# Business Model # Business Models for the NFE Microgrid OS > \[!info\] Current working basis — August 2026 > **Tariff:** UEDCL Code 10.2, Q3 2026: Peak 666.5 / Shoulder 562.1 / Off-Peak 429.7 UGX/kWh. The full tariff schedule must remain a configurable model input because ERA/UEDCL rates change periodically. **Sezibwa base case:** zero export value until site eligibility, interconnection terms, and the applicable export credit are confirmed. All commercial offers should be re-run with site-specific load, tariff, capex, financing, and reliability assumptions before commitment.

**Update, October 2026: Commercial RPMs.** For commercial facilities NFE has chosen the NFE-owned energy-service model, with reliability included in a utility-matched tariff (no separate reliability fee) and a 50% minimum monthly payment. That differs from the "explicit reliability tier" and "partner-led first" guidance below, which came from the Sezibwa community case. See [NFE Commercial Reliable Power Microgrids](https://bookstack.nearlyfreeenergy.com/link/146) and the [Commercial RPM decision log](https://bookstack.nearlyfreeenergy.com/link/187).

# Purpose NFE Microgrid OS needs a business model that creates durable recurring revenue while improving the economics and reliability of community energy systems. Two ownership models remain viable, but the current Sezibwa evidence suggests a clear sequencing principle: **scale first through partner-owned deployments, while using NFE-owned projects selectively where the project return and strategic value justify tying up capital.** The operating system is the common product in both models. Ownership determines who funds the assets, carries performance risk, and receives the energy margin. # Current evidence from Sezibwa The Sezibwa analysis is not a universal tariff card, but it gives NFE a grounded reference case for product and commercial decisions:
QuestionCurrent findingBusiness implication
What should be added to the existing inverter and 8.9 kWh battery?Solar first. Adding battery at today's price lowers NPV.The OS should recommend marginal investments based on the assets already installed, not sell a standard hardware bundle.
How much solar?**4 kWp is the low-spill efficiency sweet spot;** **5 kWp is the maximum-NPV optimum.**The recommendation depends on the customer's objective: capital efficiency and minimal curtailment versus maximum total value.
What does battery arbitrage contribute?About 22,000–40,000 UGX/month before fully pricing degradation, depending on backup reserve.Arbitrage is incremental value from sunk hardware, not the core investment case. Dispatch must exceed marginal degradation cost and preserve the agreed reserve.
When does more storage pay?At the 5 kWp anchor, the first added unit breaks even around 367,000 UGX/kWh, versus about 626,000 UGX/kWh today.Treat battery price as a procurement trigger. Additional storage may still be justified separately for outage resilience.
Can surplus solar be sold?Not assumed in the Sezibwa base case.Use zero export value until site-specific arrangements are confirmed; show export as a separate upside scenario.
