Computational fluid dynamics for predicting performance of ultraviolet disinfection – sensitivity to particle tracking inputs

Abstract

A three-dimensional (3-D) computational fluid dynamic model that predicts the performance of a full-scale medium-pressure lamp ultraviolet (UV) reactor for disinfection of drinking water is described. The model integrates velocity field, fluence rate distribution, and particle trajectory calculations with a microorganism inactivation kinetic model to arrive at predictions of reduction equivalent dose and microorganism inactivation for MS2 coliphage. A rational approach to determining an appropriate number of fluid particles that would generate the required computational precision is presented. Predictions of inactivation and equivalent dose were found to be sensitive to computational mesh geometry (hexahedral versus tetrahedral) but were less sensitive to the value of the Lagrangian empirical constant used in the random walk model and to choice of turbulence model (κ – εε versus Reynolds stress). Non-steady-state (dynamic) simulations produced results that were similar to those of steady-state simulations. Utility of the model for evaluating different lamp operating modes and alternative physical arrangements of the baffles and lamps was demonstrated.Key words: ultraviolet, UV reactor, disinfection, water, computational fluid dynamics, modeling.