This is a simulation of a 4 MG Cylindrical tank with a tangential inlet pipe and a nearby, slightly larger outlet pipe. The tank is run in a simultaneous inflow/outflow mode. The simulation was done primarily as a validation of the CFD technology for water storage tanks. This case was also used to study grid resolution and choice of time step for modeling tanks in general. Sampling data comes from a June 1994 report of an April 1994 sampling study (Technical Memorandum No. 3, Light and Water Department, City of Azusa).

Comparing the mixing time statistics for this case with other cylindrical tanks shows that this tank mixes more slowly than the others, even though its fill rate is nearly twice the other cases. All the other cylindrical tanks we studied have vertical inlets.

The tangential inlet in the Ed Heck reservoir creates a flow that results in a relatively quiet zone in the center of the tank. The water age in that portion of the tank ranges from 20 to 24 hours at a mean flow rate of 6000 gallons/min (~8.6 MGD).

Below we present a summary of the results for this reservoir.

Geometry and Mesh for Tank (1100x850 GIF 60K).

This image shows the tank geometry and mesh. Inlet and outlet area are shown with additional detail. This mesh has ~50,000 cells.

Fluoride Concentration (1100x850 GIF 25K).

This image shows the distribution of a fluoride tracer fed into the tank for ~24 hours. This is the distribution after ~2.5 hours. The horizontal display plane is near the tank's floor. The vertical display planes are at 12:00, 3:00, 6:00, and 9:00 O'clock (inlet/outlet pipes are at ~6:00).

Secondary Velocity Field (1100x850 GIF 41K).

This image shows the velocity field in a vertical cut plane (90 degrees from the inlet). These circulation patterns transport the fluoride tracer from the tank floor and outer wall to the Tank's surface and inward toward the tank's center.

Secondary Velocity Field (1100x850 GIF 50K).

This image shows the velocity field in a vertical cut plane (270 degrees from the inlet). These circulation patterns transport the fluoride tracer from the tank floor and outer wall to the Tank's surface and inward toward the tank's center.

Test of Different Simulation Methods (850x710 GIF 12K).

This image shows the fluoride tracer history at the tank outlet for several different simulations. Simulating the flow without a turbulence model does not reporduce the outlet fluoride history. Similarly, changing the time step (before the flow field has fully developed) does not reproduce the outlet fluoride history. Reducing the number of iterations per time step does not compromise the simulation.

Validation Study (Fluoride Concentration History over 24 hours) (850x710 GIF 12K).

This image shows the fluoride tracer history in the tank over a 24 hours period. The tank's mean concentration is shown for reference (black). Field data from a sampling study in April 1994 is shown in red. The CFD results are shown in Blue.

Other tanks showed that the outlet concentration tracks the mean concentration to within 10%. That is not the case here. The CFD results reproduce the outlet history quite well.

Mixing Statistics (850x710 GIF 13K).

The mixing statistics show that this tank takes ~10.3 hours for the standard deviation to reach 10% of the mean tracer concentration. This mixing time is very comparable to the ~10.2 hours needed to fill the tank of this size at 6000 gallon/minute. As comparison, cylindrical tanks with vertical inlets mix to this level of uniformity 3 to 6 times faster than their respective fill times.

[ SSESCO Website ] [ CFD Water Tank Home Page ]

Copyright (C) 1998 by Supercomputer Systems Engineering and Services Company.