Indian University Campus Could Cut Grid Power Use by Two-Thirds, Study Finds

Shivaji University in Kolhapur, India, already runs 930 kilowatts of rooftop solar panels, but a new planning study argues that the campus could go much further. Researchers developed a framework called CMR-ESAF that models the university’s transition from a grid-dependent consumer to an optimized hybrid renewable energy system microgrid, and their simulation suggests grid electricity imports could fall by 67.8 percent compared with a grid-only baseline.

The study, published in the journal Carbon Neutral Systems, is unusual among campus energy analyses because it does not start from a blank slate. Instead of treating the existing solar fleet as a hypothetical greenfield asset to be re-optimized, the authors treat the 930 kilowatt installation as a fixed baseline and ask a different question: how much additional benefit can storage, flexible demand, coordinated control, and extra renewables deliver beyond what the panels already provide? The framework separates three operating states, labeled S0, S1, and S2, so that planners can see the value already delivered and the incremental value of the next investment stage independently.

State S0 is the reference case: a campus supplied entirely by the utility grid, with no renewables, storage, or demand management. Under the study’s annual hourly simulation assumptions, this baseline requires 3650 megawatt-hours of grid electricity per year, emits 2993 tonnes of carbon dioxide, and reaches a maximum demand of 1080 kilovolt-amperes. State S1 represents the campus as it exists today, with the 930 kilowatt rooftop photovoltaic system connected to the grid. In the simulation, this existing installation lowers grid imports to 2430 megawatt-hours per year, a 33.4 percent reduction in grid dependency, and avoids 1000.4 tonnes of carbon dioxide annually. Notably, the peak demand drops only modestly, from 1080 to 1000 kilovolt-amperes, because rooftop solar generation does not always coincide with the evening or non-solar periods when the campus billing peak occurs.

State S2 is the proposed optimized configuration, and this is where the study’s central finding emerges. The selected design combines 1580 kilowatts of total photovoltaic capacity, meaning roughly 650 kilowatts of expansion beyond the existing fleet, with a 150 kilowatt wind generator, a 500 kilowatt and 1500 kilowatt-hour battery energy storage system, and 130 kilowatts of demand-side management participation, all coordinated by an energy management system. In the model, this combination reduces annual grid imports to 1175 megawatt-hours, cuts maximum demand to 695 kilovolt-amperes, achieves a 72.3 percent renewable fraction, and reaches 65.8 percent renewable self-sufficiency. The modeled annual utility cost falls from 413.6 lakh rupees to 145.7 lakh rupees, a saving of about 267.9 lakh rupees per year.

The decomposition of energy savings is one of the study’s most decision-relevant results. The existing solar plant already saves 1220 megawatt-hours of grid electricity per year, but the optimized hybrid system delivers an additional 1255 megawatt-hours beyond that present-day condition. According to the authors, this additional saving cannot be achieved by expanding solar capacity alone; without storage, increased export and curtailment limit the benefit of extra panels. Battery storage absorbs surplus daytime generation and discharges during peak grid-import periods, while demand-side management shifts or reduces flexible loads such as pumping, schedulable heating and ventilation, and common-area services. Critical laboratories, safety systems, and information technology infrastructure are explicitly excluded from the controllable pool, and the 130 kilowatt figure is described as a planning envelope rather than a claim that all campus loads are interruptible.

The optimization itself uses an adapted version of the NSGA-III evolutionary algorithm, modified for a many-objective, heavily constrained problem. Six independent objectives are minimized simultaneously: annualized campus cost, annual grid import, billing maximum demand, renewable curtailment, a self-sufficiency deficit, and a normalized network stress index combining transformer, feeder, and voltage utilization. Carbon emissions are deliberately excluded from the objective vector because, with a constant grid emission factor, they are exactly proportional to grid-import energy and would not add an independent optimization direction; instead, emissions are reported as a post-optimization indicator. The algorithm incorporates normalized constraint violation measures, a dynamic threshold that allows near-feasible boundary solutions to participate early in the search, angle-penalized reference-vector association, dual archives separating convergence and diversity, and a fuzzy membership rule to select a single best-compromise design from the final Pareto archive.

A key architectural choice keeps the computation tractable. Rather than placing thousands of hourly dispatch decisions in the evolutionary chromosome, the optimizer handles only five continuous capacity variables: additional photovoltaic capacity, wind capacity, battery power and energy ratings, and the demand-side management limit. For every candidate capacity vector, a deterministic hourly energy management system evaluates all 8760 hours of the year using a fixed priority sequence, in which available renewables serve the post-management load first, surplus charges the battery, remaining surplus is exported up to the point-of-common-coupling limit or curtailed, and the battery discharges during tariff and peak windows before the grid covers the residual load. The campus network is represented as a reduced 45-bus planning model with six explicit distribution transformers, and voltage, feeder loading, transformer loading, and reverse-power-flow constraints are checked at every hourly point.

The authors are careful to flag the limits of their evidence. The load profile was reconstructed chronologically to preserve measured annual energy and billing peak, because complete synchronized 8760-point measurements for every bus were not available; the solar series is a typical-year profile scaled to recorded yield, and the wind series is a planning-level profile rather than an on-site measured record. The 150 kilowatt wind unit is therefore conditional on a subsequent hub-height measurement campaign, and procurement would require at least a year of site-specific wind data, wake assessment, and manufacturer power-curve validation. The network layer is a screening model, not a protection-grade power-flow study, and hourly resolution cannot capture sub-hourly ramps or short-duration billing peaks. The battery degradation model is planning-level, triggering one module replacement in year 12 when usable capacity reaches 80 percent of nameplate, rather than an electrochemical aging model.

The economics are calculated incrementally from the existing solar state to the optimized design, treating the 930 kilowatt plant as a sunk asset. The simple payback period is 6.4 years, and the levelized campus electricity cost, a metric the authors distinguish from a conventional generator LCOE because it includes residual grid purchases, is estimated at 4.85 rupees per kilowatt-hour against an assumed grid tariff of 10 rupees per kilowatt-hour. A discounted cash-flow analysis over 25 years, including the year-12 battery replacement, yields a net present value of approximately 4.32 crore rupees at an 8 percent discount rate and an internal rate of return near 14.1 percent. Sensitivity tests indicate the design remains viable under a 10 percent reduction in solar generation, which raises grid imports to 1305 megawatt-hours while keeping the renewable fraction above 65 percent, and under combined adverse cases involving load growth and higher battery costs.

For the wider community of Indian higher-education institutions, many of which have large daytime loads, available rooftops, and high-tension supply connections, the study’s message is that stand-alone rooftop solar is only the first stage of a campus energy transition. The proposed pathway for Shivaji University involves retaining and monitoring the existing plant, expanding rooftop capacity, installing storage, implementing demand-side management within conservative limits, deploying coordinated dispatch, and considering voltage-support equipment such as a 300 kilovolt-ar D-STATCOM, which in a representative case raised the minimum bus voltage from 0.944 to 0.972 per unit. The authors emphasize that the reported values are simulation outputs under stated assumptions and that detailed site surveys, protection studies, and procurement-stage engineering must precede any implementation.

Subject of Research: Techno-economic modeling of a hybrid renewable energy microgrid transition for an Indian university campus

Article Title: From conventional campus grid to optimized HRES microgrid: a grid-constrained techno-economic and environmental assessment of renewable energy integration at Shivaji University, India

Article References: Kamble, A. C., & Jadhav, H. T. (2026). From conventional campus grid to optimized HRES microgrid: a grid-constrained techno-economic and environmental assessment of renewable energy integration at Shivaji University, India. Carbon Neutral Systems, 2(1), Article 21. https://doi.org/10.1007/s44438-026-00041-3

Image Credits: AI Generated

DOI: 10.1007/s44438-026-00041-3

Keywords: microgrid, hybrid renewable energy, solar photovoltaics, battery storage, demand-side management, NSGA-III, campus energy, Shivaji University, grid constraints, peak demand, decarbonization, energy modeling