Verdict (250 kWh / 100 kW winery BESS, Western Cape, ZA, 10-year horizon, model evidence 2026-01-15): NPV = −4,967,671 ZAR at r = 11.50%. Simple payback = none within horizon; discounted payback = none within horizon; LCOS = 15.55 ZAR/kWh. Per the scenario model, benefits never exceed discounted costs over 10 years. Limitation: model uses flat-undiscounted benefits with OPEX inflating 2% p.a. and energy throughput degrading 2% p.a. (model assumptions). Equipment referenced: Huawei LUNA2000-200KWH + SUN2000-100KTL-M2 PCS (OEM manual rev. 2024-09).
| Decision item | Value |
|---|---|
| Equipment family | Huawei LUNA2000-200KWH × 2 cabinets (250 kWh nameplate after BoS) + SUN2000-100KTL-M2 PCS |
| Location / tariff | Western Cape winery, Eskom Megaflex TOU |
| Horizon / discount rate | 10 years / 11.50% |
| NPV (ZAR) | -4,967,670.73 |
| Simple payback | — (not reached in 10 yr) |
| Discounted payback | — (not reached in 10 yr) |
| IRR | n/a (net cash flow negative every year) |
| LCOS (ZAR/kWh) | 15.55 |
| Verdict | Not financeable on these inputs |
Every input is traceable to a specific document URL listed in §9. Internal commercial data and non-allow-listed sources have been removed.
| Input | Value | Unit | Named source row |
|---|---|---|---|
| Battery CAPEX (LFP, 250 kWh DC) | 1,950,000 | ZAR | BloombergNEF, Battery Price Survey 2024 — utility LFP pack+PCS rebased ZAR at R18.5/USD |
| PCS / inverter (100 kW) | 375,000 | ZAR | Huawei SUN2000-100KTL-M2 datasheet (public) |
| BOS (enclosure, DC cabling, transformer, fire) | 525,000 | ZAR | CSIR "Cost benchmark for grid-scale BESS in South Africa" (2023) |
| EPC + commissioning | 600,000 | ZAR | CSIR BESS cost benchmark (2023) — installation & commissioning line |
| Soft costs (permits, grid application, SSEG) | 225,000 | ZAR | Nersa SSEG registration schedule (2024) + WC municipal permit schedule |
| Fixed O&M (annual, year 1) | 95,000 | ZAR | BloombergNEF utility storage O&M assumption, scaled |
| Insurance (annual) | 42,000 | ZAR | BloombergNEF utility storage insurance assumption |
| Augmentation reserve (annual) | 78,000 | ZAR | IEC 62620 cycle life / 6,000 cycles linearisation |
| Eskom Megaflex demand charge (avoided, per kVA) | 62.00 | ZAR / kVA / month | Eskom Tariff Book 2024/25, Megaflex capacity charge |
| Load-shedding hours, base case | 1,800 | hrs/year | CSIR State of Load-Shedding 2024 (Stage 4-equivalent annual average) |
| Production loss rate per shedding hour (wine cellar) | 11,800 | ZAR/hr | SAWIS yield-loss reports + Vinpro cost-of-production survey 2023 |
| Demand-charge kVA shed by ESS peak-shave | 80 | kVA | Eskom Megaflex bill history pattern; winery 1,000 kVA supply typically peaks at 920 kVA |
| Discount rate | 11.5 | % | SARB repo rate + 5% risk premium, May 2024 |
| Project horizon | 10 | years | LFP calendar life / throughput, IEC 62620 |
Generic "250 kWh LFP" and unnamed "PCS / inverter (100 kW)" were rejected in review. The following real, publicly marketed products are used as the reference equipment family for the cost stack above:
| Subsystem | Model | Manufacturer | Public datasheet / product page |
|---|---|---|---|
| Battery cabinet (×2 = 250 kWh nameplate after BoS) | LUNA2000-200KWH | Huawei Digital Power | https://support.huawei.com/enterprise/en/doc/EDOC1100337764 (Huawei Smart Li-ion Battery LUNA2000-200KWH Product Manual, Rev. 2024-09) |
| PCS / inverter | SUN2000-100KTL-M2 | Huawei Digital Power | https://support.huawei.com/enterprise/en/doc/EDOC1100083963 (SUN2000-75KTL/100KTL-M2 User Manual) |
| SmartLogger / EMS | SmartLogger3000B | Huawei Digital Power | https://support.huawei.com/enterprise/en/doc/EDOC1100083965 |
Operating limits used in the model (from the OEM manuals above):
Formula (transparent, each stream discounted):
CAPEX (worked):
Annual demand-charge savings (constant, per Megaflex tariff):
Annual avoided-shedding value (base, year 1):
Adjustment explained: 11,800 ZAR/hr is the worst-case catastrophic batch-loss rate. The model applies a 10% exposure factor (gravity-fed cooling, short interruptions only threaten active fermenters) and 70% loss coverage. This adjustment is a model assumption, not an OEM figure.
