Abstract
Tertiary lagoons have generally been viewed as an effective and low-cost method of removing pathogens from wastewater. A review of the literature showed that there is inconclusive information on removal mechanisms for all parameters. The majority of lagoons in use today have been designed using equations developed by Marais and Shaw (1961) based on first-order kinetics for the removal of faecal coliforms, but there are wide variations in performance. Mathematical models have also been developed, but these are difficult to use for new lagoons as they require an estimate of several complex parameters, pay insufficient attention to the hydraulic regime, and do not consider the systems in three dimensions. Computational fluid dynamic (CFD) modelling has therefore been proposed as a more practical approach to lagoon modelling and design, and the models developed so far are reviewed. Monitoring of tertiary lagoons at three sewage treatment works in the UK was carried out to investigate their performance, their hydraulic regime and the vertical distribution of physical parameters. Strong seasonal variation was seen in the performance of the lagoons. None of the sites were able to produce effluents which complied with the EC bathing water directive or the WHO guidelines for wastewater reuse, although their existing discharge consents were easily met. Tracer studies showed that there was significant short-circuiting occurring in all lagoons, and the hydraulic regime was dispersed plug-flow. Depth profiling suggested there was greater vertical variation in physical parameters than in the horizontal plane. Three-dimensional steady flow CFD modelling of a facultative lagoon demonstrated that there was significant short-circuiting. The addition of baffles improved the hydraulic regime, and indicated that plug-flow conditions produced better performance. Time-dependent modelling, to include thermal effects, found that thermal stratification exacerbated short-circuiting. Baffles did improve the situation, but the retention time was still much shorter than the nominal retention time.