Factors influencing suspended solids concentrations in activated sludge settling tanks

Abstract

A significant fraction of the total mass of sludge in an activated sludge process may be in the settling tanks if the sludge has a high sludge volume index (SVI) or when a hydraulic overload occurs during a rainstorm. Under those conditions, an accurate estimate of the amount of sludge in the settling tanks is needed in order to calculate the mean cell residence time or to determine the capacity of the settling tanks to store sludge. Determination of the amount of sludge in the settling tanks requires estimation of the average concentration of suspended solids in the layer of sludge ( X SB) in the bottom of the settling tanks. A widely used reference recommends averaging the concentrations of suspended solids in the mixed liquor ( X) and in the underflow ( X u) from the settling tanks ( X SB=0.5{ X+ X u}). This method does not take into consideration other pertinent information available to an operator. This is a report of a field study which had the objective of developing a more accurate method for estimation of the X SB in the bottom of the settling tanks. By correlation analysis, it was found that only 44% of the variation in the measured X SB is related to sum of X and X u. X SB is also influenced by the SVI, the zone settling velocity at X and the overflow and underflow rates of the settling tanks. The method of averaging X and X u tends to overestimate the X SB. A new empirical estimation technique for X SB was developed. The estimation technique uses dimensionless ratios; i.e., the ratio of X SB to X u, the ratio of the overflow rate to the sum of the underflow rate and the initial settling velocity of the mixed liquor and sludge compaction expressed as a ratio (dimensionless SVI). The empirical model is compared with the method of averaging X and X u for the entire range of sludge depths in the settling tanks and for SVI values between 100 and 300 ml/g. Since the empirical model uses dimensionless ratios, the regression parameters are also dimensionless and the model can be readily adopted for other activated sludge processes. A simplified version of the empirical model provides an estimation of X SB as a function of X, X u and SV f and can be used by an operator when flow conditions are normal.