Abstract
Identification and investigation of problems of water treatment plants with slow sand filter overload was the main focus of this study. Initial studies in St. Lucia at Dennery treatment works in 1987, showed turbidity levels ranging from 400 NTU, resulting in frequent filter blocking and consequently to interruption in the water supply and loss of sand during cleaning operations. Phase 1 of this project in 1989 brought about the construction and evaluation of an upflow prefilter and the evaluation of the key operational criteria of filter performance for turbidity and faecal coliform removal. Initial overall average turbidity removal by the prefilter was 60% and 34% for faecal coliform removal in 1989 whilst the slow sand filters produced 68% reduction in turbidity and 66% reduction in faecal coliform counts. This led to further improvements in both prefiltration and slow sand filtration systems. The average raw water turbidity for the year was 47 NTU and at the out let of the slow sand filter was 6 NTU, which is above the WHO guideline value of 5 NTU. Dissatisfaction with these results, led to further intervention, phase 2 in late 1989. Phase 2 was the construction of flow control chambers for the slow sand filters. Unfortunately evaluation of phase 2 was conducted during a dry year in which the average turbidity for the whole year was 14 NTU more than 3 fold lower than the previous year. As a consequence the law of diminishing returns effected a lower performance than in 1989. By contrast faecal contamination was higher and consequently showed removal correspondingly greater in both prefilter (46%) and slow sand filters (78%). Because the turbidity levels were still above the WHO guideline values (5 NTU), a third series of interventions (phase 3 & 4) were planned and executed in 1990- 91. These involved the construction of the second prefilter to reduce the flow rate and to replace the lost sand and improve filtration, thus improve performance. Unfortunately rainfall in 1990-91 was even lower, about (43%) below 1987. In the first 5 months in 1991, the average turbidity was also correspondingly lower (6.92 NTU). Again the law of diminishing returns did not show the prefilters at optimum efficiency (28 & 31%) with an overall average mean of 29.8 % to its advantage. The slow sand filters on the other hand achieved individually for turbidity removal 19 and 16% efficiency, with an overall mean efficiency of 17.7 %., although the average turbidity in the final water was 4.0 NTU which at last was within the WHO guideline values, the performance efficiency was well below the 90% target expected from as slow sand filter. In the same period the average faecal coliform contamination in the raw water was 78.8 faecal coliform /100 ml, about 2 fold below that of the previous year 1990. In spite of this low faecal coliform loading the prefilter removed on average 48.8% and the slow sand filters 38.7%. In 1992, the filters did much better in bacteriological removal. Both the prefilters and slow sand filters achieved overall performances of 55 and 76% respectively. It was believed that the slow sand filters performed poorly in the initial stages of the study because the water was considered to be 'thin'. The term thin water is defined as oligotrophic (nutrient-poor), which supports very little living organisms, hence the poor colonization of biological predators to assist in the purification process.