| Sumario: | Adsorption onto activated carbon in a fixed bed reactor has been found to be an effective technology for removing many hazardous organics from water and wastewater. Current fixed bed reactor design procedures can be both expensive and time consuming. Mathematical process modeling can reduce the cost and time by requiring less laboratory and pilot scale experimentation. A model will provide the most utility if it is stable and accurate for a wide variety of system conditions and has a high degree of computational efficiency. The numerical techniques that have been used over the years in modeling the activated carbon adsorption in a fixed bed reactors are finite differences and the method of orthogonal collocation. The method of orthogonal collocation is currently favored over finite differences because the later is computationally intensive. This method, however, leads to an algorithm which is still computationally expensive and introduces oscillation in the spatial direction. Furthermore, the stability analysis of such an algorithm can be quite difficult. Consequently, the objectives of this work were to develop stable, efficient and convergent finite element models for single and multisolute activated carbon adsorption in a fixed bed reactor. For single-solute systems the use of the semi-discrete Galerkin finite element method with an asymmetric interpolating function was found to eliminate the spatial oscillation. Most important of all, the model was found to be stable, computationally less intensive than orthogonal collocation, and the model predictions were in good agreement with the experimental results. Two finite element models were developed for multisolute adsorption; the semi-discrete Galerkin model with asymmetric basis function and the Characteristic-Galerkin model. Both the models were found to be stable, computationally less intensive than orthogonal collocation, and convergent. The semi-discrete Galerkin finite element model with asymmetric basis function, however, was found to describe better the experimental results for a bisolute mixture of p-chloro phenol and dodecylbenzenesulfonate.
|
|---|