Unknown macro: {composition-setup}

cloak.toggle.exclusive=false

Unknown macro: {toggle-cloak}
Abstract

Unknown macro: {cloak}

The Lab Flocculator research team has an important focus on investigating the relationship between fluid shear (G) and its effect on overall flocculator mixing (Gtheta). Floc formation and breakup rates depend greatly on the Gtheta parameter, which is intrinsically related to the fluid shear present inside a reactor. The semester research goals of the Lab Flocculator research team was to measure the relationship between G and Gtheta by performing experiments using a coiled-tube flocculator experimental apparatus. The team decided to first perform experiments to determine the appropriate alum dose to achieve optimal floc formation for the experimental apparatus. Experimental data showed that a 25 mg/L concentration of alum produced the best results for a range of influent turbidity between 50-150 NTU and a humic acid concentration of 35 mg/L with minimal marginal improvements thereafter; thus, subsequent experiments were conducted with a constant 25 mg/L alum dose. Throughout the semester, the team was faced with numerous obstacles originating from the physical apparatus and the various software used by the team. Sedimentation of clay inside feed tubing and the flocculator inhibited various experimental controls including influent turbidity, flow rates, and also data collection. Measures were taken to overcome this issue, but sedimentation should ideally only occur in the settling column during the appropriate operational state. The flocculator team has overcome a large learning curve of needing to understand the experimental setup, Process Controller, and the MathCAD data processor tools. Much data was collected of different combinations of important variables such as flow rate, influent turbidity, and flocculator length. Currently, only a qualitative analysis of the data has been performed and the team is satisfied with the data set obtained. During coming semester, the team will focus on developing an analytical model to better analyze the data as well as to present it in intelligible figures and charts in preparation for publication.

Unknown macro: {cloak}

Unknown macro: {toggle-cloak}
Introduction

Unknown macro: {cloak}

The goal of finding a relationship between G and G¿ is not simple; it requires many experiments and a lot of data analyses. It is an important relationship to determine because it is hypothesized that with the increase in the length of the flocculator, the max G required to continue to build flocs decreases. Keeping the flocculator at a constant G throughout the tube may be compromising the floc strength and growth, as the flocs will build throughout the flocculator and then break up from the shear forces at the end of the flocculator. Floc growth depends on G, alum concentration, particle concentration, and the residence time of the flocculator. As particles floc together and the floc become larger, the strength of the floc becomes weaker with volume, so it is more prone to break up at the end of the flocculator. By reducing the shear forces on the flocs at the end of the flocculator, the flocs will remain intact, keeping the effluent turbidity low.
The effectiveness of alum dosage, flocculator residence time, and G is measured by the size of flocs after flocculation. Since the equipment needed to observe and measure the sizes of flocs formed after each experiment does not exist, the rate at which flocs settle under quiescent conditions must be analyzed along with the final settled turbidity. Since the experiments were run under identical conditions, the mechanisms creating the flocs are identical from run to run; therefore, the composition of the flocs (their shape and density) should be relatively uniform. This fact allows us to directly relate the settling rate of the flocs to the size of the flocs, while neglecting factors such as electrostatic interactions and porosity (which would affect drag coefficients and densities). This principle is the basis of the laboratory flocculator research.

Unknown macro: {cloak}

Unknown macro: {toggle-cloak}
Methods

Unknown macro: {cloak}

The flocculator setup was used to run the set of experiments. G was incremented from 40 sec-1 to 100 sec-1 by an interval of 5 sec-1 at each turbidity of 50, 75, 100, and 125 NTU. Using this set-up, our entire experiment was automated by Process Controller and ran continuously for up to two days per entire run. This experiment was run at three different flocculator tube lengths of 10 feet, 25 feet, and 100 feet. All three various flocculator tube lengths stayed wrapped around the hollow plastic structure in order to keep the transition between experiments easy.

Unknown macro: {cloak}

Unknown macro: {toggle-cloak}
Results

Unknown macro: {cloak}

The team's goal was to investigate the relationship between the average shear in the flocculator to the effectiveness of flocculation. Flocculation effectiveness can be qualified by observing the settling rate of the flocs during quiescent clarification and also the final turbidity after a period of time corresponding to the settling velocity and volume of the settling column. In conventional water treatment design, sedimentation tanks are designed to facilitate the settling of flocs by designing to the critical settling velocity. In the experiments performed this semester, a constant 600 second settling time was set for all runs because it was observed that it was a long enough period to observe final settling turbidities. This parameter should be adjusted according to the flow rate and the initial influent turbidity. The critical settling velocity can be modeled with first principles:

Unable to find DVI conversion log file.

where,

Unable to find DVI conversion log file.

