal rate of return (IRR) 17.93% / 17.75% / 17.53% for MG 1 / MG 2a / MG 2b respectively."
  source_location: "Table 11, p.11302"
- claim: "Techno-economic component specifications (2021 NZD): Senwei SWT 50 kW wind turbine NZD 65,000/unit (25-yr life); JA Solar Half Cut PERC mono PV panel 0.33 kW NZD 335/unit (25-yr life); Tesla Powerwall 14 kWh battery pack NZD 15,000/unit, replacement NZD 11,000, O&M NZD 30/yr (15-yr life); Eaton DG IP21 21 kW inverter NZD 8,000/unit (20-yr life); SolarEdge 7.6 kW EV charger NZD 4,000/unit (20-yr life). Project lifetime 25 years; real interest rate 4%; average 2021 exchange rate 1 USD = 1.41 NZD."
  source_location: "Table 2, p.11293; Table 3, p.11297; footnote 6, p.11293"
- claim: "The artificial hummingbird algorithm (AHA) consistently ranked first against PSO and GA across all three case studies and outperformed both by approximately 6% on average (best-run total discounted cost) over the three cases; the AHA also converged in fewer iterations."
  source_location: "Abstract, p.11288; Section 5.2 and Table 5, p.11298-11299"
- claim: "Coordinating (shifting) EV-charging loads materially lowers off-grid microgrid cost: relative to the late-evening/early-morning business-as-usual schedule, deferring the entire MG 1 EV load to the lightest-load hours (12 a.m.-5 a.m.) reduces TNPC by ~9% (~NZD 38,000-39,000), whereas shifting EV charging to afternoon hours (12 p.m.-4 p.m.) increases TNPC by ~20% (~NZD 83,000)."
  source_location: "Abstract, p.11288; Section 5.3.2, p.11299-11300; Tables 7, 8 and 9, p.11300-11301"
- claim: "Energy-flow breakdown for MG 1 (best AHA run, business-as-usual EV charging): ~38% of total generation from solar PV and 62% from wind turbines; on the consumption side residential/commercial loads ~66%, EV-charging loads ~22%, with ~12% attributable to non-ideal scenarios, renewable curtailments and unserved EV loads (EV reliability index ELF = 0.005)."
  source_location: "Section 5.5 (Energy flow analyses), p.11301; Fig. 7, p.11302"
- claim: "EV-charging loads were modelled as private Nissan Leaf EVs (6.6 kW charging power, 40 kWh battery, ~270 km range) and utility LDV EV-80 EVs (6.6 kW, 56 kWh, ~190 km range); MG 1 was assigned 10 private + 5 utility EVs, with private EVs requiring a full charge every alternate day and utility EVs daily."
  source_location: "Appendix A.3 (EV-charging loads), p.11304; Appendix A, p.11304-11305"
- claim: "For MG 1, increasing forecast wind speed by 40% lowers TNPC by ~34% versus baseline, while wind-speed deviations of -40% to -10% result in no wind turbine being selected (PV + battery only); for MG 1 a 10% increase in load demand causes rejection of PV panels (wind + battery preferred), and a +40% load deviation raises TNPC by ~43% versus baseline."
  source_location: "Section 5.3.1 and Table 6, p.11299; Section 5.3 and Appendix B (Tables B.4-B.6), p.11306"

Neobiome Intelligence relevance

This is the closest published NZ analogue to a fully off-grid, EV-integrated community energy design, and it lands several anchors the NI energy skill can use directly:

  • Component cost anchors in NZD (2021). Table 2 gives manufacturer-named, per-unit capital, replacement and O&M costs for PV (JA Solar Half Cut PERC, NZD 335/unit at 0.33 kW LIT_031), battery (Tesla Powerwall 14 kWh at NZD 15,000/unit LIT_031), wind (Senwei SWT 50 kW at NZD 65,000/unit LIT_031) and inverter (Eaton 21 kW at NZD 8,000 LIT_031). The PV figure implies ~NZD 1,015/kWp panel-only and the battery ~NZD 1,071/kWh pack-only — both consistent with, and slightly cheaper than, the installed benchmarks in CR_008 (which include installation uplift the LIT_031 figures may not). Treat LIT_031 as equipment-cost anchors, not installed-cost anchors.

