This work is carried out in collaboration with Dunea and TU Delft. The aim of this study was to optimize the ripening period of slow sand filters and to identify the indicators of ripening. Combination of full scale and column scale slow sandfilters were used to achieve the goals. Two full scale and eight column scaleslow sand filters were used in the study.

Ripening period is required for the formation of biological community (schmutzdecke layer) over the sand layer and within the top layers of the sand bed during which filter performance is sub-optimal. The ripeningperiod depends on factors such as influent water quality (nutrient loading), temperature and filtration rate. In order to optimise the ripening period of slowsand filter, two possible approaches were investigated. First, how can we retain the maximum biological activity within the filter bed at the time of scraping (optimise scraping) and second how to accelerate the growth of microorganism when a filter is put into operation by changing operational parameters.

In order to ensure maximum biological activity is retained within the filters at the time of scraping, biomass concentration in different layers of sand bed which is responsible for head loss was quantified followed by determining the inactivation potential and biological activity in different layers of sand bed.The biomass concentration was determined by measuring Adenosine Triphosphate(ATP) content and biological activity by using combination of total cell count and ATP of the sand samples in both column and full scale SSF. Inactivation potential of different layers of sand bed was determined by carrying of spiking experiments after removal of different layer of sand bed. Spiking experiments were done only in columns SSF.  Spatial Distribution of biomass on the filter bed was also investigated.  

In order to accelerate the growth of microorganism’s, three possible solutions were investigated in column SSF.  First was the use of additional nutrients, second was to increase the filtration rate and third was the use of microbial inoculum (schmutzdecke) from a matured filter. The effect of different operational parameters on the efficacy of column SSF was determined by measuring influent and effluent parameters such as particle counts, turbidity, dissolved organic carbon and total nitrogen. Along with this spiking experiment of E.Coli WR1 and MS2 bacteriophage were carried out during stages of filter operation. Physical, chemical & microbialparameters that were used to assess the efficacy of SSF were correlated to each other and most suitable indicators of ripening were identified.

In total there were 8 columns, running in duplicates with six of them running at filtration rate of 0.1 m/hrand two at 0.5m/hr. Two columns with 0.1 m/hr were used a reference for comparison. Two columns running at 0.1 m/hr were seeded with microbial inoculum from one of the full scale filters and other two at 0.1 m/hr were seeded with additional nutrients.

The Biomass concentration decreased with depth in both the full scale and column slow sand filters. More than 80% of biomassaccumulation takes place in schmutzdecke and top 2 cm of sand bed.  Position of the inlet valve source affects the spatial distribution of biomass on the filter surface due to lateral gradients and leads to uneven biomass growth. Cell Count follows the similar patters as biomass distribution. Biological activity was present throughout the 10 cm of sand bed.  

Decimal Elimination Capacity of column SSF decreased after the removal of subsequent sand layers, with most significant reduction in DEC was observed after the removal of schmutzdecke (>1 log). Columnwith added microbial inoculum were able to mimic the full scale filters. After removing of schmutzdecke and top 2 cm of sand bed where most of the biomassaccumulation takes place, columns were still able to achieve more than 3 logremovals for bacteria and 1 log for virus. This is higher than the values required by current full scale SSF’s in their operation.  Effluent turbidity and particle counts wereless than 0.1 NTU and 200/ml even after the removal of schmutzdecke in columns with inoculum.

Combining the results of the biomassdistribution and spiking experiments carried out in the columns with microbialinoculum, it can be concluded that ripening period of the SSF will bedrastically reduced if the scraping of only 4 cm of sand bed  takes place including schmutzdecke.

To reduce the start up time of a new filter, addition of microbial inoculum (schmutzdecke from a matured filter) is the better solution in comparison to addition of nutrients or increasing the filtration rate. Although the purpose of adding inoculum or nutrients or increasing the filtration rate was same: that is to increase the biological activity in the sand bed. Columns with inoculum reached more than 2.5 log removals in first 30 days of operation as compared to others which have less than 1.85 log  and took only 24and 27 days to reach median levels of turbidity of 0.1 NTU and particle count of  less than 200/ml.

Reductions of bacteria, viruses, turbidity and particle counts increase substantially with time as filters ripens. No such pattern was observed in the DOC and TN removal, they were more a function of the influent water quality and independent of the ripening period. Correlation between different water quality parameters resulted Particle count and turbidity can be used as indicators of ripening.  Although particle count was a better surrogate than turbidity as an indicator of ripening, using them together would provide a better insight as the correlation between them increases as the filter ripens. DOC and TN cannot be used as indicators of ripening.