Introduction: Urban large-scale construction projects, while playing a prominent role in the economic and physical development of cities, generate extensive and complex environmental impacts that require intelligent decision-making tools for effective management. The construction industry alone accounts for approximately 40-50% of global carbon emissions, 20-50% of natural resource consumption, and 50% of total solid waste worldwide. With the acceleration of urbanization and rapid population growth, these projects have become one of the most significant sources of pressure on the environment. Traditional environmental impact assessment (EIA) approaches, particularly in developing countries, often lack systematic mechanisms for risk identification, impact mitigation strategy evaluation, and real-time decision support. In recent years, advances in artificial intelligence and knowledge-based systems have opened new horizons for improving environmental decision-making in construction projects. However, a significant gap remains in the development of comprehensive, knowledge-driven decision support frameworks specifically designed for environmental impact assessment of urban large-scale construction projects.
Methodology: This research employed a mixed-methods approach with a sequential exploratory design, combining qualitative and quantitative techniques. The qualitative phase involved semi-structured interviews with 30 experts from the Environmental Protection Organization, academic institutions, and the construction engineering sector. Participants were selected through purposive and snowball sampling methods, with interviews continuing until theoretical saturation was achieved. Interview data were analyzed using open, axial, and selective coding procedures following the grounded theory approach proposed by Strauss and Corbin (1990). To ensure trustworthiness, the study employed multiple strategies including data triangulation, member checking, peer debriefing, and inter-coder reliability assessment (calculated at 83%, exceeding the acceptable threshold of 60%).
Based on the qualitative findings, a questionnaire was developed to assess the causal relationships and hierarchical structure among identified factors using a 10-point Likert scale. The quantitative sample consisted of 234 experts and specialists in environmental and construction engineering fields in Tehran province, selected through stratified random sampling. For data analysis, two complementary techniques were employed: Fuzzy DEMATEL for identifying causal relationships and determining the influence and dependence of each factor, and Interpretive Structural Modeling (ISM) for leveling the factors and establishing their hierarchical relationships. The DEMATEL method was implemented through four steps: (1) forming the direct relationship matrix (Z), (2) normalizing the matrix using N=S×ZN=S×Z, (3) calculating the total relationship matrix using T=N×(I−N)−1T=N×(I−N)−1, and (4) determining the threshold value and constructing the causal graph. The ISM method involved forming the reachability matrix based on significant relationships from DEMATEL results, followed by level partitioning through intersection of reachability and antecedent sets.
Results: The qualitative analysis identified 24 environmental impact factors organized into five paradigmatic categories: causal conditions (9 concepts: personal factors, pricing, quality, organizational mission, communication/public relations, structure, economic factors, individual project factors, advertising and awareness), intervening conditions (5 concepts: increasing competition, creativity and innovation, profit margin, access limitations, cost increases), core phenomenon (environmental impact factors of large-scale construction projects), strategies (6 concepts: reactive measures, incentive schemes, feedback, planning, technology, training), and consequences (3 concepts: financial outcomes, structural outcomes, cultural outcomes).
Conclusion: The proposed framework offers practical benefits for various stakeholders: environmental organizations can utilize it to systematize EIA procedures and enhance assessment quality; municipalities can integrate it into urban planning processes to support sustainable development; construction firms can employ it to identify and mitigate environmental risks during project planning and execution; and policymakers can leverage it to develop evidence-based environmental regulations and standards. The framework serves as a foundation for developing intelligent systems based on fuzzy neural networks and other AI methods in future research, contributing to the broader goal of sustainable urban development.