Modelagem hidrológica de um módulo experimental de pavimento permeável poroso utilizando o EPA SWMM

Abstract

There is limited documentation on sensitivity analysis using SWMM LID module parameters. Recent studies reiterate the need for research to perform calibration, validation and sensitivity analysis using this software using its LID module. The objective of this work was to perform the computational modeling of a permeable pavement module using the EPA SWMM software. The experimental module of permeable pavement was covered with an interlocking block of porous concrete, and implemented in the Campus of the Federal Technological University of Paraná (UTFPR) in Pato Branco.The characterization of the materials used in filling the pavement layers and their dimensioning was carried out. The second stage comprised the construction and monitoring of the module. Precipitation data were obtained by a rain gauge installed on site and water depth data were obtained with the aid of pressure transducers installed in the reservoirs downstream of the bottom drain and in the surface runoff reservoir. Modeling of the porous pavement was performed using the SWMM LID module. From the intensity duration frequency (IDF) equation of Pato Branco, artificial rains were created using the alternating block method to perform the sensitivity analysis and for the steps of model calibration and validation, rains and blades measured in the field were used. The sensitivity analysis indicated that precipitation characteristics can influence the sensitivity coefficient of the parameters. The most sensitive parameters for this modeling were the drain offset, the porosity of the soil layer and the void ratio of the storage layer. The performance of the computational model was evaluated using the NashSutcliffe (NS) coefficient, the Root Mean Square Error (RMSE) and the Mean Absolute Error (MAE). Two precipitation events were used for calibration. In the best calibrated event, a NS coefficient of 0.8804, RMSE of 0.0003 and EMA of 0.0037 were obtained. Then, the validation was performed with two precipitation events, the best of which was obtained a NS of 0.8790, RMSE of 0.02575 and EMA of 0.0315. Good validation results were obtained when using a similar calibration event. The results of this work show that, in the sensitivity analysis, attention should be paid to the return time and rainfall intensity, as the rainfall characteristics can influence the result of the sensitivity analysis. The model was considered validated for rainfall similar to that used in the calibration itself.