Predicting the capability of carboxylated cellulose nanowhiskers for the remediation of copper from water using response surface methodology (RSM) and artificial neural network (ANN) models
Predicting the capability of carboxylated cellulose nanowhiskers for the remediation of copper from water using response surface methodology (RSM) and artificial neural network (ANN) models
复制标题
DOI：
10.1016/j.indcrop.2016.05.035
复制
发表时间：
2016-12-25
影响因子：
5.9
通讯作者：
Gomes, Rachel L.
中科院分区：
文献类型：
--
作者：
Hamid, Hazren A.;Jenidi, Youla;Gomes, Rachel L.
本研究观察了温度，初始Cu（II）离子浓度，和吸附剂的用量对Cu（II）从水基质中去除使用表面氧化纤维素纳米晶须（CNWs）轴承羧酸功能的影响。此外，本研究针对某污水处理厂的实际情况。电导滴定的CNWs悬浮液显示的未改性和改性的CNWs，分别为54和410 mmol/kg的表面电荷，这表明，改性的CNWs提供了一个相对较高的表面积比未改性的CNWs每单位质量。此外，在不同条件下测试了改性碳纳米纤维的稳定性，证明了功能基团是永久的，不降解。响应面法（RSM）和人工神经网络（ANN）模型，以优化系统，并建立一个预测模型，以评估Cu（II）去除性能的改性CNW。ANN和RSM模型的性能进行了统计学评价的决定系数（R-2），绝对平均偏差（AAD），和均方根误差（RMSE）预测的实验结果。此外，为了确认模型的适用性，对不属于训练数据集并且位于训练集边界内外的14个新试验进行了看不见的实验。结果表明，人工神经网络模型（R-2 = 0.9925，MD = 1.15%，RMSE = 1.66）在预测Cu（II）去除率时，其R-2、AAD和RMSE均优于响应面模型（R-2 = 0.9541，AAD = 7.07%，RMSE = 3.99），因而更可靠。采用Langmuir和Freundlich等温模型对平衡数据进行拟合，结果表明，Langmuir等温模型（R-2 = 0.9998）比Freundlich等温模型（R-2 = 0.9461）具有更好的相关性。实验数据也进行了测试，在动力学研究方面，使用伪一级和伪二级动力学模型。结果表明，准二级动力学模型能较好地描述吸附过程。(C)2016爱思唯尔B. V.保留所有权利。
This study observed the influence of temperature, initial Cu(II) ion concentration, and sorbent dosage on the Cu(II) removal from the water matrix using surface-oxidized cellulose nanowhiskers (CNWs) bearing carboxylate functionalities. In addition, this study focused on the actual conditions in a wastewater treatment plant. Conductometric titration of CNWs suspensions showed a surface charge of 54 and 410 mmol/kg for the unmodified and modified CNWs, respectively, which indicated that the modified CNWs provide a relatively high surface area per unit mass than the unmodified CNWs. In addition, the stability of the modified CNWs was tested under different conditions and proved that the functional groups were permanent and not degraded. Response surface methodology (RSM) and artificial neural network (ANN) models were employed in order to optimize the system and to create a predictive model to evaluate the Cu(II) removal performance of the modified CNWs. The performance of the ANN and RSM models were statistically evaluated in terms of the coefficient of determination (R-2), absolute average deviation (AAD), and the root mean squared error (RMSE) on predicted experiment outcomes. Moreover, to confirm the model suitability, unseen experiments were conducted for 14 new trials not belonging to the training data set and located both inside and outside of the training set boundaries. Result showed that the ANN model (R-2 = 0.9925, MD = 1.15%, RMSE = 1.66) outperformed the RSM model (R-2 = 0.9541, AAD = 7.07%, RMSE = 3.99) in terms of the R-2, AAD, and RMSE when predicting the Cu(II) removal and is thus more reliable. The Langmuir and Freundlich isotherm models were applied to the equilibrium data and the results revealed that Langmuir isotherm (R-2 = 0.9998) had better correlation than the Freundlich isotherm (R-2 = 0.9461). Experimental data were also tested in terms of kinetics studies using pseudo-first order and pseudo-second order kinetic models. The results showed that the pseudo-second-order model accurately described the kinetics of adsorption. (C) 2016 Elsevier B.V. All rights reserved.