Modeling Filter Performance and Standardizing Sand

The members of Stacked Rapid Sand Filter Theory aim to develop a mathematical model for the performance of the sand filter. This model will take input parameters like influent turbidity and coagulant dosage and measure head loss, or the amount of energy that will dissipate from the water. Flow rate is kept constant, and in order to best simulate an actual sand filter with their laboratory model, the velocity is also kept constant, at 1.8 millimeters per second.

Historically, SRSF Theory has been as much about empirics as about theory. Previous semesters built a two-column filter in order to compare surface with subsurface filtration. Surface filtration entails water entering through the top of the filtration column, while subsurface filtration refers to water entering through the side.

This semester, the team built a new model that features two 20-centimeter layers of filtration. Despite the difference between the two-layer model used in the lab and the six-layer filter in the field, prior calculations and considerations ensured that research results would translate well. For example, tube sizes for the model were decided upon based on the metrics taken from stacked rapid sand filters in Honduras.

The team also spent much of the semester implementing an algorithm for Process Controller, the software that controls the pumps and measures turbidity. The algorithm, called a proportional-integral-derivative controller, measures error in a process and using data from past experiments, attempts to correct it. In some of the earlier experiments, the influent turbidity fluctuated slightly even when it was supposed to constant. The PID was applied to mitigate any inaccuracies.

This semester’s goal is to collect data from experiments with varying coagulant dosage and constant influent turbidity, measuring for resulting head loss and effluent turbidity.

So far, AguaClara’s stacked rapid sand filters have made appearances in both Honduras and India. However, while the design of the filter remains similar despite geographical distances, the sand used almost surely isn’t. For this reason, the Sand Source and Testing team seeks to develop a set of standards for what sand can be used, and a set of a procedures for finding out whether a sample is viable or not.

In India, they’re currently vetting sand samples based only on size, and not on anything like acid solubility. This can pose a problem for their sand filter, for example, if the sand they’re using contains limestone, which dissolves in contact with water.

Tests for sand are gathered from various sources, from the America Water Works Association to the American Society for Testing and Materials, to AguaClara’s own internal guidelines. The team’s goal is to tailor the myriad tests for sand not just to fit the needs of the stacked rapid sand filter, but also so that operators anywhere in the world can easily conduct them with the resources available in their setting.

For example, a test for silica content was eliminated because it was impractical and inessential. Other tests were removed because the nature of stacked rapid sand filtration rendered them redundant. Ultimately, it was decided that tests for the acid solubility, porosity, and uniformity coefficient and effective size–essentially, how similar grains are to each other in a given sample– were most important and suited for the SRSF’s needs.

The only real issue regarding the sand used in India is solubility; given the current backwash velocity of AguaClara’s sand filters, the margin of error in regards to the grain size of a sand sample are very wide, and so the tests currently being use in Honduras and India are adequate in that regard. Meghan and Rebecca’s research will provide operators with a more precise method of not only measuring the size of sand, but it’s solubility. Their results also indicate that as long as the sand is determined to be insoluble, then it’s fairly likely that that sand is viable for use in AguaClara plants.

Improving Floc Formation for Cleaner Water

The Laminar Tube Flocculator’s current goals are based on a floc breakup theory by one of our former graduate students, Karen Swetland. Flocs are formed by unwanted particles sticking to each other to create unwanted masses in the water. These undesirable masses are then separated from the water later. Swetland's theory basically hypothesizes that when larger flocs are broken up during the flocculation process, they are given the opportunity to regrow and collect more particles, thereby resulting in a lower residual turbidity. 

The setup of the laminar tube flocculator team. The machines attached are called turbidimeters, and
measure the influent and effluent turbidity.

To test this theory, Vicki and Yining are conducting a series of experiments using clamps with openings measuring 4 millimeters and 5 millimeters, something that was decided this past summer to break up flocs and most effectively achieve the desired residual turbidity levels. Coagulant dose is varied along with clamp size, and each configuration of the two is tested multiple times in order to make sure that experiment results are consistent with themselves. As a control for the clamp test, they collected data on the effect that having no clamps had on the residual turbidity.

So far, they have found that having one clamp in the middle arrangement of the tubing does not reduce residual turbidity when compared to the base case data. Their next steps are to test with multiple clamps to see if breaking up flocs more frequently throughout the process will reduce residual turbidity.

Lab-scale turbulent tube flocculator. Newly built!

Laminar tube flocculation is useful primarily at the scale of laboratory experiments because the movement of the particles through the water is very orderly, resulting in more straightforward analyses. Compare this with turbulent tube flocculation, where the movement of particles in water is erratic and difficult to analyze.

However, turbulent tube flocculation best describes the process of the actual plants in Honduras. This semester, the turbulent tube flocculator team was primarily concerned with building a scaled version of the flocculator based off of designs by the Summer 2013 team. However, they hope to, like the laminar tube floc team, test Karen Swetland’s floc breakup research. While Karen’s research was done with the laminar system in mind, turbulence may have some effects on flocculation that aren’t reflected in testing with its laminar counterpart.

Both research teams’ efforts are centered on achieving lower residual turbidity. Regardless of whether floc breakup will help them better achieve this end, the results of their research will have implications for how the full-scale flocculator will be designed.

Creating Better Flocs Through Measurement and Mixing

A tool that allows for the analysis of flocs would aid AguaClara’s current research significantly. Flocs are the masses formed by particles in the water after having been dosed by coagulant. Conclusions drawn from the analysis of these flocs for attributes such as size distribution have implications especially for our laminar and turbulent tube floc teams. For example, the laminar tube floc team is currently working to see if breaking flocs up somewhere during the flocculation process will create better flocs and thus cleaner water. A member of the AguaClara team, Tiago Viegas, is currently conducting research that will hopefully allow them to analyze the results of their experiments more precisely.

