Liping Pang, Panan Nilprapa, Susan Lin, Aruni Premaratne, Phillip Abraham, Chris Nokes (Institute of Environmental Science and Research Ltd)
Annabelle Tham (Institute of Environmental Science and Research Ltd, University of Otago)
Adrian Cocker, Philip MacDonald, Richard Adams (Invercargill City Council)
Drinking-water contamination by pathogenic Cryptosporidium parvum and Cryptosporidium hominis poses a serious health risk. Insufficient oocyst removal in drinking-water has caused numerous cryptosporidiosis outbreaks worldwide. As Cryptosporidium oocysts are extremely resistant to traditional disinfection, filtration through porous media plays a very important role in the removal of Cryptosporidium oocysts during drinking-water treatment. To ensure safe drinking-water supplies, the filtration efficiencies of treatment systems need to be assessed to optimise filter operations and better plan disinfection steps.
Using pilot-scale experimental facilities, we assessed protozoan filtration efficiencies in 3 different rapid sand filters and 5 different point-of-use household filters commonly used in New Zealand. Glycoprotein-coated polystyrene microspheres were used as a Cryptosporidium surrogate. The modified microspheres mimic the size, density, shape, surface charge and surface macromolecules of Cryptosporidium oocysts, and have been satisfactorily validated alongside Cryptosporidium in New Zealand and overseas in laboratory and pilot-scale coagulation-filtration studies.
The pilot filtration plant was located at the Branxholme Water Treatment Plant (WTP), which treats water sourced from the Oreti River. The pilot plant’s feedwater came from the WTP’s clarifiers post-coagulation with polyaluminium chloride (PACL) or alum. The sand filters comprised anthracite, pumice or Macrolite engineered ceramic sand, and simulated a typical water treatment plant’s operational condition. The filter (15 cm diameter) contained 1 m of granular media (70 cm filter media: 20 cm silica sand: 10 cm pea gravel) and it had a 40 cm water column head. The surrogate’s log10 reduction values (LRVs) based on the peak concentrations were >3 in 100%, 70% and 41% of the trials with the ceramic sand, pumice sand and anthracite filters, respectively. The LRVs achieved in the ceramic sand filter trials (4.440.38) were significantly greater than those in the pumice sand (3.210.30) and anthracite (3.010.70) filter trials (P<0.00001).
The data describing the media filtration efficiencies could help WTP operators improve their practices for ensuring drinking-water safety. This could be achieved, for example, by using more effective filter media, allowing sufficient time for the water to run to waste after backwashing before the filter is brought back on-line and incorporating additional post-filtration treatments.
A full-scale filter test rig was custom-built to simulate typical household-use conditions (40 psi, 21.6 L water/day treated, intermittent operation). The data from 120 test runs (duplicate filters, 24 replicate runs per filter type) indicated that the surrogate particles’ LRVs were 3.93–4.54 in the 1 µm nominal activated carbon filters, 1.95–2.94 in the 2 µm nominal silver-impregnated activated carbon filters, and < 1.0 in the 1 µm nominal polypropylene, 1 µm nominal polyester and 1 µm absolute pleated-paper filters. To achieve an LRV>3, which is a requirement of domestic drinking water treatment units for protozoan reduction, 1 µm activated carbon filters are recommended. To satisfy protozoan removal requirements when using the other four filter types tested, additional treatment, for example, water boiling or ultraviolet disinfection, is necessary.