As many regions seek to supplement traditional water sources with reclaimed water, an increasing number of risk assessments are conducted for these types of applications. The most comprehensive approach is to conduct a quantitative microbial risk assessment (QMRA) combining empirical and literature data, point value estimates, and probability distribution functions (PDFs) to estimate the final risk for human health from a treatment train in quantitative terms. The variability and uncertainty of reuse systems can be more adequately assessed by probabilistic methods instead of deterministic, point value estimates. This review summarizes common assumptions in PDF selection for source water and treatment steps and dose-response models for risk assessments applied to potable and non-potable reuse scenarios. The review revealed that source water pathogen concentrations were mainly modeled using PDFs, while log reduction values (LRVs) were often derived as point estimates to describe removal efficacy of individual treatment steps. When enough point value LRVs are known, a triangular distribution is recommended to retain the stochastic characteristics of the variable being modeled. Treatments steps with the least amount of experimental data included biological activated carbon, membrane bioreactors, among others. To circumvent such lack of experimental data, an open-source, anonymized database of concentrations and LRVs could be made available for future assessments. Numerous studies mentioned that testing multiple dose-response models can help determine how the dose-response choice affects final risk. Although sensitivity analyses to determine how variables in the assessment influence final risk were performed in most studies, how PDF selection affects the final risk was not consistently evaluated. Such a discussion could help to establish more informative and comprehensive risk assessment models in future studies as the water reuse field continues to grow.
Zhiteneva et al. 2020
Neuer Zeitschriftenbeitrag von Zhiteneva et al. 2020