Chemical Engineering Science 58 (2003) 859–865 www.elsevier.com/locate/ces Computational uid dynamic (CFD) simulation of a pilot-scale annular bubble column photocatalytic reactor V.K. Pareek a , S.J. Cox b , M.P. Brungs b , B. Young c , A.A. Adesina b; * a Department of Chemical Engineering, Curtin University of Technology, GPO Box U1987, Perth, WA 6845, Australia b Reactor Engineering and Technology Group, School of Chemical Engineering & Industrial Chemistry, University of New South Wales, Sydney, NSW 2052, Australia c Department of Chemical & Petroleum Engineering, University of Calgary, 2500 University Drive NW, Calgary, Alberta, Canada T2N 1N4 Abstract The behavior of an 18-l pilot-scale photocatalytic reactor has been investigated using a computational uid dynamic (CFD) approach. The granular Eulerian model was used to describe the multiphase ow system. Solid recirculation was predicted while liquid velocity vectors were inuenced by the gas ow. The companion radiation transport equation was iteratively solved using a nite-volume-based discrete ordinate method. The rst-order photodegradation kinetics of spent Bayer liquor previously studied in the same reactor was used to evaluate the CFD simulation. A Pearson correlation coecient 0.974 between simulated and experimental data is indicative of model adequacy. ? 2003 Elsevier Science Ltd. All rights reserved. Keywords: Photoreactor; Three-phase reactors; Photocatalytic; Computational uid dynamics (CFD) 1. Introduction There has been a urry of research activities in the de- velopment of photocatalyzed destruction of organic wastes in aqueous media within the last two decades (Pelizzetti, 1999; Yue, 1997). In most cases, TiO 2 is used as a photo- catalyst in the presence of pure oxygen or air. Consequently, the performance of a photocatalytic reactor is not only inu- enced by gas–liquid mixing but also by distribution of cat- alyst particles in the reaction space. In fact, the distribution of photocatalyst particles in the reactor determines the light intensity distribution and the local volumetric rate of energy absorption (LVREA), and hence the rate of photo-activated step(s). Therefore, in order to design and analyze the per- formance of a photocatalytic reactor for possible industrial wastewater treatment, it is important to take cognizance of the interaction between hydrodynamic attributes and the ra- diation characteristics (Pareek, Brungs, & Adesina, 2001a). Multiphase ow in reactors can be modeled using either of the two approaches—Eulerian–Lagrangian (EL) and ∗ Corresponding author. Tel.: +61-2-9385-5268; fax: +61-2-9385- 5966. E-mail address: a.adesina@unsw.edu.au (A.A. Adesina). Eulerian–Eulerian (EE) (Ranade, 2002).TheELapproachis more computationally intensive because it allows the track- ing of particles (or bubbles) as discrete entities in the con- tinuous (liquid) phase. However, the EE model regards both bubble and liquid as a continuous liquid phase and therefore requires less computer memory. As a result, it is appropriate for large-scale multiphase reactor modeling (Ranade, 2002). The solution of the radiation transport equation (RTE) gives the light intensity distribution inside the heterogeneous pho- tocatalytic reactor. Since the RTE is an integro-dierential equation, an exact analytical solution is possible only for highly ideal one-dimensional situations. Carvalho and Farias (1998) have presented a state of art review of various methods developed to numerically solve the RTE. Spadoni, Bandini, and Santarelli (1978) used a Monte-Carlo ap- proach to simulate the light scattering in photocatalytic reactors. Recently, a discrete ordinate model (DO model) was proposed for the solution of the RTE in photocat- alytic systems (Romero, Alfano, & Cassano, 1997; Sgalari, Camera-Roda, & Santarelli, 1998). In this paper, we present an integrated computational uid dynamic (CFD) model for a three-phase photocatalytic reac- tor. The interaction between three-phases in the photoreac- tor has been modeled using an Eulerian–Eulerian approach and a nite-volume variant of the DO model has been used 0009-2509/03/$ - see front matter ? 2003 Elsevier Science Ltd. All rights reserved. doi:10.1016/S0009-2509(02)00617-6