Why Solar kWh and Your Utility Bill Don't Match

Method note: modeled worked example from published NREL field data

This article is a modeled worked example, not a SolarTech customer project. System sizes, annual energy totals, and grid-cost findings come from the National Renewable Energy Laboratory (NREL) technical report *Savings in Action: Lessons Learned from a Vermont Community with Solar Plus Storage* (McKnight Lane, Waltham, Vermont). Monthly production for the same coordinates was fetched from the European Commission Joint Research Centre PVGIS API on 2026-07-20 (NREL PVWatts API was unreachable from this build environment). SolarTech summarizes and tabulates; NREL and JRC did not author this page. Rebuild every input with your own bills and a site-specific model before decisions. Method companion: solar ROI payback method.

Direct answer: what the NREL community data shows

Across the McKnight Lane homes NREL analyzed, each unit has about 6 kW rooftop PV and a 4 kW / 6 kWh battery (one home uses 8 kWh). For Home 1 in 2017, NREL reports 3,452 kWh household use, 5,507 kWh PV output, 2,579 kWh taken from the grid, and 4,469 kWh exported. Across eight datasets, average PV output is about 5,508 kWh/year and average use about 4,584 kWh/year. NREL also finds average annual grid cost to serve native load drops from about $0.045/kWh without solar-plus-storage to about $0.010/kWh with it, or roughly $157 per home per year in grid-cost savings in that study. The teaching point: annual PV can exceed annual use on paper, yet grid imports stay large because load and generation do not line up hour by hour. Batteries change when energy is available, not only how many kWh the roof makes.

Hypothesis

If a 6 kW south-facing rooftop array at Waltham, Vermont (44.152 N, 73.232 W) serves a low-to-moderate all-electric load near the NREL Home 1 2017 total, then modeled annual export will be large on an energy-match basis, while published meter data still show thousands of kWh of grid import, because evening and winter timing (and battery dispatch) dominate bill outcomes more than nameplate kW alone.

Inputs (each row sourced)

PVGIS annual yield (~1,226 kWh/kW) is higher than Home 1's measured 2017 yield (~918 kWh/kW). That gap is expected: different weather years, snow, soiling, orientation, and inverter behavior. Use the meter table for "what happened," and the PVGIS table for a transparent climate-model month shape at the same place.

Monthly modeled production (PVGIS, kWh)

Monthly energy-match offset (teaching table)

Consumption annual total equals NREL Home 1 2017 (3,452 kWh). Monthly shape is a stated winter-heavy assumption for an all-electric Vermont-style load (not published by NREL as a monthly series). Production is the PVGIS row above. Formula: self-consumed = min(production, consumption); exported = max(0, production − consumption); imported = max(0, consumption − production). This ignores hour-level timing and battery charge/discharge.

Compare to NREL meters for the same home-year: 2,579 kWh from the grid and 4,469 kWh exported. The energy-match table nearly eliminates imports; the meters do not. That difference is the lesson. Real homes import after sunset and on cloudy days; batteries and rate design change value. Read self-consumption explained before treating any annual kWh match as a bill forecast.

Annual benefit index (no invented payback currency)

Self-consumed kWh × 1.0 + exported kWh × export_weight. For the energy-match year above, export_weight 0.6 is a labeled variable (not a claim about your tariff). NREL cites study rates near retail for netting and a $0.053/kWh performance incentive; those are Vermont/GMP context in the report, not a global constant.

Installed cost is omitted as a single point value. Simple payback = cost ÷ annual money benefit only after you insert a real quote and your utility's written export rule. See complete guide to solar ROI and solar proposal checklist.

What the battery changes (from NREL, not invented)

NREL models and field context show batteries at McKnight Lane support resilience (study average about 148 hours of sustained load for the 6 kWh packs under their method), peak-related utility value, and lower average grid cost to serve the homes. Household bill savings under the study tariff are not identical to grid-cost savings; NREL separates those perspectives. Do not copy the $157 figure as your personal bill reduction. Use it as evidence that solar-plus-storage value is multi-metric: customer bill, utility peak, resilience, and emissions.

What to do with this model

1. Pull 12 months of your own kWh from bills. 2. Run PVWatts or PVGIS for your coordinate, tilt, and losses; keep the run date. 3. Build the monthly energy-match table, then expect real imports to be higher than the simplified import column. 4. If you add storage, track self-consumption and resilience separately, as in complete guide to solar batteries. 5. Stress-test export weight and −10% production before comparing quotes with solar proposal checklist.

Closing

A transparent 6 kW Vermont worked example grounded in NREL meter totals and a real PVGIS monthly run shows why annual kWh balance is not the same as bill outcome. Production can exceed use while grid imports remain thousands of kWh. Storage and tariffs decide when those kWh matter. Rebuild the tables with your data; do not treat this page as a savings guarantee.