Characterization of P-nitrophenol Adsorption Kinetic Properties in Batch and Fixed Bed Adsorbers

P-nitrophenol（PNP） adsorption in batch and fixed bed adsorbers was studied. The homogeneous surface diffusion model（HSDM） based on external mass transfer and intraparticle surface diffusion was used to describe the adsorption kinetics for PNP in stirred batch adsorber at various initial concentrations and activated carbon dosages. The fixed bed model considering both external and internal mass transfer resistances as well as axial dispersion with non-linear isotherm was utilized to predict the fixed bed breakthrough curves for PNP adsorption under the conditions of different flow rates and inlet concentrations. The equilibrium parameters and surface diffusivity（Ds） were obtained from separate experiments in batch adsorber. The obtained value of Ds is 4.187×1012 m2/s. The external film mass transfer coefficient（kf） and axial dispersion coefficient（DL） were estimated by the correlations of Goeuret and Wike-Chang. The Biot number determined by HSDM indicated that the adsorption rate of PNP onto activated carbon in stirred batch was controlled by intraparticle diffusion and film mass transfer. A sensitivity analysis was carried out and showed that the fixed bed model calculations were sensitive to Ds and kf, but insensitive to DL. The sensitivity analysis and Biot number both confirm that intraparticle diffusion and film mass transfer are the controlling mass transfer mechanism in fixed bed adsorption system.

An Application of Nonlinear Autoregressive(NARX)Model to Predict Adsorbent Bed Temperature of Solar Adsorption Refrigeration System

In any solar adsorption refrigeration system,there are three major components:a solar collector adsorbent bed,a condenser and an evaporator.All of those components operate at different temperature levels.A solar collector with a tubular adsorbent configuration is proposed and numerically investigated.In this study,a nonlinear auto-regressive model with exogenous input is applied for the prediction of adsorbent bed temperature during the heating and desorption period.The developed neuronal model uses the MATLAB Network toolbox to obtain a better configuration network,applying multilayer feed-forward,the TANSIG transfer function,and the back-propagation learning algorithm.The input parameters are ambient temperature and the uncontrolled natural factor of solar radiation.The output network contains a variable representing the adsorbent bed temperature.The values obtained from the network model were compared with the experimental data,and the prediction performance of the network model was examined using various performance parameters.The mean square error(MSE)and the statistical coefficient of determination(R2)values are excellent numerical criteria for evaluating the performance of a prediction tool.A well-trained neural network model produces small MSE and higher R2 values.In the current study,the adsorbent bed temperature results obtained from a neural network with a two neuron in hidden layer and the number of the tapped time-delays d=9 provided a reasonable degree of accuracy:MSE=1.0121 and R2=0.99864 and the index of agreement was 0.9988.This network model,based on a high-performance algorithm,provided reliable and high-precision results concerning the predictable temperature of the adsorbent bed.