Analyzing the Risk of Producing Non-compliant Water in a Water Treatment Plant Using Probabilistic Risk Analysis Methodologies

摘要

A variety of risk analysis methodologies are available to assess the risks experienced in different disciplines. However, determining the probability that a properly operated water treatment plant will produce water that is not compliant with a stated threshold presents a unique challenge because of both the complicated nature of water treatment and the natural variation of water quality and quantity. One methodology that could be used is the consequence frequency assessment (CFA), which is used in microbial risk assessment. In this research, a full-scale anthracite/sand filter unit was analyzed using the CFA, based on the influent and effluent turbidity. The risk output from the consequence frequency assessment showed some deficiencies including a possible overestimation of the risk of producing high turbidity water and an inability to incorporate more than one regulated parameter. The difficulties of using the consequence frequency assessment called for a different approach to perform the analysis. The approach used in this research combines water treatment plant modelling with probabilistic risk analysis. To compare the two methods, the same filtration unit that was analyzed using the consequence frequency assessment was then analyzed using the new methodology. As an example of results obtained, the consequence frequency assessment determined that the probability of producing effluent turbidity greater than 0.30 NTU was 1.0% while the probability of producing effluent turbidity greater than 0.10 NTU was 10%. The new methodology determined that the probability of producing effluent turbidity greater than 0.30 NTU was 0.6% while the probability of producing effluent turbidity greater than 0.10 NTU was 0.7%. Along with the numerical difference between the methodologies, the new method can deal with the problems experienced while using the consequence frequency assessment and incorporate multiple regulated parameters into the analysis. The demonstrated disparity between the results obtained with the two risk analysis methodologies shows the potentially significant role that an analysis methodology can play in the analysis results.