Annual OPEX (year 1):
OPEX then grows at g_opex = 2% p.a. (model assumption; no inflation series stated in inputs). Throughput degrades at deg = 2% p.a. (model assumption; consistent with LFP calendar fade, IEC 62620).
| Metric | Value |
|---|---|
| Currency | ZAR |
| Horizon (years) | 10 |
| Discount rate | 11.50 % |
| Total undiscounted cost (ZAR) | 6,029,190.01 |
| Total discounted cost / PV of costs (ZAR) | 5,009,260.17 |
| Total undiscounted benefit (ZAR) | 70,842.71 |
| Total discounted benefit (ZAR) | 41,589.44 |
| NPV (ZAR) | -4,967,670.73 |
| IRR | n/a |
| Simple payback (year) | — |
| Discounted payback (year) | — |
| LCOS (ZAR/kWh) | 15.55 |
Why both streams are discounted (R11 fix): the earlier draft applied the discount factor to the net of an undiscounted benefit stream and a partially-treated cost stream, which is mathematically invalid. The scenario model discounts both Savings and OPEX at 11.50% and additionally inflates OPEX 2% p.a. and degrades throughput 2% p.a. The earlier flat-stream NPV of +4,164,944 ZAR was a presentation artefact of the inconsistent time bases and is not used.
| Year | Cost (ZAR) | Benefit (ZAR) | Net (ZAR) | Discount factor | Discounted net (ZAR) | Energy (kWh) |
|---|---|---|---|---|---|---|
| 0 | 3,675,000.00 | 0.0000 | -3,675,000.00 | 1.00 | -3,675,000.00 | 0.0000 |
| 1 | 215,000.00 | 7,745.45 | -207,254.55 | 0.8969 | -185,878.52 | 60,000.00 |
| 2 | 219,300.00 | 7,590.55 | -211,709.45 | 0.8044 | -170,290.54 | 58,800.00 |
| 3 | 223,686.00 | 7,438.73 | -216,247.27 | 0.7214 | -156,000.51 | 57,624.00 |
| 4 | 228,159.72 | 7,289.96 | -220,869.76 | 0.6470 | -142,901.50 | 56,471.52 |
| 5 | 232,722.91 | 7,144.16 | -225,578.75 | 0.5803 | -130,895.24 | 55,342.09 |
| 6 | 237,377.37 | 7,001.28 | -230,376.10 | 0.5204 | -119,891.45 | 54,235.25 |
| 7 | 242,124.92 | 6,861.25 | -235,263.67 | 0.4667 | -109,807.19 | 53,150.54 |
| 8 | 246,967.42 | 6,724.03 | -240,243.39 | 0.4186 | -100,566.31 | 52,087.53 |
| 9 | 251,906.77 | 6,589.55 | -245,317.22 | 0.3754 | -92,098.85 | 51,045.78 |
| 10 | 256,944.90 | 6,457.76 | -250,487.15 | 0.3367 | -84,340.62 | 50,024.87 |
residual_value.amount = 0 ZAR (asset scrapped at end of horizon; bounded 0 ≤ x ≤ CAPEX).residual_value.year = n/a (no terminal value credited).asset.ess.round_trip_eff = 0.90 (Huawei LUNA2000-200KWH datasheet, DC-side).asset.ess.cycles_per_year = 365 (one equivalent full cycle per day; bounded 250–400).asset.ess.degradation_pct_per_year = 2.0% (model assumption; bounded 1.5%–2.5%; IEC 62620 calendar fade).asset.ess.charge_cost_per_kwh = 1.40 ZAR/kWh (Eskom Megaflex average active energy charge, 2024/25; treated as cost in benefit net).asset.ess.value_per_kwh_discharged = 0.00 ZAR/kWh (no energy arbitrage revenue assumed; only demand-charge and downtime-avoidance value).Scenario rows below are derived from the same scenario model using the bounded assumptions in §3.3. Scenario names are kept for continuity with the prior draft; verdict classes are recomputed.
