Francois Daudelin
M.ASc. Student
Civil Engineering
Queen's University
2018 - 2020
Supervisor(s): Pascale Champagne, Warren Mabee
Research Project: Transient Heat Flux Models for Uncertainty based Waste Stabilization Pond Design
The design of waste stabilization ponds (WSP) can make use of four different types of models: Rules of thumb, Regression equations, First order kinetic models and Mechanistic models. Out of the four, only the last two are able to adapt to local climatic conditions and therefore produce the smallest variability in pond sizing estimates. The first order kinetic model can be used with a probabilistic approach to design which quantifies uncertainties in both the inputs and the model’s estimator to offer an associated uncertainty with the model’s output. A review of WSP design methods concluded that probabilistic designs offer a more accurate method of design by capturing local climate variability and should be further developed in the future. The only probabilistic WSP design method which appears in literature and makes use of first order kinetic models is the use of Monte Carlo Simulations. This approach requires a probability density function for the mean daily pond temperature (MDPT) which was identified as a critical variable in these types of simulations. In practice this variable can be estimated by using an empirical equation to get the mean monthly pond temperature and associating to it an uncertainty to capture daily variations (ex:+-10%) which produces a uniform distribution for the MDPT. Alternatively, an appropriate transient heat flux model could be used to estimate a time series of daily mean pond temperatures from which statistical properties could be extracted. This second method would not only provide an a more accurate estimate for MDPTs but would also provide information regarding their autocorrelation. The autocorrelation between MDPTs is of importance in these simulations because retention times of these systems are longer than a day, meaning they require multiple MDPTs to be sampled per Monte Carlo iteration.