... ZVI has proven to be an effective material for removing multiple contaminants, including arsenic species [13][14][15], NO 3 − [16][17], NO 2 − , hexavalent chromium [18], heavy metals [19,20], as well as halogenated and nitrated organic compounds [21]. In the last 20 years, numerous studies have been conducted to assess the dominant processes of contaminant removal in ZVI/H 2 O systems [14,17,[22][23][24][25]. The investigated experimental conditions include iron source, particle size, presence of dissolved oxygen, contaminant concentration, solution chemistry, chemical modification of the particle surface, the use of inert materials mixed with ZVI in different proportions, and the material Abbreviations: ANN, artificial neural network; CA, cluster analysis; CL, corrosion layer; D, input data matrix; E, square error; f, number of neurons in the input layer; HIPR, holdback input parameter randomization method; HRT, hydraulic retention time; LS, learning set; MSE, mean square error; m ZVI , ZVI loading; NH, number of neurons in the hidden layer; P, loadings matrix in PCA; PC, principal component; PCA, principal component analysis; r, number of neurons in the output layer; RSSCT, rapid small-scale column tests; T, scores matrix in PCA; TS, test set; VS, validation set; X 1 , X 2 , X 3 , controlled factors; Y, response variables; ZVI, zerovalent iron; δ ZVI , ZVI packing density. ...