In this study, the drag exerted by an accelerating fluid on a stationary 2D circular cylinder is numerically investigated using Fluent 19.2 based on the finite-volume method. The SST k–ω model is chosen as the turbulence model because of its superiority in treating the viscous near-wall region. The results are compared to literature, and the numerical methods are validated. The acceleration of the inflow is analyzed for the range of 0.0981–9.81 m/s², and the drag for each acceleration is compared. Additionally, the effect of the initial velocity on the drag acting on the circular cylinder is investigated at two initial velocities. As a result, a supercritical region, typically found under steady state conditions, is observed. Furthermore, vortex shedding is observed at a high initial velocity. This flow characteristic is explained via comparison with respect to the recirculation length and separation angle.

This study aims to enhance the accuracy and interpretability of flood susceptibility mapping (FSM) in Seoul, South Korea, by integrating automated machine learning (AutoML) with explainable artificial intelligence (XAI) techniques. Ten topographic and environmental conditioning factors were selected as model inputs. We first employed the Tree-based Pipeline Optimization Tool (TPOT), an evolutionary AutoML algorithm, to construct baseline ensemble models using Gradient Boosting (GB), Random Forest (RF), and XGBoost (XGB). These models were further fine-tuned using Bayesian optimization via Optuna. To interpret the model outcomes, SHAP (SHapley Additive exPlanations) was applied to analyze both the global and local contributions of each factor. The SHAP analysis revealed that lower elevation, slope, and stream distance, as well as higher stream density and built-up areas, were the most influential factors contributing to flood susceptibility. Moreover, interactions between these factors, such as built-up areas located on gentle slopes near streams, further intensified flood risk. The susceptibility maps were reclassified into five categories (very low to very high), and the GB model identified that approximately 15.047% of the study area falls under very-high-flood-risk zones. Among the models, the GB classifier achieved the highest performance, followed by XGB and RF. The proposed framework, which integrates TPOT, Optuna, and SHAP within an XAI pipeline, not only improves predictive capability but also offers transparent insights into feature behavior and model logic. These findings support more robust and interpretable flood risk assessments for effective disaster management in urban areas.

This study investigates the impact of building insulation standards on indoor thermal environments and the risk of heat-related illnesses during heatwaves in South Korea. Indoor temperatures were measured in residential buildings located in Chuncheon and Gwangju during the 2022 heatwave, with outdoor temperature data sourced from the Korea Meteorological Administration. Probability distribution fitting was used to estimate the likelihood of indoor temperatures exceeding the critical threshold of 27 °C. Additionally, a linear regression analysis was conducted to examine the relationship between the probability of exceeding the threshold and heat-related illness data from 2017 to 2023 provided by the Korea Disease Control and Prevention Agency. The findings reveal significant variations in indoor thermal conditions during heatwaves, influenced by factors such as building type, year of construction, and climate region, which affect the thermal insulation performance. Buildings with a lower thermal insulation performance were associated with higher indoor temperatures, increasing the likelihood of exceeding the critical threshold and contributing to a higher incidence of heat-related illnesses, particularly in provincial non-metropolitan areas. These results underscore the need for region-specific building insulation standards that address both winter energy efficiency and summer heatwave resilience. Enhancing thermal insulation in vulnerable regions could significantly reduce the risk of heat-related illnesses and improve public health resilience to extreme heat events.

Mt. Baekdu’s eruption precursors are continuously observed and have become a global social issue. Volcanic activities in neighboring Japan are also active. There are no direct risks of proximity-related disasters in South Korea from the volcanic eruptions at Japan or Mt. Baekdu; however, severe impacts are expected from the spread of volcanic ash. Numerical analysis models are generally used to predict and analyze the diffusion of volcanic ash, and each numerical analysis model has its own limitations caused by the computational algorithm it employs. In this study, we analyzed the PUFF–UAF model, an ash dispersion model based on the Lagrangian approach, and observed that the number of particles used in tracking substantially affected the results. Even with the presence of millions of particles, the concentration of ash predicted by the PUFF–UAF model does not accurately represent the dispersion. To overcome this deficit and utilize the computational efficiency of the Lagrangian model, we developed a PUFF–Gaussian model to consider the dispersive nature of ash by applying the Gaussian dispersion theory to the results of the PUFF–UAF model. The results of the proposed method were compared with the field measurements from actual volcanic eruptions, and the comparison showed that the proposed method can produce reasonably accurate predictions for ash dispersion.

This study aims to improve the accuracy and interpretability of flood susceptibility mapping (FSM) by integrating automated machine learning (AutoML) with explainable artificial intelligence (XAI) frameworks. Using spatial data from Seoul, South Korea, ten topographic and environmental factors were selected as model inputs. Initially, the Tree-based Pipeline Optimization Tool (TPOT), an evolutionary AutoML algorithm, was employed to construct baseline models using Gradient Boosting (GB), Random Forest (RF), and XGBoost (XGB). Subsequently, Bayesian optimization with Optuna was applied for hyperparameter tuning, resulting in enhanced performance, with the GB model achieving the highest AUC of 0.966. To ensure interpretability, SHAP (SHapley Additive exPlanations) was used to analyze both global and local feature contributions. The SHAP summary and dependence plots identified elevation (DEM), slope, stream distance, stream density, and built-up areas (LULC_7.0) as primary drivers of flood risk. Force and waterfall plots provided instance-level insights into model behavior, validating prediction logic across various geospatial contexts. In addition, Optuna’s visualization tools, such as optimization history, parallel coordinate plots, and hyperparameter importance graphs, offered transparency into how parameter spaces affect performance. These findings highlight the critical role of hyperparameter tuning in FSM and demonstrate the synergy of combining TPOT, Optuna, and SHAP within an XAI framework. The proposed approach not only improves predictive performance but also ensures interpretability, supporting more transparent and data-driven flood risk management.

Investigating flow separation regions on the surfaces of three-dimensional bluff bodies in turbulent flows is important because these regions can induce significant aerodynamic loads. Separation bubbles can generate extreme pressures, making the roof components of low-rise buildings vulnerable. In this study, proper orthogonal decomposition (POD) was applied to wind-induced roof pressures to elucidate the physical significance of the dominant modes. Based on the interpretation of the first mode from the POD, the mean reattachment length of the roof separation bubbles on a low-rise building model in turbulent flow was determined. The mean reattachment length derived from the POD was then compared with the length obtained from an aerodynamic database. For the centerline of the roof, the mean reattachment length based on the POD aligned well with that from the aerodynamic database, showing a difference of less than 5%. This study highlights the efficacy of POD as a powerful tool for estimating the reattachment length of separation bubbles on bluff bodies.