On a clear midday, a solar site may be producing below its panels' potential because grid or export limits leave the electricity nowhere useful to go. For operators of co-located batteries, that is a specific loss: power that could have been stored and discharged during the evening peak is wasted instead.
Curtailment is often treated as an outside constraint to plan around. A closer look shows that many events are foreseeable. By combining solar generation forecasts with dispatch, a system can position batteries to absorb a meaningful share of curtailed energy before it happens.
Why curtailment happens at distributed sites
At small distributed solar sites, export limits are the usual cause. Distribution network operators cap the power a site may send to the grid at any moment. That cap is often the contracted grid connection capacity, which is frequently well below installed panel capacity at residential and small commercial sites.
An 8 kWp residential array with a 3.7 kW export limit will curtail on clear midday hours when production exceeds on-site use plus the export allowance. If consumption is 0.4 kW at 13:00 and solar output is 7.5 kW, 7.1 kW is available for export. With a 3.7 kW limit, roughly 3.4 kW is throttled by the inverter or can charge a battery if capacity is available.
Commercial export limits may be negotiated higher, but they remain fixed. A 50 kWp rooftop system with a 25 kW export agreement curtails when production exceeds on-site demand plus 25 kW. On high-irradiance summer days, this may occur every day for several hours.
Grid operator curtailment works differently. When the distribution operator sends an instruction, it usually acts directly on the inverter or connection point and may block both export and battery charging during the instruction window. Recovery is genuinely harder because the energy is lost at the panel rather than remaining available for local storage.
The dispatch conflict at the core of the problem
Arbitrage dispatch and curtailment recovery can conflict. Operators should understand that clearly before assuming a co-located battery will capture all curtailed generation.
Under simple overnight arbitrage, a battery charges from the grid at low overnight prices and discharges during the evening peak. After overnight charging, it may be full by midday, leaving no room for solar being curtailed by the export limit. Curtailment occurs, but the full battery cannot absorb it.
Curtailment recovery therefore needs dedicated dispatch logic, not merely an add-on to arbitrage. The overnight charge target must reflect expected midday solar. With a clear-sky forecast, the battery should reach midday with useful headroom, giving up some overnight arbitrage to create space for free solar that can be discharged later at peak prices.
The tradeoff is overnight grid energy purchased at roughly 30 to 40 EUR/MWh versus solar at near-zero cost. If the solar irradiance forecast is accurate, charging from curtailed solar and selling at the evening peak beats overnight grid charging followed by sale at that same peak. Forecast accuracy determines how confidently the overnight target can be lowered.
What a dispatch-aware system can actually recover
Consider the limits. A 10 kWh battery with 4 kWh of headroom created by lowering its overnight target can absorb up to 4 kWh of curtailed midday solar. If clear-day curtailment is 8 kWh per day, the battery recovers half, while the other 4 kWh is lost.
More recovery requires more headroom, which means cutting overnight arbitrage charging further. Eventually, forgone overnight margin costs more than the added recovery is worth. The optimal overnight target balances both values and changes daily with the solar forecast.
At sites in southern Portugal and Spain, summer clear-sky irradiance is high and residential export limits are often 3.5 to 7.4 kW. With suitably sized batteries, a well-calibrated dispatch system can recover 20 to 40 percent of annual curtailment. That is meaningful revenue, not a rounding error, but curtailment does not vanish. Battery capacity remains the hard limit.
The solar forecast integration requirement
Manual dispatch cannot reliably perform this optimization at fleet scale. Reviewing weather and changing overnight targets for 20 or 50 sites each evening is not viable. A dispatch system can combine irradiance forecasts, panel capacity and orientation, current battery SoC, and the next day's price profile to calculate each site's overnight target automatically.
Solar forecast accuracy directly affects the headroom calculation. Overestimating cloud cover on a clear day creates too little headroom, so the battery is full when curtailment occurs. Underestimating cloud cover creates too much headroom, leaving the battery partly empty at midday with little solar to refill it, while grid arbitrage was forgone without benefit.
For our early-access operators using Pollen in MIBEL zones, the next day's irradiance data updates the solar forecast integration overnight. Where Pollen can access actual inverter generation, the following overnight target uses recent output from that site, not only regional averages. Site-level calibration substantially improves the headroom estimate over uniform regional forecasts.
Realistic framing for curtailment as a revenue source
Operators assessing a battery investment can overstate curtailment recovery in the business case. The usual mistake is assuming the battery will absorb all curtailed generation every day, without allowing for headroom or its interaction with arbitrage.
A more defensible view is that curtailment recovery is an additional revenue layer for operators whose batteries primarily perform price arbitrage, provided dispatch integrates solar forecasts and adjusts overnight targets. Revenue depends on curtailment volume, battery size relative to that volume, and forecast accuracy when setting headroom. It can become meaningful over a full year, but should not be forecast as 100 percent of theoretical curtailment.
In this context, automated dispatch mainly adds consistency. Manual operators miss headroom adjustments when other work intervenes. A dispatch system does not.
