Researchers Senthil Krishnamurthy and Nteka Maletsie Mojela have evaluated an adaptive mathematical method known as the Dynamic Arithmetic Optimization Algorithm to schedule energy dispatch across a hybrid renewable microgrid model. The investigation, published in Discover Energy, examines a simulated 24-hour operational cycle designed to support reliable and affordable rural electrification while decreasing reliance on imported electricity from the national grid. By coordinating photovoltaic panels, wind turbines, battery energy storage systems, and bidirectional grid connections, the numerical framework seeks to balance fluctuating consumer demand with intermittent local generation. The computational dispatch model demonstrated that dynamically adjusting exploration and exploitation phases allows an optimization solver to maintain power balance while managing system operational expenses, battery degradation, and power curtailment.
Microgrids intended for rural electrification frequently encounter significant operational challenges due to the intermittent nature of solar irradiance and wind velocity, combined with nonlinear battery behavior and multidimensional operating costs. Traditional optimization techniques such as linear programming and conventional metaheuristics like Grey Wolf Optimization often face computational limitations when dealing with severe renewable fluctuations and complex system constraints. To address these operational issues, the researchers applied the Dynamic Arithmetic Optimization Algorithm, an advanced variant of the Arithmetic Optimization Algorithm originally formulated in 2021. While the foundational arithmetic algorithm employs standard arithmetic operators to mimic search spaces, the dynamic variant incorporates adaptive time-varying control parameters that allow search steps to adjust across computational iterations, preventing early algorithm stagnation and improving convergence reliability.
The mathematical formulation of the microgrid dispatch environment incorporates an explicit 24-hour operational horizon with hourly scheduling intervals. In this system framework, the decision vector accounts for dispatched photovoltaic generation, dispatched wind power, battery charging power, battery discharging power, grid import power, grid export power, battery state of charge, and unserved load shedding. The microgrid operates under fundamental physical constraints, ensuring that renewable generation does not exceed available solar or wind resources. Furthermore, the operational limits restrict battery charging and discharging rates to prevent simultaneous operation, while battery state-of-charge dynamics are governed by nominal storage capacity, cyclic terminal boundaries, and charging and discharging efficiencies.
Economic parameters within the study encompass multiple distinct objective functions integrated into a normalized multi-objective formulation. The primary economic objective accounts for grid electricity exchange costs, derived from hourly purchase tariffs and export tariffs, alongside non-negative throughput-based battery degradation expenses calculated from equivalent full charge cycles. Additional objective components quantify environmental emission costs tied to grid import emission factors, financial penalties associated with renewable power curtailment, and reliability costs linked to energy not served using the value of lost load. Because monetary expenditures, carbon dioxide emissions, and curtailed kilowatt-hours possess vastly different numerical magnitudes, the optimization algorithm normalizes each objective term before applying weighting coefficients to evaluate overall system performance.
The Dynamic Arithmetic Optimization Algorithm executes its search mechanism by transitioning dynamically between exploratory and exploitative arithmetic phases. Exploratory movements utilize multiplication and division operators to broadly scan the solution space during early iterations, driven by a high dynamic search pressure coefficient. As iterations advance, the dynamic control mechanism progressively shifts priority toward exploitation, deploying addition and subtraction operators to intensively refine solutions around promising local spaces. Adaptive boundary management and dynamic scale factors further reduce sensitivity to high-dimensional nonlinear interactions, enabling the solver to respect strict power-balance equalities, battery state-of-charge constraints, and ramp-rate limits without requiring exhaustive manual parameter tuning prior to execution.
Across the simulated 24-hour test horizon, the total consumer load demand reached approximately 91,400 kilowatt-hours, varying from a minimum of 2,200 kilowatts at the first hour to a peak demand of approximately 5,400 to 5,600 kilowatts at midday. The simulation results revealed that wind energy provided the dominant local renewable contribution, yielding 37,550 kilowatt-hours of dispatched energy from an available potential of 65,473 kilowatt-hours. Meanwhile, solar photovoltaic generation supplied 2,990 kilowatt-hours of dispatched energy between daylight hours 8 and 17, peaking at 435.79 kilowatts during peak midday irradiance. Together, dispatched wind and solar power delivered approximately 40,530 kilowatt-hours, representing a 44.3 percent renewable contribution directly satisfying total daily electricity consumption.
The microgrid storage subsystem was operated selectively across the scheduling period rather than undergoing continuous charging and discharging cycles. Scheduled charging was restricted to hour 1 at 210.53 kilowatts and hour 13 at 11.08 kilowatts, accumulating 221.61 kilowatt-hours of total charging energy. Battery discharging occurred exclusively at hour 12 and hour 24, with each interval delivering 100 kilowatts to support power demands, resulting in a daily discharged total of 200 kilowatt-hours. This constrained utilization pattern limited mechanical battery degradation to a minor cost of 12.648 dollars. The authors noted that prioritizing direct generation resources over intensive battery cycling preserves battery lifespan and minimizes unnecessary cycling losses within rural energy storage infrastructure.
