Commercial RPM economics and bankability The numbers behind the Commercial RPM: where NFE's margin comes from, the Worship Harvest Phase 2 model, the customer-size threshold, and how a project can be financed. These are internal screening figures, not quotes or investment commitments. Model built 18 Sep 2026; terms updated to 21 Oct 2026. Owner: Aaron Tushabe. Summary NFE charges the utility tariff, so its margin is the grid electricity it no longer has to buy : solar used directly in the daytime, and solar stored for the 18:00–24:00 peak. The margin sits inside the customer's existing price, not on top of it. Worship Harvest base case (120 kWh / 50 kW / 40 kWp; CAPEX modelled at UGX 142.8m): revenue UGX 47.0m/year, avoided grid purchases UGX 34.8m/year, cash flow available for debt service (CFADS) UGX 24.5m/year, project return (IRR) about 15% over 15 years. Financing works with shilling debt at 12% or less over 10 years or more (debt service cover, DSCR, of 1.6–1.9x). At ordinary commercial rates of about 18–19%, DSCR falls to about 1.25x and the equity return to about 10%. First projects: 50–55% debt , a 6-month debt service reserve, and a 15-year customer contract. Later, pool sites into a portfolio SPV. Terms are defined in the NFE Commercial RPM Finance Glossary . 1. Where the margin comes from RPM margin per kWh = (customer revenue − grid purchases − RPM operating costs) ÷ kWh sold Grid electricity NFE passes through at the same tariff earns nothing. Solar used directly replaces shoulder-rate grid electricity (UGX 562.1/kWh). Solar stored in the battery and used in the evening replaces peak-rate electricity (UGX 666.5/kWh), after about 10% round-trip losses. The battery cycles every day, so reliability, solar shifting and peak avoidance all come from the same asset. NFE has to be the billing layer for the facility's electricity. If the customer kept paying UEDCL for normal grid power and bought only backup power from NFE, this margin would not exist. Charging the battery from the grid off-peak to sell at peak adds a little, but it is not the investment case (see Business Models for the NFE Microgrid OS ). 2. Model structure Customer → Load → System → Dispatch → Economics → Financing Customer: tariff, monthly bill, generator costs, outage frequency and cost, credit quality Load: annual and monthly kWh, peak/shoulder/off-peak split, maximum kW, critical kW, interval profile System: solar kWp, battery kWh and kW, inverter kW, CAPEX, useful life Dispatch: solar used directly, battery throughput, grid and generator kWh, curtailed solar, unserved energy Economics: revenue, energy costs, O&M, replacement reserve, CFADS, project IRR Financing: debt share, rate and tenor, DSCR, debt service reserve, equity IRR, buyout schedule There is no shared spreadsheet yet; the calculations so far were done in Aaron's ChatGPT working session. 3. Base assumptions Assumption Base value Validate with Customer tariff Applicable Code 10.2 time-of-use tariff. Q3 2026, before VAT: peak 666.5, shoulder 562.1, off-peak 429.7 UGX/kWh. Q4 2026 reportedly unchanged. ERA schedule at each proposal date Load (Worship Harvest) 82,940 kWh/year; 29% peak / 52% shoulder / 19% off-peak 12 months of bills and Phase 1 interval data Solar yield 1,500 kWh/kWp/year (a Soroti solar plant averaged ~1,670) Site-specific PV simulation Battery 90% round-trip efficiency; 90% depth of discharge Supplier datasheet and warranty Inverter efficiency 95% Datasheet Reliability reserve 15% on battery sizing SLA definition Routine O&M UGX 2.14m/year (~1.5% of CAPEX) O&M quotes NFE monitoring, billing and admin UGX 1.2m per site per year NFE operating costs Battery and inverter replacement reserve UGX 6.91m/year (~5% of CAPEX) Warranty terms, cycle life, replacement prices CAPEX UGX 120–145m (model run at 142.8m) At least two EPC/vendor quotes Contract term 15 years; no residual value; no export income Contract Tariff escalation None (real terms) Indexation clause Target project return 15% Investment decision 4. Worship Harvest results Energy flows (40 kWp) Flow kWh/year Customer consumption 82,940 Solar production (40 kWp × 1,500) 60,000 Peak load supplied through the battery 24,207 Solar used directly against shoulder load ~33,100 Grid purchases replaced ~57,300 