The timing matters. Under the European Renewable Energy Directive and its revisions, together with the “Fit for 55” package, renewable energy communities have been formally recognised as market actors entitled to own, develop and operate local renewable projects. Italy transposed these principles through Legislative Decree 199/2021, refined them in the Integrated Text on Distributed Self-Consumption approved by the regulator ARERA in 2022, and finalised the incentive scheme with the 2023 ministerial decree known as CACER. Under the definitive arrangement, each community member is remunerated hourly for the fraction of renewable generation that the community virtually self-consumes, calculated ex post from smart-meter data and valued through a premium tariff called TIP. The problem is that all of this hourly accounting requires hourly data that emerging communities simply do not have at the pre-feasibility stage.

The scale of the deployment gap is striking. According to the official dataset of Gestore dei Servizi Energetici, the state agency managing Italian renewable support schemes, only 212 active renewable energy communities were mapped as of March 2025, with 22.4 megawatts of installed renewable capacity — a mere 0.4 percent of the subsidy capacity allocated by law. For comparison, Italy’s cumulative photovoltaic fleet reached roughly 37 gigawatts by the end of 2024, meaning community-based capacity sits well below 0.1 percent of the national total. Previous studies have attributed this shortfall to administrative burdens borne by untrained volunteers, limited expertise in small municipalities, and the difficulty of assessing economic feasibility before resources are committed. Existing tools, from public calculators to commercial design software, tend to either oversimplify or demand skills and computational effort beyond the reach of citizen-led initiatives.

The core of the new methodology is a five-step statistical model, implemented entirely in a spreadsheet, that turns billing data into a full 8760-hour electricity demand profile. The procedure starts by harvesting statistical hourly consumption patterns from ARERA’s freely accessible database, which provides profiles for typical weekdays, Saturdays and holidays, broken down by month, province, contracted power and dwelling typology — 864 hourly values per residential profile annually. These profiles derive from measured withdrawals recorded through Italy’s national metering infrastructure. In the second step, billing information arriving in various formats — annual totals, monthly totals, or monthly totals split into the Italian F1, F2 and F3 billing time slots — is harmonised into a consistent matrix, with missing time-slot values filled by proportional disaggregation using the statistical shares. Steps three and four build comparable monthly-day-type-time-slot matrices from the statistical data and the real bills respectively, and the fifth step rescales each statistical hourly value by the ratio between billed and statistical consumption for the corresponding month, day type and time slot. The result preserves realistic hourly shapes while forcing the annual and monthly totals to match the customer’s actual bills.

For non-residential members — bars, hotels, retail shops, farms — comparable statistical profiles do not yet exist, so the researchers instead built hourly patterns from recognised standards such as SIA 2024 and ISO 18523-1, refined with local billing data and operational details like opening hours and days off. Photovoltaic production for every existing and planned system in the community was simulated hourly using the BIMSolar dynamic modelling software with local weather files, capturing real roof orientations, inclinations and capacities.

Crucially, the team did not simply assert that their reconstruction was adequate; they validated it against ASHRAE Guideline 14, the industry benchmark for building energy model calibration, which requires a Normalized Mean Bias Error within plus or minus 10 percent and a Coefficient of Variation of the Root Mean Square Error at or below 30 percent for hourly models. Using 15-minute smart-meter records from two residential users collected between January and May 2025, the validation produced remarkably low bias — NMBE of −0.2 percent and −0.07 percent for the individual users, and −0.13 percent for their aggregated profile. The CVRMSE, which is highly sensitive to individual consumption spikes, told a more nuanced story: one user exceeded the threshold at 47.8 percent, the other passed at 29.6 percent, and the aggregated profile came in comfortably within limits at 24.8 percent. This aggregation effect is analytically central. Individual household behaviour is intrinsically noisy and unpredictable, but when profiles are summed at community scale, random user-level variability partially cancels out. Since the economic fate of an energy community depends on the collective hourly balance between generation and demand rather than on any single household’s evening spike, the method’s sweet spot — aggregated, community-level assessment — is exactly where accuracy matters most.

