Solar Energy Solutions

Solar PV is misunderstood as a panel-and-inverter purchase. It is in fact a complete electrical system: array, DC isolation, MPPT optimisation, inverter, AC isolation, metering, storage, and distribution. The panel rating in watts is barely a quarter of what determines whether the plant pays back inside 5 years or 15.

Site irradiance is the first input. Nairobi sees ≈ 5.5 kWh/m²/day average GHI; Northern Kenya pushes 6.5; Mombasa coastal humidity drops effective generation. NASA POWER and Solargis databases give locational means but only on-site pyranometer data — even one month — captures soil, shade and microclimate.

Tilt and orientation: in Kenya (latitude 1° S to 4° N) the equator-facing tilt of 10–15° is mathematically optimal but visually almost flat; many architects insist on 15–25° for self-cleaning. The trade-off is annual yield vs water-shedding — a 5-degree mismatch costs ~1% annual generation, soil loss to dust costs 3–5%.

Module choice today is between mono-PERC, TOPCon, and HJT cells. Bifacial modules add 5–15% rear-side gain when mounted over high-albedo surfaces. Polycrystalline still appears in cheap imports but is now obsolete — efficiency 15% vs 21–24% for current mono-PERC, with worse temperature coefficient.

String voltage matters more than panel count, and it is set by the coldest morning rather than the hottest afternoon. Open-circuit voltage rises as cell temperature falls: a modern mono-PERC or TOPCon module has a Voc temperature coefficient of roughly −0.25 to −0.30 %/°C, so at 5 °C in the Kenyan highlands at dawn — twenty degrees below the 25 °C STC reference — Voc sits about 5–6% above its nameplate figure. Size the string on that corrected figure: N × Voc(stc) × (1 + coefficient × ΔT) must stay below the inverter's maximum DC input voltage, with margin. Note that the module datasheet coefficient is the number to use; applying a generic multiplier instead is how strings end up either over the inverter limit or needlessly short. Exceeding maximum DC voltage does not simply reduce output — the inverter trips on DC overvoltage and sustained excursions damage the input stage, which is not a warranty claim you will win.

Inverter topology: string inverters (Sungrow, Huawei, SolarEdge, Fronius, GoodWe) for residential and small-commercial; central inverters (SMA, Sungrow, Power Electronics) for utility scale; microinverters (Enphase) for residential with shading. Hybrid inverters (Victron, Deye, Sungrow Hybrid, GoodWe ET) integrate batteries on the same DC bus.

Battery technology in 2026 is dominated by LiFePO₄ (LFP) cells. Lead-acid is gone from new commercial installs — its 50% depth-of-discharge ceiling and 1,500-cycle life can't compete with LFP's 80–90% DoD and 6,000 cycles. Pylontech, BYD, BSLBATT, Dyness and Huawei LUNA are the workhorses; Tesla Powerwall remains premium residential.

DC isolation is non-negotiable. IEC 60364-7-712 and the AS/NZS 5033 derivative require lockable DC isolators within 3 m of the array (rooftop) and at the inverter end. We see imported Chinese kits omit the rooftop isolator — that single missing component blocks any compliance certificate and voids fire insurance.

Earthing and lightning: PV arrays are large equipotential mats that attract direct strikes. Class I SPDs at array entry and class II at inverter AC output are mandatory. Equipment earth bonding must achieve < 0.1 Ω across the array structure. Inadequate bonding is the leading cause of inverter electronics frying after thunderstorms in the highlands.

Long-term performance is dominated by soiling and degradation. Modules degrade 0.4–0.7%/yr (LID + LeTID); soiling can cost 5–10%/yr if uncleaned. A robust O&M contract includes monthly cleaning, quarterly thermography (cell hot-spots indicate cracks), and annual IV-curve tracing. Sites without it lose 15–20% generation by year 5 vs spec.

The financial model decides the design, not the other way round. Kenya's Energy Act 2019 provides for net metering on systems up to 1 MW, but the practical rollout has been slow and the terms depend on your supply agreement, so the sound approach is to size for self-consumption first and treat any export credit as upside rather than as the basis of the business case. That single assumption changes the answer: a self-consumption design puts the array where the daytime load is, keeps the DC/AC ratio near 1.2, and only adds storage where the evening load or outage exposure justifies it. Systems sized on annual kWh instead of the hourly load curve routinely generate power at noon that nobody on site is using.

Fire safety on rooftop PV is the part that gets skipped. A live array cannot be switched off by the fire brigade — daylight alone keeps the DC side energised — which is why rooftop DC isolation, clear array-boundary marking, and a documented shutdown procedure at the main entrance matter as much as the electrical design. Roof structure deserves the same rigour: a ballasted flat-roof array adds meaningful dead load plus wind uplift, and uplift, not weight, is what removes arrays from Kenyan industrial sheet roofs. Get a structural sign-off in writing before the first rail is fixed.