Grid purchases remaining ~25,600 Share of consumption met by solar and battery ~69% Cash flow (bankability model, 40 kWp) Item UGX m/year Customer energy revenue, before VAT 47.0 Grid electricity NFE still buys (12.3) Avoided grid purchases (gross margin) 34.77 Routine O&M (2.14) NFE monitoring, billing and admin (1.20) Battery and inverter replacement reserve (6.91) CFADS 24.52 CFADS per kWh sold ~UGX 296 Project IRR, 15 years, before financing ~15.1% Solar size sensitivity (first, lighter model) Solar CAPEX (UGX m) Annual contribution (UGX m) Margin per kWh (UGX) 15-year IRR 32 kWp 135.5 23.4 282 15.2% 36 kWp 139.1 26.7 322 17.5% 40 kWp 142.8 30.0 361 19.6% ~46 kWp 148.6 35.2 425 22.6% The first model combined O&M and replacements at 2.5% of CAPEX. The bankability model funds battery and inverter replacement explicitly, which lowers the 40 kWp return from about 19.6% to about 15.1%. The "around 20%" return discussed at the 19 Sep check-in came from the first model. 5. Customer-size threshold The model was scaled to hypothetical customers with Worship Harvest's load shape, sizing the battery, inverter and solar for each and checking the 15-year return (first, lighter model): Monthly consumption Result at the 15% target 1.0 MWh Below target 1.25 MWh Below target ~1.5 MWh About 15% (break-even ~1,490 kWh/month) 2.0 MWh Attractive (for example ~30 kWh battery, 15 kW inverter, ~12 kWp solar) 3.0 MWh Strong 5.0 MWh Very good Downside case Break-even consumption Base ~1.49 MWh/month Solar yield 1,350 kWh/kWp ~1.68 CAPEX +15% ~2.09 Required return 20% ~2.24 Reliability reserve 30% ~1.64 Site overhead UGX 2m/year ~1.81 Several reasonable downside cases land near 2 MWh/month, which is why it became the screening threshold. Caveat: these runs used the lighter replacement assumption. Re-run them with the bankability assumptions and the 50% minimum payment before treating 2 MWh/month as firm. 6. Financing Structures compared CAPEX UGX 142.8m, CFADS UGX 24.52m/year, 10-year loan with level annual repayments: Debt share Rate Debt service (UGX m/year) DSCR Equity IRR 60% 12% 15.16 1.62x ~15.7% 60% 18.7% (commercial bank) 19.54 1.25x ~10% 50% 18.7% (commercial bank) 16.28 1.51x ~11% 60% 10% 13.94 1.76x ~17.4% 60% 8% 12.77 1.92x ~19.2% 55% 12% 13.90 1.76x ~15.6–16.0% 50% 12% 12.64 1.94x Not calculated Equity IRR treats the 6-month debt service reserve as cash tied up for the term. In the 60% / 12% case, LLCR is about 1.62x and PLCR about 1.95x. Rates context: Bank of Uganda reported weighted shilling lending rates of about 18.7% in Q4 2025, and shilling loans averaged 17.9% in August 2026. On 14 Oct 2026 the Central Bank Rate rose from 9.75% to 12% ( Macro outlook ), so commercial debt is likely dearer now. That strengthens the case for concessional finance. Volume risk (60% debt at 12%) Useful energy vs today CFADS (UGX m) DSCR Base 24.52 1.62x −5% 22.78 1.50x −10% 21.04 1.39x −15% 19.30 1.27x −20% 17.56 1.16x A lender wanting DSCR of at least 1.30x needs about 86% of today's throughput (about 5.95 MWh/month). At 50% debt, a 20% drop still leaves about 1.39x. The 50% minimum monthly payment (agreed 6 Oct) adds a revenue floor. This table hasn't been re-run with it yet. Recommended structure for the first projects Item Position Debt 50–55% of CAPEX, shilling, 10 years or more, 12% or less Equity (NFE and investors) 45–50% Debt service reserve (DSRA) 6 months of debt service Customer contract 15 years Customer deposit About 2 months of electricity bills Minimum payment 50% of the reference monthly energy charge (agreed 6 Oct) Early termination Payment covering outstanding debt, demobilisation and a defined share of unrecovered investment Export income None in the base case At 55% debt: debt UGX 78.5m, equity plus DSRA about UGX 71m, debt service UGX 13.9m/year, DSCR 1.76x. Capital stack to aim for 50–60% concessional or project debt at 8–12% in shillings, over 8–10 years or more 30–40% NFE and investor equity 0–20% climate grant, guarantee or first-loss capital One candidate is Uganda Development Bank's Climate Finance Facility, which describes concessional debt, equity, guarantees and grants, and supports solar mini-grid projects. Once 5–10 sites have an operating history, pool them in a portfolio SPV (for example 20 sites × UGX 150m ≈ UGX 3bn) with portfolio-level DSCR and a central reserve. Lenders prefer that to financing one site at a time. 7. Ownership option and buyout The year-5 purchase price must at least repay the outstanding senior debt and any prepayment costs, plus the NFE residual value agreed at signing. Outstanding debt in the 60% / 12% / 10-year case: End of year Senior debt outstanding 3 UGX 69.2m 5 UGX 54.7m 8 UGX 25.6m 10 0 8. Cash-flow waterfall The customer pays NFE for metered electricity (or the minimum payment). NFE (or the SPV) pays for grid electricity and other energy costs. Routine O&M and operating costs are paid. The battery and inverter replacement reserve is funded. What remains is CFADS. Scheduled senior debt service is paid. The debt service reserve is topped up if needed. The rest goes to NFE and equity investors. 9. Risks and mitigations Risk Mitigation Customer uses less electricity 50% minimum monthly payment; lower debt on early sites; credit review; early termination payment Tariff changes Mirror the Code 10.2 structure; define indexation in the contract Late or missed payments Payment history, deposit, remote metering, clear default procedures, a diversified portfolio Battery and inverter replacement Explicit reserve, manufacturer warranties, standard equipment, remote health monitoring Solar or storage underperforms Conservative yield, commissioning tests, reserve margins Peak demand underestimated Phase 1 metering after the grid/generator changeover; size the inverter from measured kW Regulatory ERA certificate of registration for the microgrid before operation Construction Competitive quotes, fixed-price scope where possible, acceptance tests, EPC warranties Currency Equipment is priced in dollars; the shilling has fallen about 12% this year ( Macro outlook ) 10. Equipment cost benchmarks (Sep–Oct 2026) Item Price Source and caveat 61.44 kWh high-voltage LFP battery (Felicity) ~UGX 32.77m incl. VAT Uganda retail listing, Sep 2026 50 kW three-phase hybrid inverter (Felicity) ~UGX 14–15m Uganda retail listings, Sep 2026 580 W solar module (Felicity) ~UGX 405,000 (~UGX 700/W) Uganda retail listing, Sep 2026 600 W mono solar module UGX 370,000 (~UGX 617/W), VAT not shown Muden Systems quote, 19 Oct 2026 ( page ) Packaged 120 kWh / 50 kW / 55 kWp C&I system ~UGX 144.69m incl. VAT Uganda retail listing, Sep 2026 Packaged 60 kWh / 50 kW / 26.4 kWp system ~UGX 96m Uganda retail listing, Sep 2026 Protection, changeover, mounting, cabling, switchgear ~UGX 8–12m Estimate NFE metering, controls and installation ~UGX 6–9m Estimate SRNE equipment for a ~120 kWh / 60 kW / 40 kWp system ~US$30,750, before the remaining installation costs NFE's SRNE wholesale price list (Uganda), 23 Jul 2026 These are budget-brand retail prices. Hillary pointed out on 12 Sep that commercial customers and EPCs may expect SMA- or BYD-class equipment, so get quotes for that too. Dollar-priced equipment follows the shilling within weeks. Installed planning allowances used for the Oct 2026 funding target: about US$15,000 for a smaller commercial site (30–40 kWh), US$25,000 for a medium site (60–80 kWh) and US$45,000 for a Worship Harvest-scale site (~120 kWh), each including solar, inverters and installation. Worship Harvest is one of the larger sites, not the typical one. See Funding target . How the Worship Harvest CAPEX moved: UGX 118m in the first estimate (~28 kWp) → UGX 135.5m (32 kWp) → UGX 142.8m (40 kWp, with more conservative installation, project and contingency allowances). Present it as UGX 120–145m until EPC quotes arrive. 11. Before an investment decision Replace CAPEX with at least two EPC quotes, including an industrial-grade option 12 months of bills plus Phase 1 interval data, including generator-supplied load Site-specific solar simulation Lender term sheets (rate, tenor, minimum DSCR, reserve requirements) Re-run the threshold and volume-risk tables with the 50% minimum payment Move from annual averages to a monthly or hourly dispatch model Never quote the IRR, DSCR or 2 MWh threshold without the assumptions behind them; they are outputs of this model, not constants Source: Aaron's ChatGPT working session " Commercial RPM Modeling " (18 Sep–21 Oct 2026). Last reviewed 21 October 2026 by Aaron Tushabe.