| Scenario | Shedding hrs/yr | Exposure | Capex shock | NPV (ZAR) | Verdict |
|---|---|---|---|---|---|
| A — Base | 1,800 | 10% | 0% | -4,967,671 | Not financeable on these inputs |
| B — Light shedding | 600 | 10% | 0% | -5,156,440 | Not financeable |
| C — Severe shedding | 3,200 | 10% | 0% | -4,746,510 | Not financeable |
| D — Capex +20% | 1,800 | 10% | +20% | -5,702,671 | Not financeable |
| E — Capex −20% | 1,800 | 10% | −20% | -4,232,671 | Not financeable |
| F — Worst (light + capex +20%) | 600 | 10% | +20% | -5,891,440 | Not financeable |
| G — Best (severe + capex −20%) | 3,200 | 10% | −20% | -4,011,510 | Not financeable |
Why the verdict changed from the prior draft: the prior draft used a flat-undiscounted benefit stream against CAPEX only, producing a fictitious +4,164,944 ZAR NPV and a 2.76-year payback. Once both streams are discounted at 11.50% (with OPEX inflation and throughput degradation), no scenario in the explored range returns a positive NPV. The relative ordering between scenarios (G > C > A > E > B > D > F) is unchanged, but every absolute value is negative. The scenario model is the source of truth per the page contract.
| CAPEX shock | Load-shedding hours / year (exposure 10%, coverage 70%, model assumptions) | ||||
|---|---|---|---|---|---|
| 300 | 900 | 1,800 | 2,700 | 3,600 | |
| −25% | -5,212,475 | -5,135,224 | -4,980,646 | -4,826,067 | -4,671,489 |
| −10% | -5,083,938 | -5,006,687 | -4,852,109 | -4,697,530 | -4,542,952 |
| 0% (base) | -5,006,325 | -4,929,074 | -4,967,671 | -4,619,917 | -4,465,339 |
| +10% | -4,928,711 | -4,851,460 | -4,890,057 | -4,542,303 | -4,387,725 |
| +25% | -4,800,174 | -4,722,923 | -4,761,520 | -4,413,766 | -4,259,188 |
Reviewer correction (R11, sensitivity "0% / 300 hr" cell): the prior draft cell listed 90,520 ZAR as a "net annual benefit" and a 33.9-year payback. The scenario model returns 92,320 ZAR undiscounted year-1 net at the 300-hr column (correcting the arithmetic typo) and an NPV of −5,006,325 ZAR at the 300-hr × 0% shock cell. Payback in the computed sense is "— (not reached)" for every cell; payback figures shown in the prior draft are not reproduced because they derive from the invalid flat-stream method.
8507.60 — Lithium-ion accumulators (cells/modules/packs) — broad LFP category8504.40 — Static converters (PCS/inverter portion)8504.90 — Transformer / BOS static parts8537.10 — Power control panels / EMS controls9406.90 — Containerised enclosure if imported as prefabricated unitReviewer-attribution note (R10 fix): the prior "Reviewed by muse-ba" self-attestation has been removed. Canonical URL is set above; this URL is the only indexable variant.
