How can an operator dispatch a residential portfolio whose sites are small but collectively substantial? Each installation typically holds 6 to 15 kWh, yet forty sites averaging 10 kWh provide 400 kWh of distributed battery capacity. Every location also has its own solar profile, export limit, household load pattern, and local grid conditions.
Automated dispatch is essential for residential portfolios because no operator can manually monitor forty sites at 15-minute resolution. The practical issue is defining what the system must handle at residential scale. Residential assets differ from commercial installations in ways that directly shape dispatch decisions.
How residential sites differ from commercial installations in dispatch terms
Commercial sites generally present stable, predictable load profiles. A light industrial facility or office building often follows a repeatable weekday pattern, giving dispatch systems a workable planning basis. Residential demand is different. Consumption can shift every 15-minute interval as a washing machine or dishwasher runs, an electric vehicle begins charging, a heat pump starts, or a family returns home on a weekday afternoon.
This variability is less important for arbitrage than for applications requiring the battery to retain a specific SoC reserve for possible load spikes. A residential battery operating on pure arbitrage follows price signals, so unpredictable household consumption mainly changes how quickly stored energy is used between charge cycles. Where operators also configure peak shaving or self-consumption maximization, load variability becomes part of the dispatch decision itself.
Battery size relative to grid connection is another distinction. A residential site with a 6.5 kWh battery and a 3.7 kW inverter faces tight limits on both energy per cycle and charge or discharge power. If a low-price window lasts only 45 minutes, the system may lack sufficient charge power to reach its target SoC before the window ends. Dispatch must model each site's charge rate constraint and set achievable targets.
The uniform rules problem at residential portfolio scale
Many portfolio systems send every site the same charge instruction: when MIBEL intraday session 3 prices exceed a threshold, all sites discharge; when overnight prices fall below a threshold, all sites charge. This is easy to administer and captures much of the available spread when sites follow similar SoC paths.
Residential sites rarely share the same SoC trajectory, even when solar generation is fairly consistent across a region. Site A may reach 14:00 at 40% SoC after its household used a heat pump at midday. Site B may reach 14:00 at 75% SoC because nobody was home. Giving both sites the same discharge instruction at 14:30 lets site B complete a useful cycle while site A completes only half a cycle.
The daily revenue difference is modest. Across forty sites over a year, however, these uniform-rule mismatches consume a meaningful share of the theoretical maximum revenue.
Per-site dispatch at 15-minute resolution: what it requires
Per-site dispatch gives every site charge and discharge instructions using its current SoC, solar generation rate, battery charge power limit, and current price signal. At forty sites and 15-minute resolution, that equals 160 individual dispatch evaluations per hour, or approximately 3,840 per day.
Computation is not the limiting factor. Data is. Every evaluation requires current SoC from the battery management system, current solar generation from the inverter, and the latest price signal from MIBEL or the site's participating market. Stale inputs produce decisions that are wrong in proportion to their age.
Residential inverters differ substantially in the freshness and reliability of API data. Some report state data every minute, while others update every five minutes. API instability can also take an inverter offline briefly, after which returned values may not match the hardware's actual state. A residential dispatch system must manage these conditions by flagging stale inputs and using a conservative default instead of acting on unreliable state data.
Solar generation as a real-time input to dispatch decisions
At residential sites with co-located solar and batteries, current solar generation is a particularly important input. Its rate indicates whether the battery should charge from solar or preserve SoC for an intraday price window. A battery at 50% SoC at 10:30 under strong solar conditions will likely reach 75% by 12:00 without a grid charging instruction. Ignoring that generation and charging from the grid at 10:30 pays for energy that solar would have supplied within 90 minutes.
Dispatch must combine solar generation, the price signal, and SoC to choose among active charging, holding SoC, and discharging into an available price window. With strong morning solar on a clear-sky day, most residential sites should hold SoC in the morning and let solar fill the battery, then assess discharge opportunities as MIBEL intraday sessions update through the afternoon. On an overcast day, solar contributes little, so grid charging should follow price signals alone.
Fleet-level reporting for residential portfolio operators
An operator responsible for forty residential sites will not open each inverter portal separately every morning. They need a fleet view showing the information required for daily decisions: which sites underperformed yesterday against the available spread and why, which sites have hardware problems affecting dispatch compliance, and the aggregate fleet SoC before the next trading day.
Reporting that turns per-site performance into actionable fleet summaries, instead of only raw exports, materially reduces management overhead. An operator checking forty individual sites for four hours per day can spend thirty minutes reviewing fleet exceptions when the report identifies sites needing attention rather than requiring a full scan.
The realistic improvement from automated dispatch at residential scale
Precision matters because overstating results can mislead investment decisions. For residential sites conducting pure intraday arbitrage in MIBEL, automated per-site dispatch at 15-minute resolution typically raises revenue capture by 15 to 30 percent versus manual dispatch based on daily day-ahead price checks. The gain comes from three sources: responding to intraday session updates that manual processes miss, managing SoC correctly so batteries reach peak windows with sufficient charge, and incorporating solar generation to avoid grid charging that solar would have supplied.
Gains vary by day. When prices closely track the day-ahead forecast and intraday session revisions are immaterial, automated and manual dispatch can deliver similar results. The difference is concentrated on days with intraday divergence, partial cloud cover affecting sites differently, or unusual SoC paths caused by household loads. Such days are not rare; with significant renewable variability, they make up most operating days.