Where Fg is a force of gravity, Fb is the buoyancy force, and Fd is the force due to fluid drag. After a brief period, the particle velocity will reach steady state (dV/dt = 0) and the critical settling velocity can be given as:

Unable to find DVI conversion log file.
If we also assume that all particles are nearly spherical and move slowly enough that stokes flow (Re=O(0.1)) can describe the motion of the particles, the critical settling velocity can be further simplified to the following:
  Unable to find DVI conversion log file.

There are, however, shortcomings to the derivation of this equation: For instance, the assumption that all particles are of spherical shape and therefore are affected by drag in the same manner as a sphere is far from reality. Particles are likely to have an average size distribution with a non-zero variance that are affected by mixing. Also, the assumption that all flocs will have the same density is also not very accurate. Finding a governing equation which explicitly defines the relationships between certain parameters to the settling curve would not only help us quantify the relationship between flocculation effectiveness, but it may also give us insights about how parameters (such as density, size, and shape) effect settling.

Figure 1: Effluent Settling for various G values at Influent Turbidity of 50 NTU and Flocculator Length of 50 feet.

Figure 2: Effluent Turbidity during Settle State - Influent turbidity of 50 NTU, flocculator length of 50 ft, 600 sec settle time

Figure 3: Effluent Turbidity during Settle State - Influent turbidity of 50 NTU, flocculator length of 50 ft, 600 sec

Figure 4: Effluent Turbidity during Settle State - 50-125 NTU, 25 feet

Figure 5: Effluent Turbidity during Settle State - 50-125 NTU, 25 ft

Figure 6: Effluent Turbidity during Settle State at 50 NTU, 10 feet

Figure 7: Effluent Turbidity during Settle State at 50 NTU, 10 feet

Figure 8: G¿ vs. flowrate - 50 NTU, 10 ft

Figure 9: Effluent Turbidity in Settle State - 50-125 NTU, 10 ft

Figure 10: Effluent Turbidity during Settle State - 25-125 NTU, 10 ft

Figure 11: G¿ vs flow-rate for various flocculator lengths

After the first several runs at 50 NTU, it was obvious that not enough time was given to allow adequate sedimentation. Figure 1 shows the settling curve of the first several runs. One can see that if more time was given to settling, the curves should have reached lower turbidities. The settling time was then extended to 10 minutes to give plenty of time for particles to settle out and for the settling column to reach an asymptotic steady-state minimum.

Figure 2 shows the settling curve for an experiment with the same parameters, but with a 10 minute settling state instead of a 5 minute one. The curves reach a point where the downward slope of the settling curve becomes nearly flat along the time axis. If given enough time, the colloids will all come out of solution; since they are not soluble, gravity will eventually pull them out of suspension. After 600 seconds, all of the data sets show that most of the colloids have settled out and a plateau in the setting curve is reached.

Figure 3 is a different viewing perspective of the same experiment plotted in Figure 2. Figure 3 exhibits two of the biggest challenges the team faced when analyzing the data. The most troublesome part of the results is that while the influent turbidity for each of the 12 flow rate iterations was set at the same value, the initial effluent turbidities measured at the start of each settle state were different. Since the initial effluent turbidities are different, it is difficult to compare the results of the different flow rate runs. Figure 3 shows that the final settled turbidity reached a value roughly around 5 NTU for all of the runs except for three outliers (which will be discussed later). It appears from this experiment that flow rate does not affect the final effluent turbidity, but that the initial effluent turbidity is affected by flow rate. The data seems to indicate that some of the clay that was introduced into the influent stream does not make it out of the system in the effluent stream. Sedimentation is an unavoidable issue in large scale flocculators where local velocities can reach very low values; but in the tube flocculator, sedimentation inside the tubes can introduce unwanted obstacles that can change the behavior of the flow. However, while observing the experiment during operation, it was not clear where the sedimentation was occurring because there was not a significant amount of flocs settling out in the clear tubes that make up the flocculator. Even at the lowest flow rate (1.45 mL/sec), only very little sedimentation was observed inside the flocculator; therefore, the sedimentation must have been occurring in the tube connecting the clay stock to the influent stream.

