11n13 ijaet0313450 revised

Page 1

International Journal of Advances in Engineering & Technology, Mar. 2013. ŠIJAET ISSN: 2231-1963

ARTIFICIAL NEURAL NETWORK APPROACH FOR MODELING OF ADSORPTION OF NI (II) AND CR (VI) IONS SIMULTANEOUSLY PRESENT IN AQUEOUS SOLUTION USING ADSORBENT SYNTHESIZED FROM AEGEL MARMELOS FRUIT SHELL AND SYZYGIUM CUMINI SEED S. L. Pandharipande1, Aarti R. Deshmukh2 1

Associate Professor, 2M. Tech Third Semester, Department of Chemical Engineering, Laxminarayan Institute of Technology, Rashtrasant Tukadoji Maharaj Nagpur University, Bharat Nagar, Amravati Road, Nagpur, India.

ABSTRACT Rapid industrialization is seriously contributing to the release of heavy metals into the main water bodies. Increasing concentration of toxic heavy metals in the water sources constitutes severe health hazard due to their toxicity. Adsorption is one of the methods in effective removal of metal ions from water. Adsorption isotherms are one of the ways of expressing equilibrium relationship between adsorbate and adsorbent however very little has been reported in the literature related to the equilibrium relationship between two sorbets and two adsorbents in one isotherm. Estimation of adsorption isotherms which are adsorbent & adsorbate combination specific, become prerequisite in designing adsorption processes. The conventional mathematical model fails very often in developing correlation because of complexity and nonlinearity governing this process. Artificial Neural Network is the black box modeling tool that can address to the modeling of the operations involving multivariable nonlinear relationship and also can incorporate linguistic variables in coded number form. In present work Artificial neural network has been applied for adsorption of heavy metals Ni (II) & Cr (VI) simultaneously present in aqueous solution using the synthesized adsorbents. Adsorption studies are performed for aqueous solutions in the metal ion concentration range of 0.4938 to 1.2345 mg/ml for Ni (II) and 0.1944 to 0.4418 mg/ml for Cr (VI) with adsorbent dosing of 1-5 gm. Three ANN models ACS, ACM & ACC with different topology have been developed using elite-ANN with sigmoid function for estimation of % adsorption, equilibrium concentration and amount of adsorbent adsorbed per unit adsorbent .These estimated values for all the parameters are compared for their prediction accuracy for both the training and test data sets. The results are indicative that the ANN model ACM with two hidden layers & 10 neurons in each hidden layer has high accuracy compared to other two models. The novel feature of this work is in highlighting the potential of ANN models in substituting adsorption isotherms that will enable the user to incorporate linguistic variables coded with numbers for multiple adsorbents & adsorbates into a single model with ease & high accuracy.

KEYWORDS:

Artificial Neural Network, modeling, heavy metal ion, adsorption, aegel marmelos, syzygium

cumini.

I.

INTRODUCTION

The presence of heavy metals in water is a major concern due to their adverse effect on human health. The discharge of waste water containing heavy metals in the water bodies is thus worrying for toxicological reasons. Industries such as steel, metal plating, mining, textile, explosive manufacturing, ceramic & glass, leather, paint, etc., are some of the sources for heavy metal effluents. The heavy

114

Vol. 6, Issue 1, pp. 114-127


Turn static files into dynamic content formats.

Create a flipbook
Issuu converts static files into: digital portfolios, online yearbooks, online catalogs, digital photo albums and more. Sign up and create your flipbook.