This paper provides a critical review and analysis of the commonly accepted relationship between filament content and SVI, and demonstrates that a basic misconception of this relationship, propagated for the last three decades, has been based on an artifact of data presentation. This is appears to be an original hypothesis that corrects what appears to be an error in the commonly accepted theory of activated sludge bulking, and so it seems likely to have a broad appeal to both researchers and practioners. The work is about 90% complete.
It has been commonly stated as a demonstrated fact that as filament content increases in activated sludge samples, it reaches a “threshold” content, below which it does not greatly effect on settleability, and above which it has a disproportionately large affect on settleability (for example, Palm et al., 1980; Blackbeard et al., 1986; Hermanowicz, 1993; Lee et al., 1983; Jenkins et al., 2003; Martins et al., 2004; Liao et al., 2004). This idea has been based on data sets of filament content, measured by various means, and the sludge volume index (SVI), or variations of the SVI test. For example, one of the early papers to advance this idea was Palm et al., 1980. Based on this data, was suggested that SVI values increased rapidly above SVI values of 150 mL/g and filament contents of 107 ìm/mL. This claim has been repeated numerous times, both based this data set and similar presentations of other data. The objectives of this paper was to critically evaluate this hypothesis, and to develop a more accurate representation of the relationship between filament content and SVI.
Several data sets from previously published studies were reviewed and analyzed to address the research objectives. In most cases, data was manually taken from published tables or figures for the analyses presented herein. Statistical analyses were carried out using the freely distributed software gnuplot 3.7.
It is the nature of semi-log data plots that linearly correlated data appear to be curvilinearly related. When the Palm et al., 1980 data is presented with filament content on a linear scale, it is apparent that the relationship between filament content and SVI is approximately linear. In particular, when the data is plotted the range of SVI from 0 to 300 mL/g, it is evident that there is no discontinuity in the SVI/filament relationship at SVI of 150 mL/g or at 107 ìm/mL filament
This analysis has been conducted for 8 different data sets, with filament content measured in a variety of ways, including traditional microscopic examination and 16s RNA gene copies, and the results have consistently indicated that the “threshold effect” does not exist. In fact, 6 of the 8 data sets indicated that as filament content increased, the effect on SVI tended to decrease. For example, data taken from Jenne et al., 2005, who operated four sequencing batch reactors with synthetic feed, and found remarkably similar trends. This data indicates a flattening curve with
increasing filament content, which is the opposite trend that would be expected in the case of a threshold-type effect.
Based on this results, it was of interest to develop a new term, the SVIF, SVI normalized to filament length, which is the settled volume from the SVI test divided by the filament length. Typical results indicated that as filament content increases, it tends to approach a constant value (about 5 mL/mm in this case), which may be considered the expected increase in SV per increase in filament length. This approach could be a useful new perspective in the consideration of
filament content effects on bulking.