Detailed evidence: [Solar Plus Battery Economic Case](https://bookstack.nearlyfreeenergy.com/link/136), [Battery Arbitrage Economic Case](https://bookstack.nearlyfreeenergy.com/link/137), [Optimal Solar and Battery Allocation](https://bookstack.nearlyfreeenergy.com/link/138), and [Battery Price Break-Even](https://bookstack.nearlyfreeenergy.com/link/139). --- # 1. NFE-owned energy-service model NFE finances and owns the supported microgrid assets and sells an energy and reliability service to a landlord, estate, community entity, or other contracted counterparty. The exact structure—PPA, lease, energy-service agreement, or another permitted form—must be confirmed against the applicable licensing, tariff, consumer-protection, and metering requirements. ## Assets and responsibilities NFE funds, owns, and operates the contracted system: submeters, solar, batteries where justified, inverters, communications, and the NFE Microgrid OS control and billing layer. NFE carries financing, asset-performance, collections, maintenance, and technology-obsolescence risk. ## Revenue and value capture - **Energy-service revenue:** contracted charges for electricity supplied or bills avoided, subject to the approved commercial structure. - **Reliability service:** an explicit availability or backup tier rather than an assumed premium hidden inside the energy price. - **Operating efficiency:** lower losses, better collections, solar self-consumption, reserve-aware dispatch, and tariff-aware control increase project margin. - **Optional services:** installation, maintenance, reporting, or financing fees where separately contracted. Optimization is not revenue by itself. It improves the margin or service quality of a contracted energy product. The commercial model should therefore distinguish customer charges from internally generated savings. ## Investment test Approve an NFE-owned project only when the site-specific model shows acceptable NPV, IRR, downside resilience, and debt-service capacity after turnkey capex, losses, degradation, maintenance, collections, taxes, and financing. At Sezibwa, the attractive next investment is solar on hardware already owned; that result should not be generalized to a greenfield site where inverter and battery costs are not sunk. ## Best fit - High-confidence load and collections data. - A creditworthy counterparty or enforceable payment mechanism. - Strong solar self-consumption and limited dependence on unconfirmed export revenue. - A strategically important reference site or a return high enough to justify NFE capital. ## Key tradeoff This model captures the most project economics and operational learning, but it is capital-intensive and concentrates construction, credit, regulatory, and long-term asset risk on NFE. --- # 2. Partner-owned managed-platform model A developer, landlord, estate, community entity, or energy-service partner finances and owns the physical assets. NFE supplies the operating platform, supported controls, deployment services, and ongoing technical operations. ## Partner role The partner funds and owns solar, batteries, submeters, inverters, communications hardware, and other site infrastructure. The partner retains asset-financing risk and receives the energy margin, subject to the commercial agreement. ## NFE role - NFE Microgrid OS licensing and supported device integration. - Metering, customer accounts, billing, collections data, and settlement reporting. - Monitoring, alerts, remote operations, and performance reporting. - Tariff-aware, forecast-aware, reserve-aware, and degradation-aware dispatch. - Site modelling, investment recommendations, commissioning, and lifecycle support. ## NFE revenue - **Deployment fee:** site assessment, modelling, integration, commissioning, and training. - **Recurring platform fee:** a base site fee plus a usage driver such as connected meters, managed assets, or energy settled. - **Managed-operations fee:** optional monitoring, dispatch, incident response, reporting, and maintenance coordination. - **Performance component:** only where savings can be baselined and audited; use a share of verified incremental value rather than gross electricity sales. A pure revenue