Smart batteries for smarter buildings

Cities are responsible for 70% of global CO₂ emissions, and yet solar energy accounts for less than 2% of Sweden's electricity supply. Closing that gap requires not just more panels on rooftops, but smarter ways to manage the energy they generate.

That's exactly what Qurrent colleague Linda Lundmark set out to investigate. In a newly published study in Science Direct's publication Applied Energy, she and her co-authors from KTH Royal Institute of Technology's Department of Energy Technology and Northvolt present something rare in this field: a forecasting and optimization model that was actually tested and validated on a real, live system, not just simulated.

The setup: a living lab in Stockholm

The research was conducted at the KTH Live-In Lab, a residential research facility where innovative technologies are tested in real-world conditions. A 186 kWh battery energy storage system (BESS) was integrated with the building's existing 76.4 kW rooftop solar installation, and real-time meter data was used to train, run, and validate the model, giving the findings a level of practical credibility that much of the existing literature lacks.

The big question: how should a battery behave?

The study tested four operational strategies and compared them against a baseline scenario with no battery at all (which had an annual electricity cost of ~210,000 SEK and 72% solar self-consumption). The findings reveal a clear and important tension:

The strategy that saves the most money in the short run isn't necessarily the best one overall.

• Optimizing for cost only (ignoring peak-power billing fees) reduced costs by 32% and pushed self-consumption to 96%. But it also drove aggressive charge-discharge cycles that degraded the battery so severely that it would need full replacement within the 25-year lifespan of the solar installation, significantly eating into the long-term gains.
• Optimizing for cost while accounting for peak-power fees reduced costs by around 14%, raised self-consumption to 98%, and kept battery degradation so low that the system outlasted the PV installation's lifespan. This scenario delivered the lowest levelized cost of storage (LCOS) and the most sustainable long-term performance.
• Maximizing self-consumption achieved 100% self-sufficiency from solar, but proved economically unviable, resulting in a net loss over the system's lifetime.

Hidden conflict in Swedish energy billing

One of the study's most policy-relevant findings is the friction between two pricing signals that Swedish electricity consumers face simultaneously: real-time hourly energy prices (which reward rapid, opportunistic battery use) and monthly peak-power billing fees (which penalize high grid draw in any single hour of the month).

For larger grid connections, like those in apartment buildings, these two signals pull in opposite directions. A battery told to maximize market arbitrage will spike its grid draw and trigger steep peak-power charges. This billing structure, the study argues, is a genuine barrier to unlocking the full financial potential of residential solar-plus-storage systems.

Why this matters beyond Sweden

The forecasting-optimization framework Linda and her team developed is adaptable. It can be retrained on data from other sites, handle different billing structures, and scale to various building types. In markets without net metering, where selling excess solar back to the grid is less profitable, the case for smart battery dispatch becomes even stronger.

The study also reinforces a point that deserves more attention from investors: battery dispatch strategies must be dynamic across seasons. Winter and summer conditions in Nordic climates produce very different solar yields, price environments, and grid interactions, and a one-size-fits-all control strategy will leave significant value on the table.

The takeaway

There is no single "best" way to operate a solar battery system. The right strategy depends on the balance between short-term cost savings, long-term system health, billing structure, and energy goals. What this study makes clear is that smarter, forecast-driven, and context-aware operation is not just theoretically superior, it works in the real world.

Congratulations to Linda and the full research team for a study well done. We're proud to have Qurrent contribute to research toward more intelligent and resilient energy systems.

Read the full open-access paper: “A validated forecasting model for optimizing dispatch schedules of PV-battery systems in grid-connected apartment buildings” — by Linda Lundmark, Farzin Golzar, Rafael Guedez & Majid Astaneh. Published in Applied Energy, Volume 404, February 2026.