Mehmet Ozgur Kayalica

Agent-based modelling (ABM) is increasingly used to analyze rooftop solar photovoltaic (PV) diffusion, yet it remains unclear how existing models operationalize socio-technical mechanisms such as peer effects, grid constraints, policy uncertainty, and energy justice. This systematic review evaluates ABM architectures for rooftop PV diffusion through a three-layer evidence design: a mapping of 876 publications (2000–2025), a core set of 101 ABM-related studies, and structured coding of 50 focal ABMs using the Spatial–Economic–Mechanistic–Policy–Justice (S-E-M-P-J) framework, which tests whether mechanisms are formally implemented rather than narratively invoked. The coding reveals an uneven architecture: behavioural mechanisms are nearly universal (96%) and policy representations common (82%), yet only 2 of the 50 focal studies (4%) combine architectural embedding of justice-related variables with justice as a core evaluation criterion, and justice remains computationally isolated from the models’ economic, spatial, and policy machinery. The field shows signs of an integration plateau: the median model implements three of the five dimensions, unchanged for a decade, and only 3 of the 41 policy-implementing models treat policy as time-varying. Current models answer “how much and where”, but rarely “for whom”. The review distils a methodological agenda into a minimal justice-reporting standard for adoption ABMs.

Falling PV costs alone do not explain who installs rooftop solar. In Türkiye, the segment with the largest technical potential—residential—shows almost no realised adoption, while industrial rooftops lead deployment; models that reduce adoption to a representative payback-driven adopter cannot explain this sectoral and regional divergence. This paper develops a spatial agent-based model of rooftop PV diffusion for Türkiye, resolving 81 provinces and three consumer segments (residential, commercial, industrial). Each province–sector agent adopts through the proposed S-E-M-P-J utility framework, a discrete-choice formulation integrating spatial suitability, economic return, peer influence, policy/grid access and financial capability, with a second-stage financial-access filter converting adoption intention into realised installation and endogenous learning, peer diffusion and grid-saturation feedbacks. The model is calibrated on reconstructed province-level installation records for 2021–2024 and independently validated against 2025 observations, reproducing aggregate dynamics and province-level spatial patterns with the strongest out-of-sample performance in the industrial segment. Results reveal a structurally asymmetric, enterprise-led diffusion regime: by 2025 the industrial segment reaches the early-majority boundary (17.8% of technical potential), whereas residential adoption remains in a pre-innovator stage (0.04%). Across seven pathways, 2040 capacity ranges from 23.7 GW under macro-financial stagnation to 34.6 GW under an upper residential-diffusion sensitivity case. Pathways that broaden residential participation also lower spatial inequality and raise avoided emissions, indicating that equity and decarbonisation objectives can reinforce each other within the modelled scenarios. The findings shift rooftop PV policy from uniform incentives towards sector-specific strategies targeting the constraints that prevent technical potential from becoming realised adoption.

This study presents a multi-stage stochastic optimization model for the design and operation of a grid-connected hybrid microgrid integrating solar PV, wind turbine, biogas power plant, diesel generator, battery storage, and hydrogen energy systems. The model focuses on minimizing total system cost while meeting energy and hydrogen demands and maximizing renewable utilization and reliability under uncertainties in solar irradiation, wind speed, ambient temperature, and load demand. The scenarios in the study were generated through Monte Carlo simulation and reduced using the K-means clustering algorithm to ensure computational efficiency. The multi-stage stochastic optimization model was solved using the Stochastic Dual Dynamic Integer Programming (SDDiP) algorithm. In this study, SDDiP significantly improved computational efficiency, reducing solution time by 97.14% for large scenario sets while still providing effective optimization results. Across five case studies, net annual cost decreased from 3.692 M€/y in the small-scale baseline to 1.597 M€/y in the industrial hub case, and the medium-scale high-price case achieved net profitability (0.894 M€/y), highlighting both the effects of green hydrogen scale and the influence of offtake agreements under constrained grid access. The optimized designs reduced the grid energy share from 30.6% to 0.2%. These findings demonstrate that the proposed renewable-based hydrogen microgrid framework can improve grid resilience, increase energy autonomy, and enable cost-effective decarbonization under uncertainty.

This study investigates the impact of energy, economic, and environmental policies on Türkiye’s energy dynamics and CO₂ emissions using climate models and an agent-based simulation (ABM) framework. By integrating climate projections with policy scenarios, it assesses how energy transitions and climate change affect renewable energy sources (RES), cooling demands, and CO₂ emissions from electricity generation. Methods include selecting suitable Global Climate Models (GCMs), evaluating climate change impacts on RES performance, and simulating policy effects through ABM across one base and nine policy scenarios from 2023 to 2040. The study highlights the critical role of policy interventions in influencing emissions trends and energy prices. It identifies renewable energy subsidies and low-carbon strategies, such as nuclear power support, as effective tools for reducing emissions and stabilizing energy costs. The methodologies and findings provide actionable insights for policymakers globally, emphasizing the importance of integrating climate data with policy planning.