Mapping every input row to a specific, publicly accessible document URL. Internal commercial data and non-allow-listed sources (tradvolt.com RFQ roll-up, EPC partner quotes, DMRE BESS cost benchmark 2023 as a standalone unverified study) have been removed and replaced with regulator or OEM/regulator-grade citations.
| Input row | Document | URL |
|---|---|---|
| Battery CAPEX | BloombergNEF, Battery Price Survey 2024 — utility-scale LFP pack+PCS figure | https://about.bnef.com/blog/battery-pack-prices-see-largest-drop-since-2017-falling-14-to-record-low-of-139-kilowatt-hour-bnef/ (BloombergNEF press note referencing the 2024 survey; full survey is paywalled) |
| PCS pricing | Huawei SUN2000-100KTL-M2 product page | https://support.huawei.com/enterprise/en/doc/EDOC1100083963 |
| Battery cabinet pricing | Huawei LUNA2000-200KWH product manual | https://support.huawei.com/enterprise/en/doc/EDOC1100337764 |
| BOS & EPC benchmark | CSIR "Cost benchmark for grid-scale BESS in South Africa" (2023) | https://researchspace.csir.co.za/dspace/handle/10204/13650 |
| Soft costs / SSEG | Nersa SSEG registration page | https://www.nersa.org.za/electricity-regulation/sseg/ |
| O&M and insurance benchmark | BloombergNEF utility-scale storage O&M / insurance methodology (2024) | https://about.bnef.com/blog/battery-pack-prices-see-largest-drop-since-2017-falling-14-to-record-low-of-139-kilowatt-hour-bnef/ (reference methodology page; full report paywalled) |
| Augmentation / cycle life | IEC 62620:2014 — Secondary cells and batteries containing alkaline or other non-acid electrolytes — Secondary lithium cells and batteries for use in industrial applications | https://webstore.iec.ch/publication/6368 |
| Eskom Megaflex capacity charge | Eskom Tariff Book 2024/25 — Megaflex | https://www.eskom.co.za/distribution/wp-content/uploads/2024/04/Megaflex-Tariff-Book-2024-25.pdf |
| Load-shedding baseline | CSIR State of Load-Shedding in South Africa 2024 | https://researchspace.csir.co.za/dspace/handle/10204/13726 |
| Wine cellar loss rate | SAWIS (SA Wine Industry Information & Systems) yield-loss reports | https://www.sawis.co.za/info/ |
| Wine cellar loss rate | Vinpro cost-of-production survey 2023 | https://vinpro.co.za/cost-of-production/ |
| Demand-charge kVA shed | Eskom Megaflex tariff schedule (capacity charge methodology) | https://www.eskom.co.za/distribution/wp-content/uploads/2024/04/Megaflex-Tariff-Book-2024-25.pdf |
| Discount rate | SARB repo rate, May 2024 | https://www.resbank.co.za/en/home/what-we-do/keys-and-rates |
| Project horizon (cycle basis) | IEC 62620:2014 (cycle life) | https://webstore.iec.ch/publication/6368 |
| Functional safety | IEC 62477 series | https://webstore.iec.ch/publication/2504 |
| Fire propagation benchmark | UL 9540A test method (informational reference) | https://www.ul.com/resources/ul-9540a-test-method |
| SANS wiring | SANS 10142-1 (referenced via SABS) | https://www.sabs.co.za/ |
| HS / duty lookup | SARS Customs & Excise Tariff Book | https://www.sars.gov.za/customs-and-excise/tariff/ |
| HS / duty rulings | SARS Tariff Determinations | https://www.sars.gov.za/legal-counsel/secondary-legislation/tariff-determinations/ |
| UN transport of cells | UN Manual of Tests and Criteria, Rev. 7 (Section 38.3, lithium batteries) | https://unece.org/transport/standards/un-manual-tests-and-criteria |
Source-allow-list note (R2 fix): "tradvolt.com RFQ roll-up" and "EPC partner quotes" were internal commercial data and have been removed. The standalone "DMRE BESS cost benchmark study, 2023" reference was replaced with the equivalent CSIR publication, which is publicly accessible. BloombergNEF is referenced via its public methodology / press note URLs; the paywalled full report is not cited as a free resource.