The workflow was then put to the test on a real initiative: CERivanazzano, a renewable energy community formally constituted as a non-recognised association on 11 April 2023 in the small town of Rivanazzano Terme in the province of Pavia, northern Italy. Its baseline configuration comprises 21 consumers and 13 prosumers spread over roughly 23 square kilometres, including fifteen residential consumers, three small shops, a bar, two farms and a hotel acting as prosumer, with about 73 kilowatts of installed photovoltaic capacity against roughly 210 megawatt-hours of annual consumption. Members were recruited through open events and workshops and supplied bills, contract details and declarations of intent to install rooftop systems.

From this baseline, the researchers defined four expansion scenarios reflecting genuine proposals discussed with the association. Scenarios one and two compared two strategies for adding roughly the same amount of new photovoltaic capacity: distributing 4.5-kilowatt systems across seven additional prosumer rooftops versus installing equivalent capacity in a single standalone community-scale plant with no direct on-site load. Scenarios three and four tested the inclusion of three small commercial stores, starting respectively from the baseline and from the standalone-plant configuration. Because hourly demand and hourly production were available for every member, the team could compute physical self-consumption, virtual self-consumption, several normalised efficiency indicators and, under the Italian framework, the TIP-based revenues for each configuration.

The results are striking in their implications for community design. Adding capacity on prosumer rooftops boosts physical self-consumption at the building level — rising from 26 percent to 31 percent in scenario one — but starves the community of energy available for virtual sharing, which drops from 52.1 to 43.9 megawatt-hours per year and pushes total TIP income down from 5,346 to 4,513 euros annually. The standalone plant does the opposite: because its generation has no co-located load, more electricity remains available for virtual self-consumption, which climbs to 67.2 megawatt-hours and lifts profitability to 31.5 euros per megawatt-hour consumed, against 24.4 in the baseline. Scenario four, combining the community-scale plant with three small commercial users, performed best of all: annual community income reached 8,010 euros with a profitability index of 32.2 euros per megawatt-hour — a 32 percent improvement over the baseline configuration.

The mechanism behind this result is temporal complementarity. Small commercial loads consume during the daytime, precisely when photovoltaic output peaks, so their presence raises the volume of energy that can be virtually shared during PV-active hours. Hourly analysis of the incentive values showed that a typical workday, with its greater daytime demand, generated TIP revenues roughly 20 percent higher than a comparable holiday in the same month. Box-plot analysis of hourly community efficiency across the year revealed stable, consolidated performance between 8 a.m. and 3 p.m., with high variability in early morning and evening hours when solar production fades while demand patterns diverge by day type. The authors note that such patterns could guide load-shifting strategies, although behavioural interventions were not explicitly modelled.

The study’s authors are careful about boundaries. The economic analysis covers only operational revenues under a consistent zonal price assumption from 2022, excluding investment costs, maintenance and minor grid-related savings, so results should not be read as full life-cycle appraisals. Validation was limited to two residential users over winter and spring months, excluding cooling-driven summer loads, and the authors flag extending the validation to larger samples and full-year data as a priority. They also acknowledge that statistically derived profiles may produce overly homogeneous load curves for similar users, and propose adding randomisation functions to introduce greater variability, alongside richer datasets for non-residential demands and, eventually, a graphical interface to lower entry barriers further.

Even with those caveats, the significance of the work lies in what it demonstrates: that a transparent, spreadsheet-based workflow requiring nothing more than monthly bills, basic building typology and location data can reliably discriminate between design choices that differ by thousands of euros in annual income. Rather than trying to outperform calibrated building simulations or mixed-integer optimisation frameworks, the method establishes a pragmatic intermediate layer between simple calculators and high-complexity modelling — one matched to the informational realities of citizen-led initiatives. As Europe pushes energy communities from legal text to physical reality, tools like this could help determine whether the movement scales or stalls, one small town at a time.