Optimal Solar and Battery Allocation (Sezibwa)

Data provenance

[!info]Underlying Conservative,data: bank-facingthe basis.aggregate Tariff:meter UEDCLmeter_100 Codeplus 10.2the 10 customer sub-meters, over Q32026-03-19 2026to 2026-08-10, from the modbus energy-logger archive (Peaksources: 666.5pi-direct /and Shouldernextcloud-import), 562.1i.e. /the Off-Peakpre- 429.7and UGX/kWh).early-OpenEMS Theperiod. scheduleLoad basis is a configurableconservative model21.3 inputkWh/day.

for

Note future(2026-08-20): ERA/UEDCLthe revisions.site Turnkeyhas capex,since realmigrated weather.to ReproducibleOpenEMS, which is now the live source. OpenEMS-sourced daily figures from nfe_solar_model.py.

mid-August

show

Headline

Sweeping both dimensionsinconsistencies (solarthe sizeaggregate roughly doubled exactly at the source switchover, and batteryon capacity)some days the customer-meter sum exceeds the aggregate), so they are treated as a data-quality issue to findresolve thebefore bestuse; 20-yearthis returnanalysis gives a clear, slightly counterintuitive answer:

The most business-optimal allocation is solar-heavy and battery-light. Put the money into panelsstays on the batteryvalidated alreadyarchive. owned.Phase Addingassignments battery capacity lowersreflect the return at today's battery price.

• Sezibwa zero-export-value base case: about 5 kWp on the existing 8.9 kWh battery. Turnkey capex about 6.3M UGX, IRR about 42 percent, simple payback about 2.3 years, discounted payback about 3.1 years, 20-year NPV about 9.8M at 15 percent.
• With a usable export arrangement: about 8 kWp on the same battery. Capex about 10.1M, IRR about 46 percent, NPV about 18.5M. (This scenario assumes surplus is credited at 343 UGX/kWh — a deliberately conservative, below-retail rate consistent with how net-metering credits typically pay; the real credit value under ERA's 2024 regulations is unconfirmed for this site.)
• In both, buying a second or third battery unit lowers NPV.

Method

A 2-D sweep over solar (0 to 8 kWp) and battery (existing one unit, or plus one or two SR-SE10B units). For each combination the model runs the real load through 20 years and computes discounted savings minus turnkey capex.

Assumption	Value
Tariff	UEDCL Code 10.2 Q3 2026: Peak 666.5 / Shoulder 562.1 / Off-Peak 429.7 UGX/kWh (configurable schedule)
Horizon	20 years
Battery aging	usable capacity fades linearly to 50 percent by year 20
Weather	seeded variable-irradiance year
Discount rate	shown at 10 / 15 / 20 / 25 percent
Solar capex	about 1.26M UGX per kWp turnkey
Extra battery capex	5.568M UGX per SR-SE10B (8.9 kWh usable)
Existing inverter + 1 battery	sunk (9.4M already spent)

Result

The greenest cells sit on the bottom row, the battery NFE already owns. Moving up (more battery) turns the map redder at every solar size.

[!note] 4 kWp versus 5 kWp 4 kWp is the low-spill efficiency sweet spot: it captures most of the available bill reduction with almost no curtailed solar. 5 kWp is the maximum-NPV optimum: the fifth kWp introduces some spill, but its remaining useful generation is still worth slightly more than its marginal installed cost. The two sizes optimize different objectives.

Investment metrics and sensitivity

NPV shown across discount rates so you can see which allocation survives expensive capital (metrics ranked NPV, IRR, discounted payback, simple payback, then cumulative ROI):

Allocation	Capex	IRR	Simple / disc. payback	NPV @10%	@15%	@20%	@25%
5 kWp (no export, today)	6.3M	42%	2.3 / 3.1 yr	15.4M	9.8M	6.4M	4.1M
8 kWp (with usable export)	10.1M	46%	2.1 / 2.8 yr	28.5M	18.5M	12.3M	8.2M

Both survive a 25 percent cost of capital comfortably. (The undiscounted 20-year cumulative ROI is large, ~680 to 780 percent, but it is not a decision metric; NPV and IRR carry the case.)

Findings

1. Do not buy more battery for return. At the 5 kWp anchor, adding even the first battery unit lowers 20-year NPV by about 2.3M UGX; a second and third are progressively more negative (about −6.6M and −11.5M). The battery capex is not recovered within the horizon at today's ~626,000 UGX/kWh.
2. Solar income is resilient to battery aging; battery-only savings halve over 20 years, the solar cases decline far less.
3. Value, not penetration. Beyond ~4 to 5 kWp, extra panels spill and returns diminish unless storage gets cheaper or export becomes usable.

Battery replacement strategy

The model conservatively assumes the battery simply fades to 50 percent over 20 years. A real long-horizon plan should choose one of three strategies, and the economics differ:

• Operate the original battery for all 20 years (what the model assumes — the pessimistic bound; usable capacity, and so the evening-shift slice, keeps shrinking).
• Replace once when usable capacity hits a chosen threshold (say ~70 percent around year 10 to 12): restores capacity at the cost of a fresh battery, worth it only if battery prices have fallen (see the break-even case).
• Augment gradually by adding cellswiring as loadobserved grows rather than replacing.

Importantly, the recommended allocation (solar-heavy, existing battery) is the least exposed to this: solar self-consumption does not depend on the battery, so battery replacement is a reliability/backup decision, not a driver of the solar return.

The caveat for the bank case

This prices energy savings only. It does not price the battery's real job, backup and reliability during outages. Solar is the return story; additional battery is a resilience decision justified separately.

Recommendation

Fund about 5 kWp of solar on the existing battery to maximize NPV (about 6.3M turnkey, IRR about 42 percent); choose 4 kWp instead where the priority is the low-spill efficiency sweet spot. If a usable export arrangement materialises, scalingup to about 8May kWp2026 becomesand better.should Forbe howre-verified cheapagainst batterythe mustcurrent getOpenEMS beforesetup expandingif itany pays,connections seehave Batterybeen Pricechanged.

Break-Even. Companion cases: Battery Arbitrage, Solar Plus Battery.