A study of dissolved air flotation tank design variables and separation zone performance

Abstract

A study of velocity and turbidity distributions in the separation zone of dissolved air flotation (DAF) tanks was carried out at Franldey and Trimpley Water Treatment Works of Severn Trent Water. Sampling of velocity was made using an Acoustic Doppler Velocimeter (ADV). The instrument is capable of measuring velocities as low as 1 mmlsec and producing three dimensional velocity data. Sixty-four points set at equal intervals within the tank were monitored and the flow rate corresponding to the velocity for each point was recorded. The same points were used for the sampling of turbidity within the tank and the corresponding flow rate for each sampling point was also recorded. The aim of the study was to establish the relative importance of tank design parameters within the separation zone. The ADV probe was found suitable to be used in the investigation based on the data quality obtained. The study indicated that there are some differences in the flow patterns compared to Computational Fluid Dynamics (CFD) models found in the literature. The plan view contour plots indicated that velocities in the x, y and z directions at a quarter depth from the surface of the tank were unstable with irregular velocity patterns. However the CFD models indicated that the flow at this depth was uniform. Also at this depth the vertical velocity was predominantly downward which suggested that the solid liquid separation process is inefficient. Tank physical parameters were found to have a highly significant effect on the velocity distribution using analysis of variance (ANOVA) and analysis of covariance (ANCOVA). These analyses involved higher order interactions and independent predictor variables. The results from the higher order models are difficult to interpret. Thus simple second-order empirical models were used. Empirical models were developed using regression analyses to describe the observed velocity within the tank. These models are appropriate for design purposes. Although the models are not precise, the standard statistical techniques used for data analyses are found to be useful to compare, analyse and develop the appropriate model from the velocity data obtained during the investigation. In terms of turbidity removal, there was no significant difference in the average turbidity readings between different depths of the tank. Comparison of turbidity at different lengths of the tank indicated that the average turbidity readings were identical between three quarter length of the tank from the baffle and at the extreme end of the tank. The results confirm that there were not enough air bubbles within the separation zone for turbidity removal. The size of tank at Franidey can also be reduced by 15% so that the difference between the average flow rate and the surface area between the tank at Frankley and Trimpley is the same. It is expected that the reduction in size will not affect the turbidity within the separation zone due to non significant turbidity removal within this zone.