Design Case

A design case is being analysed Flocculation CFD simulation. The limitations are stipulated by considering the design in current Agua Clara plants. The height of the flocculator in the plant cannot be as low as the optimal ratio H/S = 3 because the depth in the flocculator has to match the sedimentation tank for construction of the plant. The width in the flocculator is fixed to allow people to fit in the baffle channel. CFD cases for higher H/S ratios show the greatest amount turbulent energy dissipation occurs after the turns in the flocculator and decreases as the water travels up or down the length of the baffle. Tall baffles allow the turbulent energy dissipation from the turn to decrease substantially before the next turn in the flocculator. Collision potential of flocs is dependent on the turbulent energy dissipation, so the turns in the flocculator induce many collisions of flocs and during the length of the tall baffles there are fewer collisions of flocs. A flocculator with high H/S has decreased collision potential between turns.

Simulation of obstacles in the flocculator is being done for one of the larger H/S ratios. H/S = 10 was used earlier in the semester, so this ratio is used for the obstructed flow for ease of comparison. The obstruction considered for the flocculator is a circle, a flat plate, and a half circle. The simulations are in 2D so these shapes represent a cylinder, half cylinder, and a bar. The first examination is the effect of the size of the shape on the flow in the flocculator.

Sketch generation:

The shape is put in the sketch the way the baffles are denoted in the Fluent sketcher. Additional procedure is required to generate the mesh appropriately with the shapes in the sketch. For the standard flocculator sketch, the sketch can be partitioned to control the size of the elements along the wall of the modeled baffle. The lines can be implemented around the shapes in the sketch to control the accuracy of the mesh around the shape. The circle has another outlined just outside of the cutout of the circle to ensure accuracy of y-plus on the surfaces of the circle. The plate is implemented by partitioning the baffle into three parts across its width. The center section is the width of the plate. The case with the half cylinder has not been meshed because of difficulty to draw boundaries around the half cylinder to allow the mesher to generate uniform quadrilaterals around the shape.

Boundary layer element sizing (click for larger picture)

Meshing:

The areas of the sketch are meshed by dividing the edges into divisions. The edges on the wall are biased so smaller elements are generated at the wall. There are more partitions for the in the mesh for the flocculator with the geometries. The mesh is biased toward the upper and lower surfaces of the plate for the case with the plate in the flocculator.

Plate Mesh, Circle Mesh, Boundary layer Mesh (click of larger picture)

Simulation

The shapes cause vortex shedding in the flocculator and the non-transient solution does not converge since the quantities are not at a constant position in the mesh. Transient analysis in Fluent analyses the properties in the simulation at multiple time intervals, so the variance of values at different times and repetitive behavior of the flow are observed. For example, transient simulation can be setup for 20 time steps at a time interval of .5 sec, giving an analysis of 10 sec of fluid flow. The properties for the CFD simulation of inlet velocity = 0.1 m/s and S = 0.1m contribute to establishing a simulation time step size.

S/velocity=(0.1 m)/(0.1 m/s)=1 sec,

This is the timescale for the simulation. The Fluent user manual states to use a fraction of the time scale for the time step size and a fraction of 10 of the timescale is a good time step size. The number of time steps gives the time elapsed in the flow. 200 time steps are used to give 20 sec worth of flow. The amount of iterations per time step is defaulted at 20. Iterations per time step between 60 and 150 were used for the multiple simulations with different positions of shapes.

Simulation of the Plate in Flocculator

Observations of the different positioning for the circle:

Layering the circle cutout in the sketch with the slightly larger circle kept the y-plus values at the surface of the circle below 5.

Diameter of the circles analysed: 0.05 m, 0.03m, 0.02m, and 0.01 m

Width of the plate analysed: 0.05 m (this simulation converged without transient analysis, possibly a result of the fixed shedding point on the plate and the plate is wide enough that the shedding flow from the edges do not interfere with one another - keeping the position of the shedding constant) .

Notes about the position of circle(s) in the simulation:

Positioning the circles before the first turn of the flocculator caused a periodic vortex shedding and variation of the turbulent energy dissipation with time in the flow around the turn that persisted and repeated in the 20 sec simulation.

Variation of Turbulent Kinetic Energy Dissipation with time (scale is too .005 m^2/ s^3)

Positioning the circle and the plate before the first turn decreased the energy dissipation zone after the first turn.