share is easy to describe but difficult to govern when tariffs, load, collections, outages, and partner capex all change. A fixed recurring fee provides predictable platform revenue; a narrowly defined performance component can align incentives without making NFE responsible for every movement in the partner's electricity income. ## Hardware support and maintenance NFE should publish a supported-hardware list and define responsibility by asset. Partner-owned equipment remains the partner's financial responsibility. NFE may provide maintenance directly or coordinate approved providers under a separate service level. Unsupported third-party assets, including generators, are excluded unless explicitly integrated and contracted. ## Key tradeoff This model produces less revenue per site than ownership but requires much less NFE capital, limits balance-sheet exposure, and can scale faster across heterogeneous sites. --- # Common modelling and operating rules 1. **Keep tariffs configurable.** Store the entire time-of-use schedule, effective dates, fixed charges, taxes, and export credit separately from model logic. Re-run forecasts whenever ERA/UEDCL publishes a revision. 2. **Model the actual site.** Use interval load, outages, collections, existing assets, weather, and installed quotations. Do not transfer Sezibwa results unchanged to another microgrid. 3. **Separate sunk and new capital.** Dispatching an owned battery can create incremental value; buying another battery requires a separate investment case. 4. **Separate energy and resilience value.** Energy NPV should stand on its own. Backup reserve, outage avoidance, and service availability should be quantified and priced explicitly where possible. 5. **Use a Sezibwa-specific zero-export-value base case.** Add export value only as a documented scenario with confirmed eligibility, interconnection limits, and credit terms. 6. **Price degradation at the margin.** Arbitrage should run only when avoided tariff cost exceeds conversion losses, marginal battery wear, and the opportunity cost of reducing backup reserve. 7. **Treat second-life storage as a hypothesis.** Recycled batteries may become attractive, but require a second-life-specific techno-economic model covering acquisition, testing and grading, BMS and pack integration, safety, warranty, heterogeneous health, remaining life, and replacement. --- # Strategic recommendation
SituationPreferred approach
Partner has capital and wants operating capabilityPartner-owned managed platform
Site has strong verified economics, reliable collections, and strategic valueConsider NFE ownership after investment committee review
Demonstration or product-learning siteNFE-owned or co-invested, with an explicit learning budget
Economics depend on unconfirmed export revenue or speculative battery pricingDo not finance on the upside case; validate first
The near-term portfolio should therefore be **partner-led and platform-first**, with a small number of deliberately selected NFE-owned or co-invested sites. This builds recurring software and operations revenue, expands the data set behind the optimizer, and preserves NFE capital for projects where ownership offers exceptional return or strategic learning. # Decisions and evidence still required 1. Define the minimum recurring platform fee and its pricing driver: site, meter, managed asset, or energy settled. 2. Define a measurable baseline and audit method before offering performance-based fees. 3. Set NFE-owned project hurdle rates, concentration limits, credit requirements, and downside tests. 4. Confirm the permitted contracting and electricity-resale structures for each customer segment. 5. Publish supported hardware, warranty boundaries, maintenance responsibility, and service levels. 6. Build the second-life-specific techno-economic model before presenting recycled batteries as a validated strategy. 