The slow sand filtration alternative

Water treatment for small systems

By Robert A. LeCraw, P. Eng., RAL Engineering Ltd.

Filter under construction.

In the late 1980s and the early 1990s there was a growing emphasis on appropriate technology for small water treatment systems. Entire workshops and seminars were dedicated to advancing the knowledge that small systems had to be simple and affordable to operate. This lesson was learned the hard way in countless sad cases of complicated mechanical and chemical systems which no local staff could maintain and operate efficiently.

Unfortunately, it appears today the principle of appropriate technology has become a casualty of the headlong rush to high technology. However, more and more authorities are asking who is going to pay for these plants with high annual operating costs and when their obsolescence sets in, a period now measured in a few years, rather than a few decades. This inevitably leads to a search for a simpler, more cost-effective, and more durable solution to the small communities' servicing needs. Maybe the search should start with our lessons of the past and improve on the systems to make them more adaptable to today's and tomorrow's water quality needs but equally durable and simple to operate.

Conventional slow sand filtration

The traditional slow sand filter is basically a box full of sand. Figure 1.1 shows a cross section of a 'conventional' filter. Water enters the filter compartment above the media and flows down through the sand and, in time, will form a thin biological layer called a schmutzdecke.

The combination of physical straining and biological treatment will effectively remove turbidity, bacteria and will remove in excess of four log of Giardia cysts and Cryptosporidium oocysts. As the bed plugs up, the water level will rise over the sand. As it approaches an established upper level, usually the overflow, it must then be partially drained and the schmutzdecke scrapped off and discarded or stored for cleaning and recycling. Filter 'runs' of one to several months are typical. Over time the sand bed depth is reduced from the scrapings and new or recycled sand must be refilled into the box.

The major advantages of slow sand filters are:

The major disadvantages of conventional slow sand filters are:

Several conventional slow sand filter plants exist in Ontario and British Columbia. A small plant in Serpent River, Ontario, has been operating since 1987 with results as indicated in Figures 2.1 and 2.2.

The filters provide excellent turbidity removal and provide the community with a safe and low maintenance treatment system. A pilot study was required and determined that the filters needed to be larger than initially expected. Finally the plant does not always remove colour to below the Province of Ontario aesthetic guidelines despite the fact that the Serpent River raw water quality is relatively good compared to most surface water sources.

Improving the slow sand process

In summary, the shortcomings of the conventional slow sand filtration process are:

A number of enhancements have been studied to improve the performance of slow sand filters both in their ability to remove naturally occurring organics (NOM) and in extending their filter run times and/or to simplify the cleaning process.

These include:

The author reviewed these options and chose to pursue the development of roughing filters together with GAC contactors and, eventually, added pre-ozonation as an enhancement for highly coloured water. Chemical pre-treatment can produce good results and has been implemented in British Columbia but it did not seem to meet the objective of reducing the complexity involved with operating a plant with chemicals. The fabric option is fascinating and may have some application but to date a practical method of removing the fabric and cleaning has seemingly not yet been found.

Pilot study

To test the proposed enhancements, a pilot study was carried out in 1993 at Sturgeon Lake, Ontario, approximately 120 km northeast of Toronto.

The raw water source is not suitable for a conventional slow sand process due to high algae, moderate colour and moderate turbidity. The results of the pilot study with respect to turbidity and colour are presented in Figures 4-1 and 4-2.

The significant result in this study is the performance of the roughing filter. Despite the fluctuations in the raw water turbidity from 1.5 to 10 NTU, the roughing filter effluent remained consistently below 0.8 NTU. Also colour was removed to below 5 NTU over the four month period of the study at a relatively short empty-bed contact time of 15 minutes. Finally, the filtration rate of the slow sand was 0.3 m/hr., more than double the rate used at Serpent River, even with the much poorer raw water quality.

The real world

The Multi-Stage process has been developed to incorporate these enhancements. When the raw water is high in organics, pre-ozonation is added to oxidize the long chain molecules making them readily removable in the biological Multi-Stage filters.

The filter configuration is illustrated in Figure 5-1.

The plant consists of:

Stage 1 Roughing Filter

Stage 2 Slow Sand Filter

Stage 3 Granular Activated Carbon Contactor

The roughing filter is cleaned once every one to six months by down-flushing with the water over the entire filter bed. Occasionally a high rate up-wash is implemented for a more thorough cleaning. The slow sand filter is cleaned using a hydraulic wet harrow method. Water is directed over the filter bed to a collection trough while the surface is sprayed with a high-pressure hose to dislodge the accumulated schmutzdecke. This is required approximately once every two to twelve months depending on raw water quality. In over five years of operation the final GAC contactor has not required cleaning.

The multi-stage filter process has been installed in five Northern Ontario communities. A summary of the existing installations is presented in Table 5.1.

Plants under construction in the summer of 2000 are: