Introduction:
Ansys Fluent has been used to obtain flow solutions for idealized AguaClara flocculators and the question has always been asked by Professor Monroe Weber-Shirk, can we check if the solution is correct? PIV experimentation by Julia Schoen with a scale model of the flocculator and headloss and flowrate measurements in an AguaClara facility in Agalteca, Honduras by Sarah Long have shown that the pressure drop coefficient in flocculator simulations in Fluent, using the available turbulence models, are too high. The formulas governing turbulent kinetic energy dissipation and headloss are:
If the pressure drop coefficient is wrong in the simulation there is a good possibility the flocculation efficiencies anaylsed from Fluent as erroneous as well. The questions in Fluent introduced ideas about why Fluent had an erroneous solution for the flocculator. First, features in the flocculator were broken down into simple fluid flow features; the flocculator has a contraction and an expansion at the recirculation zone after the turn, and severe flow curvature around a baffle. Next, Fluent can be simulating one of these features of the flow incorrectly. This report is a study of different case studies comparing Fluent simulation with analytical and experimental information to test the correctness of Fluent's solution for different areas of the flow in the flocculator. A system of benchmark simulations and comparisons were conducted to assess the correctness Fluent's predictions for different features of the flow in the flocculator.
Methodology
Being developed....
Cases
The first hypothesis is Fluent does not calculate streamline curvature into a contraction correctly. The discharge coefficient is derived from Bernoulli analysis on an orifice. The discharge coefficient is used as a flowrate measurement technique in pipes. The exact width of the vena contracta following an orifice is known in a fluid flow so the width of the orifice is used and the discharge coefficient contains the error of using the width of the orifice in the Bernoulli analysis. Fluid Mechanics by White has discharge coefficients for an orifice and an orifice matching the orifice, pipe diameter ratio on the graph in the textbook was simulated in Fluent and compared to textbook results. The orifice of β=0.6 simulated in Fluent at different Reynolds number yielded the following results:
This positive result from Fluent allowed for new investigation of the Fluent simulation. The next conjecture is Fluent does not fluid flow in an expansion correctly. The pressure drop coefficient in pipe discharging into a larger pipe can be estimated analytically from the geometry using control volume analysis. Fluent analysis can be compared to the control volume analysis using the inlet pressure and velocity, and the outlet pressure and velocity. The results yielded in this comparison are:
The next plan is to complete the analysis of the orifice. The discharge coefficient does not display any data about pressure recovery in the orifice. If Fluent truly calculates fluid flow in contractions and expansions correctly, then the Fluent solution should match orifice pressure recovery experiments. The paper "Numerical Investigation of Turbulent Flow through a Circular Orifice" was found and its results are a comparison of a computational fluid dynamics code with orifice experimental results from the dissertation "A study of 3-Dimensional flow through orifice meters". "Numerical Investigation of Turbulent Flow through a Circular Orifice" analysed the correctness of the k-epsilon and Reynolds stress turbulence models by comparing dimensionless plots of mean axial centerline velocity and wall static pressure.
The orifice was simulated using the k-epsilon realizable and Reynolds stress turbulence models in Fluent. The results of wall static pressure and axial centerline velocity match the shapes of the orifice experimental measurements. The k-epsilon realizable and Reynolds stress models calculate the correct pressure loss far downstream of the orifice. At the orifice, the k-epsilon realizable model calculates a greater pressure difference than the Reynolds stress model and both models calculate greater pressure differences at the orifice than exist in the experiment. In the pressure recovery zone immediately following the orifice the k-epsilon realizable and Reynolds stress models fit the trend of the experiments pressure recovery, however the Reynolds stress and k-epsilon realizable models still different in that k-epsilon realizable reports a greater pressure difference than the Reynolds stress model. This subtle difference in the orifice could relate to characteristic of the k-epsilon realizable model reporting a higher flocculator pressure drop coefficient than the Reynolds stress model.
Based on the plots of wall static pressure and centerline axial velocity for the orifice below, both turbulence models overestimate the velocity of the fluid after the orifice and have a faster decline of velocity in the fluid than the experiment. The k-epsilon realizable model overestimates the velocity after the orifice more and predicts a steeper decline in fluid velocity. Both turbulence models over predict the velocity at the exit of the orifice. The Reynolds stress model shows some velocity recovery before x/R = 10, which match a trend in the experimental data, however the velocities in this trend do not match.
The pressure recovery characteristics of the orifice simulated with Fluent turbulence models support the results of computational fluid dynamic simulations matching orifice experimental measurements presented in "Numerical Investigation of Turbulent Flow through a Circular Orifice" and "A study of 3-Dimensional flow through orifice meters". Ansys Fluent turbulence models can predict fluid flow in contractions and expansions correctly. Severe fluid flow curvature around the baffle in the flocculator may exceed the limits of the turbulence model's predictions.
The difference between the fluid flow features in the flocculator and an orifice is the fluid curvature around the baffle. The flows have fluid contraction and expansion which Fluent can predict correctly. The conditions at the vena contracta of the orifice and the flocculator have different conditions of velocity, turbulence kinetic energy, and turbulent kinetic energy dissipation predicted. The correctness of the Fluent predictions for the orifice and rapid expansion case say Fluent will correctly predict the fluid flow from the vena contracta and the fluid contraction into the vena contracta correctly. The error in flocculator simulation may arise from incorrectness of fluid flow prediction into the vena contracta of the flocculator resulting from incorrectness of fluid flow predictions around the baffle.
Conclusions
Being formulated....