7. Turn the current optimizer into a repeatable site proposal that reports energy value, resilience value, capex, financing, NPV, IRR, payback, tariff sensitivity, and export sensitivity separately. # Solar Plus Battery Economic Case (Sezibwa) ## Data provenance Underlying data: the aggregate meter meter_100 plus the 10 customer sub-meters, over **2026-03-19 to 2026-08-10**, from the modbus energy-logger archive (sources: pi-direct and nextcloud-import), i.e. the pre- and early-OpenEMS period. Load basis is a conservative 21.3 kWh/day. Note (2026-08-20): the site has since migrated to **OpenEMS**, which is now the live source. OpenEMS-sourced daily figures from mid-August show inconsistencies (the aggregate roughly doubled exactly at the source switchover, and on some days the customer-meter sum exceeds the aggregate), so they are treated as a data-quality issue to resolve before use; this analysis stays on the validated archive. Phase assignments reflect the wiring as observed up to about May 2026 and should be re-verified against the current OpenEMS setup if any connections have been changed. # Battery Arbitrage Economic Case (Sezibwa) ## Data provenance Underlying data: the aggregate meter meter_100 plus the 10 customer sub-meters, over **2026-03-19 to 2026-08-10**, from the modbus energy-logger archive (sources: pi-direct and nextcloud-import), i.e. the pre- and early-OpenEMS period. Load basis is a conservative 21.3 kWh/day. Note (2026-08-20): the site has since migrated to **OpenEMS**, which is now the live source. OpenEMS-sourced daily figures from mid-August show inconsistencies (the aggregate roughly doubled exactly at the source switchover, and on some days the customer-meter sum exceeds the aggregate), so they are treated as a data-quality issue to resolve before use; this analysis stays on the validated archive. Phase assignments reflect the wiring as observed up to about May 2026 and should be re-verified against the current OpenEMS setup if any connections have been changed. # Optimal Solar and Battery Allocation (Sezibwa) ## Data provenance Underlying data: the aggregate meter meter_100 plus the 10 customer sub-meters, over **2026-03-19 to 2026-08-10**, from the modbus energy-logger archive (sources: pi-direct and nextcloud-import), i.e. the pre- and early-OpenEMS period. Load basis is a conservative 21.3 kWh/day. Note (2026-08-20): the site has since migrated to **OpenEMS**, which is now the live source. OpenEMS-sourced daily figures from mid-August show inconsistencies (the aggregate roughly doubled exactly at the source switchover, and on some days the customer-meter sum exceeds the aggregate), so they are treated as a data-quality issue to resolve before use; this analysis stays on the validated archive. Phase assignments reflect the wiring as observed up to about May 2026 and should be re-verified against the current OpenEMS setup if any connections have been changed. # Battery Price Break-Even (Sezibwa) ## Data provenance Underlying data: the aggregate meter meter_100 plus the 10 customer sub-meters, over **2026-03-19 to 2026-08-10**, from the modbus energy-logger archive (sources: pi-direct and nextcloud-import), i.e. the pre- and early-OpenEMS period. Load basis is a conservative 21.3 kWh/day. Note (2026-08-20): the site has since migrated to **OpenEMS**, which is now the live source. OpenEMS-sourced daily figures from mid-August show inconsistencies (the aggregate roughly doubled exactly at the source switchover, and on some days the customer-meter sum exceeds the aggregate), so they are treated as a data-quality issue to resolve before use; this analysis stays on the validated archive. Phase assignments reflect the wiring as observed up to about May 2026 and should be re-verified against the current OpenEMS setup if any connections have been changed. # Solar Sizing Re-run, Sept 2026 (600 Wp panels) This page re-runs the solar sizing from [Solar Plus Battery Economic Case (Sezibwa)](https://bookstack.nearlyfreeenergy.com/link/136) on three months of live OpenEMS load. It answers one question: is 7 to 8 panels (600 Wp each) still the best PV size for Sezibwa, and does it cut the energy bill in half? ## Answer Yes. With 600 Wp panels: - 6 panels (3.6 kWp) cut the energy bill by half. - 7 panels (4.2 kWp) cut it by 55%. - 8 panels (4.8 kWp) cut it by 58% and give the best 20 year return (highest NPV). Past about 5 kWp each extra panel adds little. The battery is already full by midday, and without grid export the surplus