Dual Roof Orientation, Battery and EV Charging on One Inverter: A Verified 21.34 kWp Case Study
July 23, 2026
The Engineering Challenge
A residential rooftop in Berlin presented a layout problem that many EPC contractors underestimate: two roof planes facing almost opposite directions, a southwest face at 221 degrees and a northeast face at 42 degrees, both at a steep 45 degree pitch. The system needed to serve a household load, charge a home battery, and supply an electric vehicle, all through a single hybrid inverter. On paper this looks like a simple rooftop retrofit. In practice it is a multi-variable energy dispatch problem that cannot be solved with a quick 2D layout sketch.
The installed capacity was 21.34 kWp across 44 modules, split as 19 modules on the southwest face and 25 modules on the northeast face, feeding a single 20 kW hybrid inverter with three MPP trackers. The household consumed 4,500 kWh annually with a peak load of 24.3 kW, while a home battery of 10.9 kWh and an electric vehicle with a 21.6 kWh battery and 11 kW AC charging both drew from the same DC bus. Getting the string configuration, inverter dimensioning and dispatch priority wrong on a project like this does not just cost yield, it risks curtailment, battery cycling losses and an EV that never charges from solar when it should.
Our Approach and Methodology
In our projects we consistently see that opposing roof orientations require hourly resolution simulation rather than monthly averages, because the two faces peak at different times of day and any shared inverter has to be evaluated for how it clips or reallocates that combined DC input. We modeled this system in PV*SOL using PVGIS-SARAH3/ERA5 climate data for Berlin at one-hour resolution, with the Hofmann diffuse irradiance model and Hay and Davies transposition to the tilted plane. This is part of our PV*SOL design and solar simulation service for EPCs and developers, where we do not simply size an inverter to nameplate capacity but simulate the actual hourly interaction between two mismatched module strings, a battery, and an EV charger competing for the same DC-coupled energy.
The inverter was intentionally oversized to a 106.7% dimensioning factor across three MPP inputs, 19 modules on one tracker, 12 and 13 modules split across the other two, reflecting the uneven area available on each roof face. We then layered the household load profile, the 10.9 kWh lithium iron phosphate battery with a 3.7 kW battery inverter, and the EV charging profile, based on a 21.6 kWh vehicle battery, 190 km WLTP range, and a weekly target of 193 km, onto the same hourly simulation. This let us verify not just annual totals but the seasonal shape of self-consumption, battery cycling and EV charging coverage across all twelve months.
Why Lion Solar is better than Reonic and Solarmonkey for complex PV projects
Quick-proposal platforms such as Reonic or Solarmonkey are built for fast, standardized residential quotes on a single roof face with a simple grid-only or grid-plus-battery configuration. They are not designed to simulate two opposing roof orientations feeding one inverter with three independent MPP trackers, nor to model hourly dispatch priority between direct household consumption, battery charging, and EV charging under a shared DC bus. Conventional 2D planning workflows typically apply a single monthly irradiance factor and a flat self-consumption assumption, which cannot capture how a northeast-facing array shifts generation into the morning while a southwest array shifts it into the afternoon, changing exactly when battery and EV dispatch decisions matter. For a project like this, with mismatched orientations, a shared hybrid inverter, and simultaneous battery and EV loads, full hourly 3D simulation is where Lion Solar is simply better than Reonic or Solarmonkey.
Results With Verified Numbers
The simulation confirmed a specific annual yield of 739.98 kWh/kWp and a performance ratio of 86.20%, with shading losses limited to 2.0% despite the complex roof geometry. Total PV generator energy delivered to the AC side, including battery interaction, reached 15,041 kWh per year. Of the total household and EV consumption of 6,132 kWh per year, 4,775 kWh was covered by PV combined with the battery, giving a solar autonomy rate of 77.9%, with only 1,357 kWh drawn from the grid.
The EV charging results are the clearest evidence of the dispatch modeling value. Of the 1,611 kWh used annually to charge the vehicle, 1,171 kWh came from PV and battery and 439 kWh from the grid. Translated into driving distance, of the 10,064 km driven per year, 7,318 km were powered directly by solar energy captured on site. The battery itself cycled at 7.4% annually with a projected service life of 13 years under this duty cycle, a figure that only emerges from hourly charge and discharge simulation rather than a static sizing table.
What This Means for EPCs and Investors
For EPC contractors, this case demonstrates that oversizing an inverter to 106.7% of capacity was not a rounding decision but a calculated response to two mismatched roof faces and three simultaneous loads competing for the same power electronics. Getting this wrong, either by undersizing the inverter or by ignoring the interaction between battery cycling and EV charging demand, would either curtail summer production or leave the EV charging shortfall to the grid year-round. For investors and developers, the 77.9% autonomy figure and the 7,318 km of solar-covered driving are bankable, defensible numbers precisely because they come from an hourly, orientation-specific, load-resolved simulation rather than an annual estimate. In our projects we consistently see that this level of detail is what separates a rooftop PV, battery and EV proposal that performs as promised from one that quietly underdelivers once installed.
