...
Click here for the results of the series of simulation experiments and parameter analysis, completely summarized in an Excel workbook. Only parts of the results are given graphically below.
preliminary simulation experiments
- To investigate the appropriate boundary condition settings and the convergence level
Simulations with various h over b ratios
According to the results of the preliminary simulation experiments, the simulation are design in following ways:
- Mesh generation
- The mesh files for various h over b ratios are generated using the journal file
- Define the problem in FLUENT
- Click here for a report summary of all parameters of FLUENT model
- Formulations of the performance parameters
Wiki Markup Click !here^CFD tasks.ppt|width=32,height=32! for a summary of the derivation and formulation of the performance parameters
Take the values calculated from the 4th baffle, and plot each parameter as a function of h/b:
Number of baffles
Note that in the Excel workbook, the performance parameters were calculated for both the 4th and 5th baffles. There are minor differences in numerical values while the general trend stays the same. Particularly, a more pronouncing difference is observed for h/b=4. Thus simulations with more baffles and a sensitivity analysis of the number of baffles could be included in the future work.
K_baffle
As shown in the plots above (and the Excel workbook), the trend of K_baffle values is consistent with the hypothesis. It decreases rapidly for h/b = 2~5 and stabilizes around 3.2. An upward trend appear around the tail of the K_baffle vs h/b curve, which indicates the major loss in addition to minor losses is playing a more significant part in the total energy loss.
Pi_cellClick here for a summary of the calculated
parameters
Pi_cell was shown to increase continually as h/b ratio increases in preliminary experiments, which was also reported in Fall 2008 research. This trend is further confirmed with a large h/b ratio of 40. As reported in Fall 2008 research, Pi_cell value keeps increasing because energy is dissipated along the wall as well as around the 180 degree turning region.
Pi_cell/(h/b)
The plot of Pi_cell/(h/b) as a function of h/b is also consistent with the hypothesis. As a measure of active fraction of the total volume of the flocculator, it reaches a maximum of 95% for h/b=4, and keeps decreasing as h/b increases. However, the curved flattens out at the tail. This also suggests although an optimal value exists for h/b, it wouldn't be the most critical constraint for design, since even for an un-pratically tall flocculator of h/b=20, it still has an active volume fraction value around 70%.
Collision potential/(h/b)
The power function fit for collision potential (theta*epsilon^(1/3))/(h/b) was suggested by Dr Monroe Weber-Shirk. It shows a unexpectedly simple relationship with a power of about -0.5 and a coefficient of 0.33. Further investigation is needed to better interpret this relationship.
Conclusion
- The number of baffle used are not enough for the cases of h/b= 2~5 to converge.
- The formulation of the K_baffle and Pi_cell are valid and consistent. They captures some of the characteristics of flocculator performance.
- Optimal values of h/b=4 is observed in terms of Pi_cell/(h/b) value.
- Relationships between (theta*epsilon^(1/3))/(h/b) and h/b is unexpectedly simple and needs further investigation.
Future Work
- Simulations with more baffles and investigate the sensitivity to the number of baffles used, especially for h/b=4.
- More detailed comparison between symmetry boundary conditions and the wall boundary condition at the water-air interface.
- Simulations with different Reynolds numbers.
- Non-dimensionalize the results.
- Characterize the maximum energy dissipation rate to characterize floc break-up