Simple and fast conceptual models to simulate water quality in rivers.

Befekadu Woldegiorgis, Ann van Griensven, Fernando Pereira, Willy Bauwens

Thursday 2 july 2015

11:45 - 12:00h at Asia (level 0)

Themes: (T) Hydro-environment, (ST) Impacts of pollutants on the water environment

Parallel session: 11G. Environment - Impact

Efficient and reliable method of planning is a key for curbing the ever increasing water quality problems thereby maintaining a healthy aquatic environment and computer modelling plays a vital role in this regard. Most river water quality models have large simulation times because they are based on hydrodynamic simulators. This limits their applicability for planning activities involving long-term statistical information. Therefore, fast and accurate models are crucially complementary to the detailed models. Traditional QUAL2E transformation formulation can be considered as an approach with less parameters than detailed models while taking the pollutant interactions into account. This paper presents a new quasi-analytical solutions-based river water quality modelling approach that integrates ODEs of QUAL2E pollutant transformation in to dynamic CSTR. We compared it with the Euler method implemented in the widely applied SWAT model and fourth order Rung-Kuta methods. It turned out that the numerical methods show serious instabilities because of overestimations of decay processes during large residence times as well as due to abrupt changes in the input concentrations. To overcome the instabilities we derived, based on simplifying assumptions, quasi-analytical solutions of first order differential equations resulting from the integration of the QUAL2E transformation formulations in to the CSTR-based mass balance equations. Finally, we tested the stability of the quasi-analytical method for real simulations and hypothetical extreme low flow scenarios. The new quasi-analytical solution gives unconditionally stable results. It also gives fairly comparable results with the reference RWQM while running 130,000 times faster than it. Hence, it is fair to conclude that this approach is preferable to the numerical methods for water quality simulations of river and conveyance systems involving large residence times. Furthermore, we demonstrated that the dynamic quasi-analytical CSTR solution of non-conservative pollutant modelling can be implemented using the classical linear reservoir approach after applying minor modifications to the latter. In the context of decision support system, we believe that the implementation of this approach combined with conceptual water quantity modelling methods makes it vitally complementary to detailed models.