  • NZ off-grid LCOE benchmark. Optimised LCOEs of NZD 0.09-0.10/kWh LIT_031 sit below the CR_007 Totarabank band (NZD 0.19-0.27/kWh) and well below NZ retail. The difference is partly methodological (this study is fully off-grid with no grid-tie revenue, and treats the inverter cost exogenously) — useful as the optimistic end of the community-microgrid LCOE envelope.

  • EV-as-flexible-load is a first-class design lever. The ~9% TNPC reduction from deferring EV charging to the lightest-load window, versus a ~20% penalty for afternoon charging LIT_031, gives NI a concrete, NZ-context magnitude for the value of coordinated (V1G-style) EV charging. This complements the V2G/V1G flexibility hierarchy already captured under D05 from OT_024, adding an off-grid sizing-cost dimension the IEA material does not.

  • Resource-complementarity finding. For MG 1 the study finds no significant seasonal complementarity between solar PV and wind at the site — wind follows an almost constant daily pattern — so wind primarily improves energy security (diversified portfolio) rather than lowering cost LIT_031. This is a transferable caution for NI: do not assume PV+wind hybridisation reduces cost; justify wind on resilience grounds (converges with the diversification argument in RD_003 and the grid-backup logic in Interview II [INT_002]).

  • Capital-budgeting template. PI / DPP / IRR results (IRR ~17.5-17.9%, DPP 7-8.4 years LIT_031) and the explicit PI, DPP and IRR equations (Eqs. 49-51) give NI a worked NZ investment-appraisal example for community microgrid proposals.

Caveats for NI use. (1) Costs are 2021 NZD and global PV/battery prices have fallen since — discount accordingly. (2) The inverter is sized exogenously (not a decision variable) and EV-charger costs are excluded from the optimisation, so reported TNPC/LCOE understate full system cost. (3) Loads were synthesised (no reliable historical island data), and tourism-driven seasonal population swings are acknowledged but not fully captured. (4) Battery ageing is simplified (95% fixed charge/discharge efficiency, no calendar/cycle degradation in sizing). (5) Great Barrier Island’s higher wind and solar resource is not the same as the Tasman/Lower Moutere pilot site — use the method and the EV-shifting magnitudes as transferable; recalibrate the absolute resource and cost numbers.

Research targets

Documents to retrieve

  • RT_160 (doc) — Soykan et al. (2022), “Optimal sizing of an isolated microgrid with electric vehicles using stochastic programming,” Energy Grids Netw. 32:100850. Primary stochastic-programming EV+microgrid sizing reference behind this paper’s EV-flexibility framing; candidate for EV-uncertainty methodology.
  • RT_161 (doc) — Park (2021), “Aotea great barrier island — A world of its own.” Cited as the source for the island’s general climatic conditions and the prior PV+wind techno-economic feasibility assessment; potential NZ island case-context primary.

Research gaps

  • RT_162 (gap) — NZ off-grid community microgrid component costs need updating from 2021 NZD to current (2025-2026) prices, and from equipment-only to installed cost, before LIT_031’s TNPC/LCOE figures are used in NI calculations. Cross-reference CR_008 and RD_003.
  • RT_163 (gap) — Quantify the value-of-EV-flexibility (TNPC reduction from coordinated charging) for a low-density NZ mainland pilot (Tasman/Lower Moutere) where the EV count is far smaller (~2 BEVs today, ~32 at fleet turnover per RD_011) than the 15-EV island fleet modelled here — the ~9% benefit may not scale down.

Connections

Links to

Sources (6): CR_007 · CR_008 · LIT_002 · OT_024 · RD_003 · RD_011

Referenced by

Sources (9): CR_051 · DS_003 · LIT_048 · LIT_067 · LIT_068 · LIT_069 · LIT_078 · LIT_084 · OT_118

SSI indicators (3): I01: Financial & Economic Self-Sufficiency · I06: Resistance to External Shocks · I07: Fulfilment of Basic Needs

EDT domains (2): D01: Renewable Energy & Storage Systems · D05: Smart Mobility & Electrified Transport

Graph View

Backlinks