The performance of stormwater infiltration systems depends significantly on the infiltration/filtration characteristics of filter media. Sand media have been frequently used due to their abundance and acceptable pollutant removal efficiency. Bottom ash media, which was known to have a high sorption capacity, are being considered as alternative and/or complement filter media to enhance the overall treatment efficiency. To verify the applicability of bottom ash as filter media, a series of 1-D column infiltration tests were conducted in this study. The filtration and clogging characteristics of the sand-bottom ash mixtures at different mixing ratios were investigated with limitedly focusing on the Total Suspended Solids (TSS) present in stormwater runoff. As a result, the presence of bottom ash was identified to improve the TSS removal efficiency in spite of having a significant increase in clogging potential. All media configurations were observed to experience the premature clogging and the clogging state for varying and constant head flow conditions, respectively. The noticeable amount of crushed bottom ash particles brought by seepage force could be accountable for the rapid occurrence of clogging phenomena. Overall, a homogeneous sand filter is preferred as long as the lifespan of filter system is the primary concern.
31 October 2021 in Journal of the Korean Society of Hazard Mitigation
Neural network models based on deep learning algorithms are increasingly used for estimating pile load capacities as supplements of bearing capacity equations and field load tests. A series of hyperparameter tuning is required to improve the performance and reliability of developing a neural network model. In this study, the number of hidden layers and neurons, the activation functions, the optimizing algorithms of the gradient descent method, and the learning rates were tuned. The grid search method was applied for the tuning, which is a hyperpameter optimizer supplied by the developing platform. The cross-validation method was applied to enhance reliability for model validation. An appropriate number of epochs was determined using the early stopping method to prevent the overfitting of the model to the training data. The performance of the tuned optimum model evaluated for the test data set revealed that the model could estimate pile load capacities approximately with an average absolute error of 3,000 kN and a coefficient of determinant of 0.5.
31 August 2021 in Journal of the Korean Society of Hazard Mitigation
A deep learning model was developed to estimate the clogging of voids in bottom-ash-mixed sand filters. Filter clogging is characterized by a decrease in the infiltration ratio with an increase in the outflow. The estimation of clogging for time-series forecasting is a regression problem. Various algorithms for regression problems have been applied through machine learning and deep learning. Despite the various algorithm applications, an application for clogging appears to be emerging. Studies are required to develop dependable models for precisely predicting clogging. A convolutional neural network (CNN) clogging model was developed in this study to estimate the relationship between the infiltration ratio and outflow quantity using features defined by the mixing ratio between sand and ash. Significant clogging data required for the effective maintenance and stable operation of stormwater infiltration filters were obtained using the proposed model. The CNN clogging model is recommended as a supplement to conventional theoretical models and expensive experimental models.
30 June 2023 in Journal of the Korean Society of Hazard Mitigation