Press Room - Modeling of reservoir flows using STAR-CD

	Modeling of reservoir flows using STAR-CD
	Reservoirs constitute an essential part of the water supply capacity of numerous water companies. There are two issues in the management of these resources which water companies have to address: The avoidance of "dead" regions of stagnant water is important in preventing the growth of micro-organisms If a pollutant reaches the reservoir, risk of pollution of water supply has to be minimised by ensuring that transit time through the reservoir is sufficient for the pollutant to be detected and water from the reservoir stopped from reaching the main supply line. Large baffles placed within the reservoir are often used as a solution, but the implementation of this method is somewhat haphazard and based mainly on the design engineer's experience and common sense. Moreover, these baffles are not aesthetically pleasing and would hinder the use of the reservoir for recreational purposes. Computational Fluid Dynamics can help the design engineers by providing them with reservoir flow patterns that enable them to place the baffles more strategically, or by suggesting other solutions to resolve the issues. Arup modelled a water supply reservoir where CAD data was available for the reservoir bed topography, as well as a series of experimental measurements including drogue tracking and dye tracing, and concentration of in-going and out-going pollutants during a real-life event. Hence, we were able to first calibrate the CFD model using the experimental measurements, and then validate it against the real-life event.The STAR-CD model consisted of approximately 275,000 fluid cells. Firstly, flow patterns under steady-state conditions were predicted that matched the results from the drogue tracking experiments.

	Using these results, a transient simulation was then carried out. For this simulation, the dye tracing data were used, i.e. a low concentration of pollutant was introduced into the inlet water over a period of approximately 35 minutes. The concentration of pollutant within the reservoir, in particular at the outlet, was then predicted over a period totalling 20 hours. Only the pollutant scalar variable was solved for during the transient analysis. No information was available on bed roughness for this reservoir. Because the cell closest to the bed was smaller in size than the estimated roughness height, the bed roughness could not be applied via a user subroutine. Instead, bed roughness was represented by creating a porous medium layer on the bed, and matching the predicted flow patterns and pollutant dispersion rate with the measured ones. Once the porous medium parameters were adjusted, the CFD results matched the measurements quite closely, both for the maximum concentration of pollutant at the outlet (predicted 0.076mg/l, observed 0.079mg/l) and the timing of the peak concentration (predicted 6.5 hours after release of pollutant, measured 7 hours). Finally, to fully validate the calibrated model, a real-life event was modelled where both concentration and flow rate data were available. The results match quite well. The predicted time of peak was 6hours and 20 minutes later than measured. The predicted peak concentration was 167ng/l, the measured value was 163ng/l. The delay in the time to peak can be explained by the fact the we assumed a constant flow rate through the reservoir, and that this value was smaller than the flow rate actually going through the reservoir in the first part of the pollution event. Overall, the predicted data provide a good fit for the measured data. This validated CFD model could now be used to devise means of increasing the pollutant transit time across the reservoir. Various options for increasing transit time in fact became self-evident in the course of completing this project. From the various CFD runs carried out, it seems that increasing the bed roughness in parts of the reservoir would suffice to increase the transit time significantly. For further information, please contact isabelle.lavedrine@arup.com