is wasted. ## What changed since August | Input | August model | This re-run | |---|---|---| | Load data | Modbus logger archive, Mar to Aug 2026 | OpenEMS aggregate meter DTSU666 (serial 221123297561), 1 Jul to 28 Sep 2026, 76 days of data | | Average load | 21.3 kWh/day | 25.8 kWh/day | | Baseline energy bill | about 374,000 UGX/month | about 454,000 UGX/month | | Panel size | kWp only | 600 Wp panels | The model code and economics are otherwise unchanged. One correction was made: the daily average now divides by the hours actually covered by data, not the calendar span, so data gaps (for example the site internet outage of 25 to 26 Sep) no longer understate the load. ## Cost of an 8 panel install (at cost) Priced from the Rincol catalogue (August 2026 buy prices), with the railing at 421,000 UGX per 4 panel kit: | Item | Qty | Cost (UGX) | |---|---|---| | 600 Wp mono panel (GS600N-144M) at 320,250 | 8 | 2,562,000 | | Railing kit (4 panels per kit) at 421,000 | 2 | 842,000 | | 6 mm2 UV solar cable at 12,000/m | 40 m | 480,000 | | PV combiner box (MC4) | 1 | 80,000 | | DC circuit breaker 20 A | 1 | 25,000 | | Surge protector (a DC rated one is still to be sourced) | 1 | 35,000 | | Earthing (rod, clamps, conductor) | 1 | 80,000 | | MC4 connectors (estimate) | 6 pairs | 48,000 | | Labour and transport | 1 | 250,000 | | **Total** | | **4,402,000** | Panels and railing are 3.4M, and the rest of the balance of system and labour adds about 1.0M. The inverter (3.84M) and battery (5.57M) are already paid for, so these are the only new costs. ## Results (600 Wp panels, no grid export, at cost) | Panels | kWp | Saving per month (UGX) | Bill cut | Solar wasted | Cost (UGX) | Simple payback | IRR | NPV at 15% (UGX) | |---|---|---|---|---|---|---|---|---| | 5 | 3.0 | 199,406 | 44% | 4% | 3.44M | 1.4 yr | 69% | 10.9M | | 6 | 3.6 | 225,904 | 50% | 10% | 3.76M | 1.4 yr | 71% | 12.4M | | 7 | 4.2 | 247,923 | 55% | 16% | 4.08M | 1.4 yr | 72% | 13.6M | | **8** | **4.8** | **263,462** | **58%** | **22%** | **4.40M** | **1.4 yr** | **71%** | **14.3M** | | 9 | 5.4 | 273,637 | 60% | 28% | 5.14M | 1.6 yr | 63% | 14.3M | | 10 | 6.0 | 281,205 | 62% | 33% | 5.46M | 1.6 yr | 61% | 14.5M | | 11 | 6.6 | 287,350 | 63% | 38% | 5.78M | 1.7 yr | 59% | 14.6M | | 12 | 7.2 | 292,746 | 64% | 42% | 6.10M | 1.7 yr | 56% | 14.6M | Cost is built per panel count: panels, one railing kit per 4 panels, and a fixed 40 m of cable, combiner, breaker, surge protector, earthing, connectors and labour. 4 panels is left out because a 4 panel string (about 175 V) is below the inverter's 200 V MPPT minimum. The SRNE ASP48120SH3 has 2 MPPT inputs of up to 9 kW each at 200 to 650 V, so 6 to 14 panels fit on one string. A 5 panel string (about 224 V at 25 °C) drops to about 194 V on hot afternoons, just under the minimum, so the 5 panel row in the table above is optimistic. For a turnkey quote at Rincol selling prices, add roughly 20 to 25% (the August 7 panel quote was 5.02M at sell prices against 4.16M at cost). Battery only (the existing SRNE, no panels) saves about 34,000 UGX/month (8%) on the model's arbitrage policy. ## How to read it - **Bill cut**: the share of the energy bill (energy charges only, excluding the fixed service charge and VAT) that solar plus the existing battery removes. - **Solar wasted**: the share of generated solar energy that can't be used or stored, because the battery is full and there is no export. - **Extra per panel**: saving rises by about 35,000 UGX/month for the 5th panel, but only about 10,000 UGX for the 9th and under 6,000 by the 12th. The 9th panel also needs a third railing kit. - **NPV at 15%**: the 20 year value after paying the install cost. It climbs to 14.3M at 8 panels and then flattens (14.6M at 12). Panels 9 to 12 add about 1.7M of cost for almost no extra value, and IRR is highest at 7 to 8 panels. ## Recommendation Install **8 panels of 600 Wp (4.8 kWp)** on the existing SRNE inverter and battery, for about 4.4M UGX at cost. That cuts the energy bill by 58% (about 263,000 UGX/month) and pays back in about 1.4 years. 