A simple test of this theory was performed to determine if an adjustment was needed to prevent sedimentation in that connection tube. The system was set to pump a 50 NTU solution into the flocculator. After reaching a steady state influent NTU, the influent tube was rigorously perturbed to dislodge any clay that may have settled in the lines. As expected, the perturbation dislodged some amount of clay into the stream and the observed influent turbidity spiked significantly and slowly returned to roughly the previous steady state turbidity level. As a first attempt to solve this problem, the original rubber tubing used to connect the clay stock to the influent stream was shortened to about 25 cm---leaving only a couple cm of free tubing on each side of the peristaltic pump-head. Two new test runs were performed at multiple turbidities with a 25 ft and a 10 ft flocculator tube.

The other challenge was the presence of data discontinuities, which periodically produced settling curves that did not match the pattern of the other settling curves. Figure 9 shows that the data from flow rates of 2.2, 2.35, and 2.95 mL/sec are inexplicable higher than what one would expect them to be given what the trend of the rest of the curves. These discrepancies have consistently appeared to be a translation of a set of data (usually the set corresponding to a particular flow rate) to a higher overall turbidity. This was originally thought to have been an error introduced by the MathCAD data processing program when it extracted the information out of the raw-datalog, but after having manually extracted and plotted the data in excel, it was clear that the Data Processor code was working properly and that the supposed discrepancies were representative of how the raw data was collected.

Figure 4 and Figure 5 both show a three dimensional plot of an experiment run with a 25 ft flocculator and which also included the adjustments made on the truncated clay inlet tube. The experimental data display very nice settling patterns that could be described at first order decay. Unfortunately, the issue of clay settling inside the flocculator is still an issue. The initial effluent turbidities among the various flow rates are still not equal. While influent turbidities all reach an equal steady state turbidity value, it seems that while floc sedimentation inside the flocculator may not seem significant to the observer, the amount of settling accrued throughout the length of the flocculator adds up and may have become rather significant. Additionally, the effluent turbidity never reaches below 10 NTU---a less than optimal result. However, unlike the results from the experiments corresponding to Figure 8 and Figure 3, the final settling turbidities do seem to be effected by flow rate. Figure 5 shows that the settling turbidity gradually increases with increasing flow rate. Initially, this observation supports the idea that as shear intensity increases in the flocculator, flocs brake-up occurs and counteracts the effectiveness of the increased amount of mixing. However, a mathematical model should be developed to better quantify this assumption. Currently, all of the conclusions this team is making are based on educated evaluations of the graphical results from the various experimental runs.

Figure 6 and Figure 7 are plots of an experimental run performed at an influent of 50 NTU with a 10 flocculator prior to the rubber tube adjustment. Notice how both the initial effluent turbidities and the final settling turbidities increase with increasing flow rate (see Figure 7). Notice also that there is very little additional settling after the first 100 seconds of settling---the settling curve remains flat relative to the decline observed in the initial 100 seconds of the state. After the adjustment of the tube, an identical experiment was performed on the 10 ft flocculator and those results are displayed in Figure 9 and Figure 10. While it is difficult to determine which of the points or lines correspond to which turbidity run, it is clear that the settling curves show the same trend as the previous experiment. With a flocculator of only 10 ft, the flocs formed are either not large enough to settle out or an inadequate amount of flocs were formed. One way of gaining more insight on the quality and size of flocs would be to document the settling column visually either by setting up a still-frame camera that can take snapshots of the settling column periodically, or setting up a video recorder to record the experiments during the loading and settling states. A better method may be to contact the manufacturers and inquire about the possibility of bypassing the onboard data averager and just record the data directly from the optical sensor and into our datalog. The plot of the level of mixing (G¿) with respect to the flowrate for the three different flocculator lengths is shown in Figure 11.

Unknown macro: {cloak}

Unknown macro: {toggle-cloak}
Discussion

Unknown macro: {cloak}

The data obtained from the this experiment showed the expected characteristics, but more analysis must be performed. Although the results were not very conclusive, it seems as though the team is progressing in the right direction. If a mathematical model was developed to enhance the data analysis, the experimental data could become even more useful. While the experimental setup still has some flaws---especially with respect to clay sedimentation and keeping effluent turbidities constant, the setup proved to be robust and capable of gathering data that accurately represents what is happening in the flocculator at the operating parameters tested. There are also some doubts about the accuracy with which the effluent turbidimeter measures a sample with large suspended flocs. Since the flocs move in and out of the path of the light, fluctuations affect the way the turbidimeter averages and displays the turbidity. Perhaps multiple successive runs of the exact same experiment can be done to determine whether averaging multiple runs can help refine the data.

Unknown macro: {cloak}

Back to Tube Floc Research

  • No labels