Tiago’s design for the floc size measurement tool consists of a square tube called the flow cell, and a camera. The flow cell’s square shape is meant to minimize distortion. The camera, specially suited for capturing accurate images of the flocs, will provide high-quality images from which we will be able to obtain information about flocs in a variety of different situations. With these images, for example, we would be able to find the most efficient floc size distribution by feeding the sedimentation tank with different distributions and analyzing each one accordingly.

For analysis of the images, it was determined that LabVIEW would be best for managing the images and data provided by the tool’s measurements.

We’re currently looking to find the most suitable camera for the job, and we’re also working with a glassworks company to create a flow cell with dimensions that match the pipes of our flocculators.

Integral to the floc formation that Tiago’s tool will analyze, however, is our coagulant, polyaluminum chloride, otherwise known as PACl. PACl is delivered to raw water through the stock tank. As of right now, PACl is distributed to the raw water and stirred manually. While this method is acceptable, the coagulant sometimes isn’t evenly distributed throughout. Alyx Cheng and Apoorv Gupta of our Stock Tank Mixing subteam are trying to devise a method for mixing the two that will ensure even distribution of PACl.

Much of the team’s past work has been empirical; Alyx and Apoorv are working off a system built in past semesters. This semester, however, they’re more concerned with the theory and calculations behind the mixing. They’ve been able to find a uniform relationship between density and concentration of PACl with the help of a hydrometer, a tool used to measure the density of a given solution.

One of their main challenges this semester is to determine the relationship between pump speed and life height of the coagulant, taking into consideration watts of power of the human arm, potential energy, and drag force. Through their calculations so far, the team has discovered large discrepancies between their theory and their lab results concerning the relationship between energy used and lift height.

The continuous lines on the left-hand side of the graph above show the relationship between the two variables yielded through calculations, while the plots on the right show the results of actual experimentation. The baffle denoted in the legend refers to a slab of plastic that was installed in the stock tank in an attempt to make the system more efficient.

Thanks to the baffle, the team was able to conclude that the error was not just the result of an accident during testing. 

A diagram denoting the variables used in calculations.

The main issue now is that while their experimental results are consistent with each other, they're not consistent with the theoretical calculations. In order to resolve this, the team is working to make their model stock tank as efficient as possible. Design changes include the addition of a t-joint to stabilize the pump at the bottom and planks to stabilize it at the top. The direction of the output of the coagulant was also changed. 

All calculations are made to ensure that any results yielded from their model will translate well to the larger sized version used in the field. Their most important goal right now is to determine the relationship between total power required and flow rate of solution out of the arm, as that will reveal the ideal distribution of the coagulant.

Improvements to the Automated Design Tool

AguaClara’s Automated Design Tool allows communities to not only expediently obtain a detailed plan of their plant, but also to effectively gauge how feasible or appropriate AguaClara technology is for their needs. The design tool takes input values such as flow rate and location, and produces a report of the plant, a model rendered in AutoCAD, and a list of supplies, all of which are adjusted according to those initial given parameters.

Integral to the function of AguaClara’s Automated Design Tool (ADT) is the Design team, a group of students who code all the functions and algorithms that make this automation possible. Design team members work from the results of our research teams and add or edit code that reflect design changes in our Automated Design Tool. Each part of the plant has its own code; these programs are integrated into and parsed by a file called EtFlocSedFi that runs as the engine for the ADT.

The Design team for this semester is split between several different tasks. Two of our members are creating functions that will display section cuts of the plants. This was originally a tedious process that engineers had to do manually, so we’re seeking to automate it as part of our ADT. A second group is integrating a new insight concerning the design of the sedimentation tank to the current code. Specifically, they’re writing functions that will tweak the placement of the inlet jet diffuser for better floc formation and add additional supports to either side of one of the tank’s pipes.

Derrick Yee is organizing the pieces of code that comprise the design tool. Currently, the code lacks organizational comments and so it’s not immediately clear to a new user how it works. Derrick is in the process of creating guidelines that standardize the coding procedures so that future members of the design team can start working on the code without such a steep learning curve. He’s also building a calculator to go with the current design tool so that communities will not only be able to know the specifications of their plant, but also how much the project will cost them.

Heidi is building a modular version of the design tool, which means that in the future our design tool will provide communities with plans for any single one of AguaClara’s technologies that they might need. Currently, the design tool is built so that the code for each part of the plant is dependent on the code of at least one other plant component. Heidi needs to rewrite the code so that all the parts are independent. When she’s finished, not only will users be able to obtain detailed plans for any given part of the plant, but they’ll be able to input more specifications; in addition to flow rate, the modular design tool will take inputs for energy dissipation rate, among other parameters.

Our current design tool gives designs for a plant that isn’t necessarily optimal for lower flow rates, and so Julia has been writing a design tool for low-flow plants. Her design centers around the research done by the Low Flow Stacked Rapid Sand Filtration Team, but it also improves flocculation given the lower flow rate, making the sedimentation tank both more efficient and more affordable.

Julia's plant design for low flow rates.

Lastly, Paul is in the process of designing flow control solutions for the plant’s entrance tank. The idea behind this is that the plant operator should have some control over the water coming into the plant, especially in cases where the water is flowing in excessive quantities. Our current method, a bronze gate valve, allows the operator to dump the entire plant flow, but it’s too expensive to implement in larger plants. After assessing the cost and efficacy of alternative solutions that either feature different valve types, multiple smaller-sized valves in parallel, or the use of Fernco flexible fittings on the top of the drain stopper pipes in the entrance tank, the latter of the three seems to be the most cost-effective solution.