7 panels (4.1M, 55%) is the cheaper option. Going above 8 only makes sense with grid export (net metering) or more battery storage. ## Assumptions and caveats - **Weather** is modelled, not measured: 5.2 peak sun hours, 0.78 performance ratio, and a seeded day to day variation. Measured site irradiance would firm this up. - **Tariff** is UEDCL Code 10.2 Q3 2026 (off peak 429.7, shoulder 562.1, peak 666.5 UGX/kWh). The Q4 schedule takes effect on 1 Oct and should be checked. - **Capex** uses August 2026 Rincol catalogue buy prices and the 421,000 railing kit. The catalogue has no DC rated PV surge protector yet (the 35,000 line is the AC one), so budget more for that. Confirm current prices before buying. - **Load** covers July to September only, so there is no seasonal variation. - **Battery use**: with solar, the battery is charged from solar and discharged through the shoulder and peak periods. It replaces the current grid arbitrage (off peak charge, 21:30 to 23:30 discharge). A 30% reserve is held for backup. - **Per phase limit**: the inverter delivers at most 4 kW per phase on battery (see the 27 Sep outage report). Solar does not change that limit. ## Reproduce The model is versioned in the **nfe-modbus-energy-logger** repository under `scripts/solar_sim/` (issue #89, [PR #90](https://github.com/Nearly-Free-Energy/nfe-modbus-energy-logger/pull/90), in review). From the repo root: ``` python3 scripts/solar_sim/sweep_bom.py scripts/solar_sim/m100_openems.csv python3 scripts/solar_sim/nfe_solar_model.py scripts/solar_sim/m100_openems.csv ``` `m100_openems.csv` holds the site aggregate energy in 15 minute steps (site totals only, no per customer data). # Sezibwa solar purchase: final Muden quote (19 Oct 2026) # Sezibwa (Nansana) solar purchase: final quote, 19 Oct 2026 The final proforma invoice from Muden Systems for the 8-panel string, as sent in the "NFE Nansana panel installation" WhatsApp group on 19 Oct 2026. The PDF is attached to this page. The design is in [Solar Sizing Re-run, Sept 2026](https://bookstack.nearlyfreeenergy.com/link/149) and the discussion is in the Element solar thread (business room, "Sezibwa solar: where we are"). **Quote:** MS/SOLAR/10-26, dated 19 Oct 2026, addressed to Nearly Free Energy, valid 15 days or while stock lasts. Delivery to Nansana. | # | Item | Qty | Unit price (UGX) | Amount (UGX) | Warranty | |---|---|---|---|---|---| | 1 | Delivery cost | 1 | 100,000 | 100,000 | | | 2 | 600 W monocrystalline panel | 8 | 370,000 | 2,960,000 | 10 years | | 3 | Aluminium rail 5100 mm | 4 | 140,000 | 560,000 | | | 4 | End clamp | 8 | 7,000 | 56,000 | | | 5 | Inter clamp | 12 | 8,000 | 96,000 | | | 6 | L-feet | 12 | 12,000 | 144,000 | | | 7 | 32 A 4-pole DC isolator | 1 | 320,000 | 320,000 | | | 8 | 1000 V DC surge protector | 1 | 110,000 | 110,000 | | | 9 | MC4 connector | 4 | 15,000 | 60,000 | | | | **Sub total** | | | **4,406,000** | | The invoice's grand total is UGX 4,406,000. The lines add up correctly. ## Not on the invoice - **PV cable.** The earlier quote (4,426,000) included 30 m of 6 mm² cable at 360,000. This one does not, so the cable is bought separately (the 4 mm² cable was out of stock). - **Installation labour.** The line is blank (labour via Green Volta, separate). - **VAT.** The 18% line is blank, so VAT on the invoice is not shown. ## Changes from the earlier quote (7 Oct) - The 2-pole 32 A 500 V DC breaker (80,000) became a **32 A 4-pole DC isolator (320,000)**. For a single string of 8 panels (Voc about 422 V, about 437 V cold) a 2-pole device rated for the voltage is normally enough; a 4-pole isolator is often wired as two poles in series for extra breaking capacity. Worth confirming what was intended. - The 1000 V surge protector (150,000 earlier) is now 110,000. - Delivery is a separate line (100,000). ## Terms (from the invoice) - Warranty: 10 years on panels; 1 year manufacturing defect warranty on the rest, on a back-to-base basis. No warranty against misuse, surges or acts of God. - Title of goods stays with Muden Systems until fully paid. - Validity: 15 days or until stock runs out. ## To do on delivery - Check every line against this table, including panel count and rating. - Take photos of labels (panel model, serials) for the warranty record. - Confirm cable, labour and VAT with the supplier and the installer.