Bidirectional electricity exchange with the utility grid provided necessary balancing support, recording approximately 30,053 kilowatt-hours of imports and 3,689 kilowatt-hours of exports across the daily profile. Grid imports reached maximum levels during peak evening and morning periods, including 2,734 kilowatts at hour 22, 2,666 kilowatts at hour 6, and 2,641 kilowatts at hour 7. Conversely, surplus renewable energy was exported to the main grid during hours 4, 12, 15, 16, 17, 20, and 24, generating export revenues that produced negative net grid costs during those specific intervals. Over the full 24-hour cycle, total grid-related expenditure amounted to approximately 3,956.80 to 3,965.84 dollars, confirming that grid interaction functioned dynamically rather than as a static base supply.
Evaluation of the multi-objective cost breakdown demonstrated that reliability and unserved energy penalties formed the dominant component of the total objective function, representing 253,100 dollars within a combined total evaluated figure of 258,037.25 dollars. Direct grid operational expenses constituted 3,956.80 dollars, emissions costs accounted for 676.19 dollars, renewable curtailment penalties totaled 291.61 dollars, and battery degradation contributed 12.648 dollars. In terms of overall resource efficiency, solar photovoltaic utilization reached 63.873 percent with 36.127 percent curtailment, whereas wind energy utilization registered at 57.347 percent. Throughout all 24 hours of operation, hourly power-balance error values remained numerically negligible, verifying that generation, grid exchange, storage actions, and consumer demand matched precisely across the simulation.
Statistical evaluation over 20 independent optimization executions highlighted both the capabilities and the variability inherent in stochastic metaheuristic search processes. The best recorded run achieved a weighted objective score of 0.55241750 with a minimal physical constraint violation of 4.749422 times 10 to the sixth, whereas the worst recorded run reached an objective value of 1,279.8, producing an overall standard deviation of 285.99. The authors emphasized that the deterministic formulation assumed perfect 24-hour foresight for weather profiles, resource availability, and electrical loads without considering communication delays or real-time forecast uncertainties. Consequently, the study highlights the value of multi-run evaluations and points to future research needs surrounding real-world validation and stochastic intermittency management in rural microgrid environments.
At specific operating intervals, the algorithm’s decisions revealed complex trade-offs between local generation and grid dynamics rather than simple renewable maximization. For instance, at hour 17, the system fully absorbed all 4.271 megawatts of available wind capacity, whereas during hours 12, 22, and 23, available wind power measuring 1.881 megawatts, 4.349 megawatts, and 3.587 megawatts was assigned zero dispatch. Similarly, at hour 12, the microgrid simultaneously discharged 100 kilowatts from storage and exported 553.79 kilowatts to the utility grid while meeting a substantial 5,400-kilowatt load demand, a condition the authors identified as requiring careful economic justification. The data also captured an operational discrepancy at hour 9, where dispatched photovoltaic power of 346.72 kilowatts slightly surpassed the recorded available solar generation of 326.84 kilowatts.
Beyond standard metaheuristic formulations, the researchers structured the continuous optimization process to bypass binary switching variables by implementing a complementary product constraint where simultaneous charging and discharging power multiplies to zero. The dynamic arithmetic search mechanism progresses through a ten-step procedure regulated by normalized iteration intervals, time-varying dynamic search pressure, and adaptive coefficients. During initial iterations, exploration probabilities dominate as multiplication and division operators sample candidate schedules across broad operating boundaries. As the iterative process advances, the dynamic balance transitions toward exploitation, using addition and subtraction operators alongside self-adjusting learning rates and local search intensification to fine-tune hourly power schedules around optimal solutions.
The authors contextualized their findings against persistent gaps in energy planning literature, observing that previous optimization research has predominantly focused on isolated subsystems, electric vehicle infrastructure, or automated power trading rather than unified rural electrification. By structuring their evaluation across distinct microgrid configurations that span isolated resource pairings to integrated wind, solar, storage, and utility connections, the investigation provides an adaptable baseline for evaluating renewable fractions against grid dependency. However, because the computational formulation relies on deterministic forecasting of solar irradiance, wind profiles, and consumer load over an idealized 24-hour cycle without real-time communication delays or weather volatility, the researchers emphasized that expanding the framework to incorporate stochastic intermittency and field validation remains crucial for practical rural deployments.
Subject of Research: Dynamic arithmetic optimization for hybrid renewable microgrid energy management
Article Title: Performance-enhanced dynamic arithmetic optimization algorithm for energy management in a hybrid renewable microgrid system
Article References: Krishnamurthy, S., & Mojela, N. M. (2026). Performance-enhanced dynamic arithmetic optimization algorithm for energy management in a hybrid renewable microgrid system. Discover Energy, 6(1), Article 108. https://doi.org/10.1007/s43937-026-00218-4
Image Credits: AI Generated
DOI: 10.1007/s43937-026-00218-4
Keywords: microgrid optimization, Dynamic Arithmetic Optimization Algorithm, renewable energy management, solar photovoltaic, wind energy dispatch, battery energy storage system, rural electrification, power balance constraint, grid exchange, energy curtailment, metaheuristic algorithms, state of charge, load demand scheduling, battery degradation, multi-objective optimization, energy economics
News Source: Faith Mcneil. (November 20, 2026). Dynamic Arithmetic Algorithm Coordinates Power Dispatch in Hybrid Renewable Microgrid Model. Scienmag.



