• Resource Scheduling Paradigms: The study evaluates four paradigms—Deterministic, Stochastic, Robust, and Hybrid—for optimizing resource scheduling in an off-grid microgrid, emphasizing cost and reliability. • Hybrid Approach Success: The Hybrid Paradigm achieved the lowest operating cost of ₹105,216 with an effective energy not served (EENS) of 0.046, significantly improving reliability while increasing renewable penetration to 75.23%. • Scenario Reduction with k-Medoids: A total of 503 scenarios were generated using 10 years of historical data, and scenario reduction was effectively implemented through the k-medoids clustering method, resulting in 60 clusters that represent diurnal load trends. • Cost-Reliability Trade-offs: The paper reveals the trade-offs among the paradigms, highlighting that while the Deterministic approach is cost-effective, it lacks reliability; conversely, the Robust method enhances reliability but at a higher cost. • Future Enhancements: Recommendations for future research include incorporating multi-day rolling horizons for better shortfall management and refining models to include battery-aging studies and adaptive tuning for feasibility. This paper presents a DSM based reliable optimum resource scheduling for isolated, distributed generation based on off-grid microgrids with minimum operating cost considering uncertainty and reliability under realistic forecasting. The case study considers an off grid microgrid (PV 1.0 MW, WT 0.4 MW, HY 0.3 MW, DG 0.4 MW, BESS 1 MWh) with historical data set to find out the operating cost per day. The realistic study is performed under four paradigms Deterministic, Stochastic, Robust and hybrid. Here the scheduling problem is managed by Pyomo MILP model to minimize day-ahead operating cost where the adequacy is handled through unmet-energy slack and reported by EENS and LOLP. Without mixing the multiple criteria the best result is reported regarding the lowest cost with permissible reliability constraints. Deterministic study is based on point forecast while stochastic methods considering uncertainty. Under these circumstances 503 scenarios are generated based on the 10 years of historical data whereas the baseline load curve is generated using the data from June 2023–June 2024. Now these hourly trajectories (24 h horizon) are reduced by using k-medoid methods where 60 clusters are generated and one medoid is representing the cluster considering diurnal shape of the load curve with the correlation with Demand Side Management. Under the robust paradigm, three stress envelops are created to overcome the renewable shortfall and sudden demand surges after the objective functions are solved deterministically. Under hybrid paradigm clustering and scenario reduction using k-medoid methods is done like stochastic way but after that each medoids are processed under the three robust stress envelops. The main aim of the study is not to find the minimum cost but an optimum cost with EENS as low as possible with a higher renewable penetration. After the study it is seen that Deterministic scheduling is cost efficient (₹111,853) but fragile due to single point forecast (EENS = 0.334) where as Robust is reliable (EENS = 0.041) but conservative and not cost efficient (₹135,125). Stochastic paradigm also struggles with reliability (EENS = 0.294) as the unmet energy is high whereas proposed Hybrid Paradigm balances between the cost (₹105,216) and reliability (EENS = 0.046) also it will increase the renewable penetration (75.23%). This study aims not only to advance the hybrid paradigm but also to strengthen the operator’s decision-making power.
Ghosh et al. (Wed,) studied this question.