An economic evaluation of water treatment costs in the Umgeni catchment area.

Abstract

This study has two objectives: first, to identify the main contaminants responsible for high water treatment costs in the Umgeni catchment area, and second, to predict water treatment costs from observed levels of contaminants. Reliable information about the origin of high water treatment costs is required to inform both policy and planning decisions. Partial adjustment models of water treatment costs are estimated using ordinary least squares regression and principal component analysis. First a model is estimated for the DV Harris treatment plant, which draws water from Midmar Dam. This model highlights important policy issues and explains 61 per cent of the variation in chemical treatment costs. Environmental contaminants have a marked impact on real water treatment costs at the DV Harris plant. Water treatment costs increase when levels of alkalinity, sodium and turbidity fall. Conversely, real costs rise with higher levels of dissolved oxygen and water stability. Paradoxically, clean water - typical of Midmar Dam is expensive to treat. Water treatment costs also rise when concentrations of the algae, Chiorella, decline. Second, a model is estimated for the Durban Heights treatment plant, which draws water from Nagle and Inanda Dams. This model explains 68 per cent of the variation in chemical treatment costs. Biological contaminants have a marked impact on real water treatment costs at the Durban Heights plant. Again, water treatment costs increase when levels of, Chiorella fall. Apparently the level of Chiorella varies inversely with the level of other, more expensive, contaminants at both treatment plants. Conversely, real costs rise with higher levels of total kjeldahl nitrogen, temperature, Anabaena and Microcystis. Water treatment costs also rise when turbidity and concentrations of silica, suspended solids and iron increase. The model predicts actual water treatment costs well (except during occasional peak cost periods) and provides a useful tool for scenario testing. For example, a simulation exercise in which turbidity levels were held constant at 6 NTU (nephelometric turbidity units) indicated an annual saving of R54 531 in water treatment costs.