ANC CONTROL

Spring 2010 Mechanisms and Hypotheses

Beginning of the Semester

We would like to demonstrate that our lime feeder can run for at least 12 hours with the effluent saturated with dissolved lime (pH 12.6). However, the Fall '09 team was unable to reach to this target with a number of trials. This semester we initially laid out several hypotheses which we believed could account for or contribute to the effluent pH dropping prematurely.

1. The presence of carbonate ions in the tap water leads to a reaction with the dissolved calcium ions to form a white calcium carbonate precipitate which coats the lime particles and inhibits further dissolution. Calcium carbonate, with a Ksp of 4.8 x 10 ^-9, is far less soluble than calcium hydroxide, which has a Ksp of 4.7 x 10 ^-6, so we would expect it to precipitate first when there is a high concentration of calcium ions in solution. The reaction would be

This may also be thought of in terms of a replacement reaction in which hydroxide is exchanged for carbonate on the surface of the solid lime, as in figure 1. Then the surface of the particles becomes much less soluble and the lime contained within them is effectively lost. The team is unsure of the specific chemical mechanism at work here.

2. Upflow velocity is the vertical component of the flow velocity. The settling velocity of the particles must be balanced with the upflow velocity in order for particles to remain in suspension. If the fluidized bed is not well-maintained by the upflow velocity in the vertical column, the lime settles into a dense bed at the bottom of the apparatus where preferential flow paths lead to insufficient contact time for full dissolution.

3. Also related to kinetics, the concentration of solid lime particles in the fluidized bed (that is, the volume of solid lime per volume of water) may decline as solid lime is lost with the effluent, so that there is not enough available solid lime surface area for the solution to become saturated within the reactor's residence time.

4. There is a minimum contact between solid lime and water that is required for the water to approach equilibrium. This required contact time is related to depth of the fluidized bed, flow rate, solids concentration in the bed, and particle size distribution.

Evaluation and Evolution of Initial Ideas

Experiments conducted using distilled water, which should not contain significant concentrations of carbonate ions, still failed prematurely, leading the team to believe that the first hypothesis is not the primary reason the feeder does not work. Without carbonates, there should be no calcium carbonate precipitation, so another mechanism must have caused the failure. However, the single experiment was not enough to eliminate the carbonates hypothesis as a complicating factor in the lime feeder function.

By observation, the team also consistently found an upflow velocity which maintained a good fluidized bed in the vertical column. A failure to maintain the suspension, as discussed in the second initial hypothesis above, is not believed to be a major cause of failure. However, there are difficulties associated with finding the appropriate flow rates for the two reactors. From observation and alkalinity tests of the effluent, the team knows that there is often significant solid lime leaving the reactor when it is fully loaded, which indicates that the upflow velocity is too high to allow the smallest particles to settle in the inclined tube. At the same time, depending on how the dry lime is broken up prior to the experiment, there are also often particles which are large enough to settle through the jet at the entrance to form a bed at the bottom of the column. Upflow velocities which produce fluidized beds depend on the particle size distribution, which is thought to be variable over the course of an experiment. The team is looking for the range of upflow velocities which allow the reactor to produce saturated effluent while minimizing the solids lost to the effluent. An additional practical limitation of this investigation is that the suspension takes a very long time to respond to changes in flow rate (which is linked directly to upflow velocity) due to the very slow settling velocity of most of the particles, which makes measuring the effects of changing the upflow velocity very tedious.

Lastly, because the new feeder design with a large-diameter tube settler has a capture velocity sufficiently low to keep the majority of particles out of the effluent, and there is always a large amount of solid lime remaining in the suspension, the third hypothesis of particle thinning does not likely explain the failure of our experiments, although the true failure mechanism may be related, since the smallest particles are carried out under most conditions.

Particle Size

Having demonstrated in experiment 2 that the lime feeders can fail even in the absence of carbonates, the team believes that a significant part of the problem is related to kinetics. Perhaps the most important variable in determining whether the solution can reach saturation within the reactor's residence time is the surface area over which the water is contacting the solid lime. This is very closely tied to particle size distribution. To maximize available surface area, we would like to maximize the area-to-volume ratio, which is inversely proportional to the particle diameter. In other words, for a given total volume of lime, smaller particle sizes yield more available surface area for dissolution.

The team now believes that the premature drop in effluent pH we observe may be due to a shift in the particle size distribution in the fluidized bed towards much larger particles, as represented in figure 2, which decreases available surface area enough that reaction kinetics are no longer fast enough to provide a saturated effluent solution. We have come up with three hypotheses for why this could happen:

1. Lime particle flocculation has been observed within the fluidized bed. This has a dramatic effect on contact surface area. We know that it happens within several hours of the addition of lime in our tests, but we have not done enough yet to characterize this process.

2. When the lime slurry is added to the apparatus, there is a broad distribution of particle sizes. The reactor's success at producing saturated effluent initially may be due to the smallest particles in the distribution, which dissolve rapidly. The observed drop in effluent pH may come at the time when the last particles small enough to provide sufficient surface area for saturation have dissolved. The larger, more visible particles which remain may be much more limited in the extent to which they can be dissolved within the residence time of the reactor, so that alone they can only provide an effluent solution orders of magnitude less concentrated than saturation. This would be consistent with the observation that there is always a significant amount of solid lime left over in the reactor, regardless of how long the experiment ran. These large particles would take a very long time to dissolve.

3. The smallest particles, which have the lowest settling velocity, are the most prone to being lost with the effluent. Thus, when solid lime is lost, it contributes to the shift in particle size distribution towards larger particles.

A shift in the particle size distribution leading to insufficient surface area for dissolution is likely some combination of these three phenomena.

Particle Coalescence

In the "Particle Size" section above it was noted that flocculation of the particles in the fluidized bed was observed. This phenomenon of lime particles sticking together was seen again with the particles sliding down the bottom of the tube settler. The team observed a large solid buildup in the smaller reactor which plugged the branched pipe segment except for some small flow paths. It was thought that the initial buildup occurred on the rim in the pipe connection and that additional settling lime added to the formation of the aggregate solid. This behavior has only been seen when the lime is broken up with a blender prior to the experiment, so the tendency of the particles to stick together seems to be related to some extent to particle size. Other surface properties and chemistry may also play a role, but this is not well understood. Clearly, just as the formation of flocs decreases the surface area available for dissolution, the formation of a solid block of lime is detrimental to the performance of the reactor.

Preferential Flow Paths

Preferential flow paths through the "clearest" water in the reactor may also be a mechanism which contributes to the reactor's failure. For example, when the large apparatus is loaded with 200 grams of lime with a flow rate of 120 ml/min, a significant amount of lime remains suspended in the upper slanted segment, but it settles to the lower side of the tube while a clean stream of water flows up the upper side, avoiding further lime dissolution, as in figure 3. This failure mode indicates that the tube settler should not contain settled lime. If more contact time is needed the vertical tube should be made longer.

Chemistry

The carbonate replacement reaction on the surface of the lime particles, mentioned above in the "Beginning of the Semester" section, is the primary chemical mechanism believed to be related to the reactor's failure. Further controlled testing with distilled water should provide further insight into the importance of this phenomenon. In a supplemental test, it was found that even in an effluent sample well below saturation concentration with significant suspended solids the pH decreases over time. The results are shown in figure 4. The steady drop in pH may be due to the effect of carbon dioxide dissolving in the open system and precipitating the calcium. In any case, the fact that the pH decreases rather than increases indicates that the solids remaining in suspension are not calcium hydroxide, which would dissolve over time and raise the pH. This suggests that the leftover solids may be primarily calcium carbonate.