Peak shaving and price arbitrage target different outcomes with the same physical capacity. Operators of co-located solar and storage choose between reducing grid peaks and responding to price spreads. Both have valid economic rationales, but require different dispatch patterns. When they conflict, tariff structure and grid connection terms determine which one loses.
Deciding between them, or coordinating them properly, is a practical operating issue rather than an abstract strategy exercise. It appears in individual charge and discharge commands on particular days at particular sites. This article covers how both objectives work, where they clash, and why the electricity agreement determines the appropriate choice.
What peak shaving is actually trying to do
Peak shaving lowers the highest power draw a site takes from the grid during specified measurement windows. In Portugal and Spain, commercial and industrial tariffs typically include a capacity charge, a monthly fee tied to peak demand at the grid connection point. A higher recorded peak produces a higher capacity charge.
The economics are direct. When a site's capacity charge uses a 15-minute interval peak, the battery should discharge as grid consumption nears the level that would raise the recorded peak. With a 50 kW threshold and 47 kW current draw, it should discharge 10 to 15 kW to keep the interval below that threshold, even when spot prices are unattractive.
Peak shaving matters most at commercial sites with variable loads, where a brief working-hours surge can increase the capacity charge for the whole month. Depending on the tariff, one 20-minute, 80 kW spike on a Tuesday can add more to the monthly capacity charge than ten times the spike's energy cost.
What price arbitrage is trying to do
Price arbitrage earns from the gap between low and high electricity prices: charge when prices are low and discharge when they rise. Its revenue comes from the price difference, not from lowering a capacity measure. Pure arbitrage follows market prices rather than the site's load profile and therefore does not respond to consumption peaks.
For operators of distributed solar and battery sites with a straightforward consumption tariff and no significant capacity charge, arbitrage is usually the main revenue source. The battery charges during midday solar surplus or overnight low-price periods, then discharges into the evening demand peak.
In MIBEL, the pronounced Iberian duck-curve price pattern creates a recurring spread between midday and evening peak prices. Windy days compress that spread, while high-demand winter evenings widen it, but the opportunity's direction remains consistent over a year.
Where the two objectives conflict
The clash concerns available capacity. At 15:30 on a Tuesday, a battery protecting a peak shaving threshold must retain enough SoC for the predictable 16:00 peak, when a large HVAC load starts as the site closes. An arbitrage strategy, however, may prefer to empty the same battery into a MIBEL intraday session 3 price spike at 14:00.
Full dispatch during the 14:00 arbitrage window leaves the battery depleted at 15:30. When the 16:00 peak reaches 82 kW, the month's capacity charge threshold is pushed higher. The cycle's arbitrage revenue is genuine, but the added capacity charge may be greater.
This failure is not merely theoretical. It is the most frequent error we see at commercial sites where dispatch was configured without deciding which objective takes precedence when the two conflict.
The tariff structure determines which objective to prioritize
Neither arbitrage nor peak shaving is universally preferable. The correct priority follows entirely from the structure of the site's electricity agreement.
A site with high contracted capacity and substantial variable load, where capacity charges make up 30 to 40 percent of the monthly electricity bill, should put peak shaving first and use a battery sized for the expected demand variance. Arbitrage comes second, using capacity not needed for the reserve.
A flat-rate tariff or low capacity charge, especially at a residential or small commercial installation with a modestly sized connection, offers little reason for peak shaving. Pure arbitrage is the better battery use, following price signals throughout the day. Holding capacity for peak events that may never happen creates opportunity cost without a matching benefit.
The third situation requires active coordination: a commercial site where both objectives matter and the battery cannot support both at once. The dispatch system must assess each opportunity's expected value in real time and set the priority. That assessment must include the monthly capacity charge tariff, the billing period's current peak measurement, and the intraday market's current price signal.
How combining the two strategies works in practice
A system serving both objectives uses a priority stack. First, when the site is close enough to the defined peak shaving threshold, the battery discharges to control load regardless of price. Below that threshold, it follows arbitrage signals while preserving a minimum SoC reserve for peak shaving.
The minimum SoC reserve reflects the expected maximum demand excursion above threshold for the rest of the billing period, the battery's discharge power limit, and the expected length of peak events at the site. A consistently smooth load that seldom nears the threshold needs little reserve. Unpredictable industrial process loads require more.
Reserve setting is a site-specific calibration task. Applying one fixed fleet-wide reserve to every commercial site will be too cautious for some and too aggressive for others. Predictable, stable loads can use a thinner reserve and capture more arbitrage value. Variable loads need a deeper reserve to prevent capacity charge exceedances.
Practical considerations for fleet operators
For a mixed fleet, begin by auditing each site's exposure to capacity charges and the percentage of its monthly bill attributable to that component. That identifies sites requiring peak shaving logic and those suitable for pure arbitrage.
Next, confirm that dispatch configuration exists per site for the capacity threshold, billing period start date, and current period peak reading. Knowing the recorded peak lets the system determine remaining headroom before a breach and set a reserve proportional to it, rather than using a fixed quantity.
Our findings from the early-access operator group working with Pollen show that coordinating both objectives, instead of selecting only one, consistently beats single-objective dispatch over a monthly billing cycle. The gain is greater in months with high intraday price volatility, when more arbitrage opportunities appear during periods that do not consume the peak shaving reserve.
