Inside a 115 kW Solar Carport + Battery-Buffered DCFC Build: Q&A With NuWatt’s Lead Engineer

We start with session profile, not peak nameplate. At a car wash the average DCFC dwell is 8–12 minutes, and simultaneous sessions happen less than 10% of the operating day in a six-port deployment. Our sizing target is 100 kW / 200 kWh — enough to carry one full 150 kW session for ~80 minutes with solar contribution, or to stagger two overlapping sessions so neither sees grid-side draw above 30 kW. We model 6 months of 15-minute utility meter data (or synthetic load if the site is new) against expected EV traffic, then pick the BESS with the worst-case 95th-percentile peak hidden below the demand-charge threshold. Oversizing wastes capex; undersizing blows the demand tariff and kills the ROI thesis.

Everything downstream is the same — the DC bus, the BESS, and the chargers do not care about upstream service class. What changes is the utility-engineering timeline and the capex. On an existing 480V 3-phase with 400A+ of headroom, we can drop in a 200 kWh buffered system without touching the Eversource service. If the site sits on 208V or on a 480V service that is already loaded past ~70%, we are either upgrading the service (12-month queue, $200K+) or we are oversizing the battery to let us survive on existing capacity. Nine out of ten car washes have adequate 480V 3-phase because their wash tunnels already pull 100–150 kW of motor load during rush periods.

The trick is clean basis allocation at the project-accounting level. We split the bill of materials into three buckets: generation assets (PV modules, inverters, canopy steel, interconnection hardware), refueling assets (EV chargers and their dedicated make-ready), and shared balance-of-system (switchgear, trenching, site prep). 48E claims the generation bucket. 30C claims the refueling bucket. The shared bucket gets allocated by pro-rata cost between the two asset classes. Our tax counsel signs off on the allocation memo before any credit is claimed — the IRS audits these closely and an unclean split can void one or both credits.

Treating the battery as a solar appliance instead of a charger appliance. The moment you DC-couple BESS to the DCFC dispensers instead of AC-coupling it to the solar inverter, the dispatch logic becomes charger-driven, not generation-driven. Installers who come from a solar-only background default to the AC-coupled topology and then wonder why demand charges are still spiking during DCFC sessions. The AC-coupled battery has to inverter-dump to the meter, race the DCFC draw, and the grid still sees the transient. DC-coupled, the battery delivers directly into the DC bus the DCFC is already pulling from. Zero grid transient.

Three reasons, in order: daisy-chain power sharing, OCPP 2.0.1 firmware that talks to the same network the DCFC dispensers report to, and the commercial back-office we can hand off to the car wash operator day one. CP6000 supports up to 40 units on a single 200A feeder with Power Management — that is perfect for car-wash parking where customer dwell is short but simultaneous. On OCPP 2.0.1 we get session-level telemetry and pricing schedules that match what Tritium or Kempower push on the DCFC side. The operator logs into one dashboard and sees everything.

Three scenarios: (1) Sites that will run 4+ DCFC dispensers simultaneously for sustained windows — a 100 kW / 200 kWh battery cannot carry that, and scaling BESS past ~500 kWh starts costing more than a service upgrade. (2) Sites with cheap, fast utility service expansion — some rural CT territories have paid-in service upgrades available in under 90 days for under $80K. (3) Sites that plan to host V2G or export energy back to the grid during peak events — the battery-buffer topology is charger-only, so V2G export requires a different topology and a larger interconnection footprint. The buffered design is the right answer at 2–3 DCFC dispensers, existing 480V 3-phase, and private-retail use. Outside that window, run the math.

ISO-NE capacity tags and Eversource demand charges attack the site on different rhythms. The ISO-NE capacity tag is set by the single highest coincident load hour during the summer — miss that hour and the site gets stuck with a capacity obligation that rolls forward for a year. Eversource G-3 demand is measured on the highest 15-minute peak each billing cycle. Our battery controls run two schedules in parallel: a site-peak minimizer that monitors the Eversource meter and dispatches whenever interval draw exceeds a threshold, and a capacity-tag-shed mode that watches ISO-NE day-ahead forecasts and pre-charges the battery for anticipated system peak hours. Both modes run on standard commercial BESS controls; the art is in the threshold tuning.

Always the generation side before the load side. We witness-test the solar canopy with Eversource, confirm inverter output matches design, and run it on a dummy load for 48 hours. Then we energize the battery cabinet and run a charge/discharge cycle at 25%, 50%, and 100% SOC to validate the BMS. Only after both generation and storage are proven do we bring up the chargers — first the L2s (lower stakes, easy to cycle), then the DCFC dispensers one at a time. Full DCFC commissioning takes a day per unit because we run each through OCPP back-office registration, a full battery-buffered session test, and a demand-charge sanity check at the meter. Only after everything passes do we hand the car wash operator the keys.

The operator’s real constraint is customer parking — a car wash loses revenue the minute customers cannot queue. We phase the work to keep at least two-thirds of parking stalls available at any given time. Canopy footings go in overnight in two sections. The battery cabinet sits outside the customer flow entirely — usually along the back or side of the building envelope. The DCFC dispensers and L2 posts go on the same parking-stall footprint that the canopy covers, so once the canopy footings are set, everything else is above-grade and does not disturb parking. The only hard-downtime event is the service-entrance rework, which we schedule for a Sunday night 4-hour window.

Quarterly: remote BMS diagnostic via the BESS OEM’s cloud portal, DCFC firmware check, and OCPP session-log audit. Semi-annually: physical site walk — cable inspections, thermal-camera scan of BESS cabinet and DCFC power module heat sinks, torque-check on service entrance. Annually: full BMS capacity test (we discharge 100% → 0% → 100% and log Ah), inverter firmware update, ISO-NE capacity-tag performance review, and a session-pricing review against local utility rates. We offer a 10-year full-service O&M contract at roughly 1% of net capex per year, escalating 3% annually. Most operators take it because a single DCFC power-module swap is a $12K out-of-pocket event, and one swap per year pays for the contract.