Sensitivity Analysis of Model Parameters for Water Quality Simulation
Nireshini Naidoo, Muthukrishnavellaisamy Kumarasamy
More details
Hide details
Civil Engineering Programme, School of Engineering, University of KwaZulu-Natal, Durban, South Africa
Submission date: 2016-01-06
Final revision date: 2016-02-21
Acceptance date: 2016-02-21
Publication date: 2016-07-22
Pol. J. Environ. Stud. 2016;25(4):1655–1660
The Hybrid Cells in Series model consists of a plug flow cell – first and second well-mixed cells in series that represent advective-dispersive transport pollutants in river reaches. The key model parameters are the cell residence times (α, T1, and T2), which require parameter optimization to calibrate the model and to identify the sensitivity of model parameters. The least square error optimization method was used to determine the model parameters. Tracer test data was adopted to evaluate the sensitivity of the hybrid model response to a unit impulse input of a conservative pollutant. Factor perturbation was applied to determine the parameter with the greatest influence on hybrid unit output. The nth hybrid unit output was generated using a convolution technique and compared to the observed output from available field data. Traditionally, the longitudinal dispersion coefficient (DL) of a particular waterway is calculated according to the physical flow and channel properties. These properties are hardly regular, and the investigation into the use of moments to determine DL helps to eliminate uncertainties for future hydraulic modelling ventures. Parameters α and T2 are more sensitive, needing accurate estimation to correctly reproduce concentration-time profile, and to simulate river water quality.