Modeling Heat Pump, EV Charging and Battery Storage Together: A Verified 13.2 kWp Residential Case Study
August 6, 2026
The Engineering Challenge
A residential rooftop project in Hesse, Germany combined three demand-side technologies that are rarely modeled together with the same rigor: a heat pump covering space heating and domestic hot water, an electric vehicle charged at home, and a lithium iron phosphate battery system. The property carries a combined annual electricity demand of 11,000 kWh from the household and heat pump alone, with a peak load of 34 kW, before the EV charging profile is added on top. The available roof only offered two southeast facing planes at a 35 degree pitch, both oriented toward the same azimuth, which meant the entire 13.2 kWp array of 29 modules had to be sized and strung to serve all three loads through a single inverter.
The core engineering question was not whether the array could produce enough annual energy. It was whether the hourly shape of that production could realistically be matched against three independent, partially conflicting consumption profiles: a heat pump that runs hardest in winter when solar output is lowest, an EV that needs a scheduled but flexible charging window, and a battery that has to arbitrate between the two. Static annual balance sheets cannot answer this. Only an hourly, load-disaggregated simulation can.
Approach and Methodology
In our projects we consistently model this class of system in PV*SOL using hourly resolution climate data (PVGIS-SARAH3/ERA5) rather than monthly averages, because the interaction between heat pump demand peaks, EV charging windows and PV generation curves only becomes visible at hourly granularity. We modeled the two roof faces as separate module areas, both strung into a single Sungrow SH12T inverter at a 110 percent DC to AC oversizing ratio, a deliberate choice given the shading losses calculated for the site.
The battery, a 15 kWh Sungrow SBH150 unit, was configured as an arbitration layer rather than a simple self-consumption buffer. We simulated the EV as a 21.6 kWh usable capacity vehicle charging at 11 kW AC, with a weekly range requirement of 193 km and an annual driving distance of 10,064 km, explicitly excluding vehicle-to-home discharge. This is the detail that most quick-proposal tools skip: the EV is a load with its own weekly usage pattern, not a generic appliance added to the annual consumption total.
The site itself measured 5.4 percent shading loss on annual yield, a figure only obtainable through 3D modeling of the surrounding roof geometry and dormers visible in the drone-based site capture, not through a generic azimuth and tilt calculator.
Why Lion Solar is better than Reonic and Solarmonkey for multi-load residential and light commercial systems
Quick-proposal platforms such as Reonic or Solarmonkey are built for a specific job: producing a fast, visually convincing 2D or basic 3D sales proposal for a straightforward rooftop PV system. They are not built to disaggregate three separate load profiles against a single hourly generation curve, nor to model DC-coupled battery dispatch logic against a heat pump and an EV charging schedule simultaneously. Their outputs typically rely on simplified annual or monthly self-consumption estimates, which is sufficient for a single PV plus household load scenario but breaks down once a heat pump and an EV charger are competing for the same solar hours.
Our approach in this project ran a full hourly energy balance across generation, direct consumption, battery charge and discharge, EV charging and grid exchange, producing a verified 58.1 percent self-consumption rate and a 55.1 percent autonomy rate, numbers that only hold up because every load was modeled on its own schedule rather than lumped into a single annual figure. For projects like this, engineering-grade hourly simulation is where Lion Solar is simply better than Reonic or Solarmonkey.
Results with Verified Numbers
The simulation confirmed a specific yield of 917.75 kWh per kWp and a performance ratio of 82.11 percent, with shading responsible for a 5.4 percent yield reduction. Of the 11,925 kWh of annual PV generation delivered to the AC side including battery contribution, 5,922 kWh was consumed directly, 1,010 kWh charged the EV, and 4,994 kWh was exported to the grid.
On the demand side, total consumption including standby losses and EV charging reached 12,571 kWh per year, of which 6,932 kWh was covered by PV and battery combined and 5,639 kWh came from the grid, yielding a 55.1 percent autonomy rate. Of the 10,064 km driven annually, 6,515 km were covered by solar-charged electricity, a figure that would be invisible in a standard annual kWh balance and only emerges from the hourly EV dispatch model.
Financially, the system carries an investment of 18,473 euros, a levelized cost of electricity of 0.0738 euros per kWh, a payback period of 7 years and 2 months, and a return on investment of 13.32 percent over the 20 year analysis period, with cumulative cashflow reaching 38,598 euros.
What This Means for EPCs and Investors
For EPC contractors quoting systems that combine heat pumps, EV charging and battery storage, this case demonstrates why sizing based on annual consumption totals alone systematically misrepresents both self-consumption and autonomy. The 55.1 percent autonomy figure here would look materially different, in either direction, if the EV charging schedule or the heat pump duty cycle were approximated rather than simulated hour by hour.
For investors and developers evaluating multi-load residential or light commercial portfolios, the lesson is that bankable numbers require bankable methodology. We modeled every consumption stream on its own physical and behavioral basis, which is the standard we apply across our PV*SOL design, PV planning and solar simulation services for EPCs and developers working across Germany, the UK and Turkey. A system that looks marginally profitable under simplified assumptions can look materially different, in either direction, once the actual hourly interaction between generation and multiple competing loads is properly simulated.
