Abstract

Ceramic pot filters (CPFs) are an effective point of use water treatment device in developing nations due to their low cost and effectiveness. CPFs are gravity fed, typically making water production a major limiting factor to a CPF's lifetime and acceptability. Directly connecting CPFs to in-line pumping systems or systems with an elevated storage tank would allow filter usage for constant water treatment at increased pressures, increasing the quantity of treated water. Ceramic disks were manufactured for testing in a specially designed housing apparatus. Filters of varying thicknesses and clay to sawdust mass ratios were manufactured to fit tightly. Flowrate and microbiological removal efficacy (logarithmic reduction value (LRV)) were determined over the testing period at various pressures. Flowrate values ranged from 2.44 to 9.04 L per hour, significantly higher than traditional CPF technology. LRVs ranged from 1.1 to 2.0, lower than traditional CPF technology but still effective at removing most Escherichia coli and total coliform bacteria. Filters proved effective at removing total and fecal coliforms at pressures less than 70 kilopascals. The optimum filter had a thickness of 3.2 cm and clay to sawdust ratio of 6:1 by mass. Filters proved to be ineffective if flowrates were above 5 L/h.

INTRODUCTION

Access to clean drinking water in developing nations continues to be an important concern for human health. An estimated 748 million people lack access to improved water supplies (WHO & UNICEF 2014). It is estimated that hundreds of millions of that population drink water contaminated during collection, transport, and storage (Clasen & Bastable 2003). The United Nations Children's Fund (UNICEF) (2008) states that many improved water sources in developing nations do not provide safe water due to microbiological contamination from water sources with inadequate fecal contamination protection. A drinking water source can be improved using a centralized community system or by a point of use (POU) system. POU systems are designed to provide adequate water for one household and can be effective for rural communities with no centralized improved water source. An effective household water treatment device is the ceramic pot filter (CPF) (Hunter 2009). CPFs are a flower-pot shaped filter made from mixing clay, water, and sawdust and pressed to shape typically using a hydraulic press and mold. The pots are typically coated in colloidal silver that acts as a disinfectant. These pots are fired in a kiln, removing most of the sawdust and leaving pores through which water can flow. The filter is housed in a 3.8 L bucket with a lid and filtered water collects in the bottom of the bucket. It can then be poured out using a spigot at the bottom of the bucket when needed. The main bacterial removal mechanisms for CPFs, as identified by van Halem et al. (2007), are exclusion by pore size, exclusion by effective pore size (tortuosity), and deactivation of bacteria by contact with silver.

For communities with a centralized water source, this may be an inconvenient way to filter water and some households may not be able to afford a CPF. CPFs typically have a flowrate of 1–2 L/h for quality control standards when manufactured (Lantagne et al. 2010). CPFs in households also filter water in batches and must be filled multiple times per day, depending on the number of people in a household, to produce enough water for the household. A more effective method of producing enough clean water for a household would be to put the filter in-line with a pumping system or elevated storage tank so that the user would have access to safe drinking water on demand and allow an entire community to have clean drinking water. This would help communities afford using ceramic filters and help prevent filters from being broken during filling and cleaning since the filter will stay contained in an apparatus. CPFs are a viable candidate for this type of process in developing countries due to their low cost, material availability, and relatively simple manufacturing requirements. CPFs also have been proven to effectively remove harmful bacteria in water, such as total and fecal coliforms, as documented by van Halem et al. (2007), Brown & Sobsey (2010), Kalman et al. (2011), and others.

To use the CPF in a pumping system, some significant problems need to be addressed. These problems include keeping a tight seal between the ceramic filter and the housing apparatus to prevent bypass, as well as the observed problem of the fragility of the filters. Putting a standard filter under pressure not only makes the seal more difficult to maintain, it also makes the filter element easier to break since the edges of the filter are weak points and therefore are more prone to failure. Due to this, a ceramic disk was developed as the filter element for this testing.

This paper describes a study of the use of ceramic disk filters in an in-line pumping system to test their applicability in a field setting. The primary purpose of the study was to determine the effectiveness of ceramic disk filters at removing microbiological contamination and providing effective flowrates at different pressures. This testing was performed on disks with varying clay to sawdust mass ratios and thicknesses to determine the optimal ceramic disk filter design.

METHODS

All filters used in this study were manufactured on campus at the Missouri University of Science and Technology using a synthetic clay body that was designed to closely mimic the clay used for CPF manufacturing at a factory near Antigua, Guatemala. The clay body used in this study was developed in an earlier process by Hubbel et al. (2015). The sawdust used in this study was collected from a local sawmill and sieved to remove particles that did not pass through a U.S. No. 10 sieve (2 mm slot size). The sawdust was combined with the clay in a 5:1, 6:1, and 7:1 clay to sawdust mass ratio, and then mixed with deionized water in a 1.8 kg to 1 L clay to water ratio with a mixer attached to a power drill. A total of 27 filters were constructed and tested during experimentation. This consisted of three filters of the same thickness and clay to sawdust ratio by mass to have triplicates for testing. Filters were constructed in a 20 cm by 20 cm block with thickness of 2.5 cm, 3.2 cm and 3.8 cm, and clay to sawdust ratio of 5:1, 6:1, and 7:1 by mass, respectively. The clay was pressed to the above thicknesses using a 12-ton press that allowed for pressure of 2,600 kPa for 1 min to mimic the procedure developed by Oyandel-Craver & Smith (2007) and Ren & Smith (2013) at a slightly lower pressure.

Before the pressed clay was fired, the clay mixture was allowed to dry for a minimum of 48 h, and then was inserted in a soils oven set at 35 °C. The oven's temperature was gradually increased until the temperature reached 100 °C. The filters were allowed to stay at this temperature for a 24 h period to allow for the filter block to completely dry. The filters in this study had no colloidal silver applied to them in an effort to simplify the focus of this study. Oyandel-Craver & Smith (2007) and Clark & Elmore (2011) have both found that the filters are effective at removing bacteria even with no silver present.

The clay mixture was fired in an electric kiln with the same firing schedule (temperatures and times) which slowly increased the temperature of the kiln to 993 °C over the course of approximately 14 h and then cooled back to room temperature. The kiln was propped open to allow for smoke from the sawdust burning out to escape and to allow for airflow through the kiln to maximize as much complete combustion as possible. After the clay mixture was fired, the resulting filter block was cut into a cylindrical disk with a diameter of approximately 17 cm using a high-pressure water jet at Missouri University of Science and Technology's Rock Mechanics laboratory to allow a consistent diameter and clean cut on the filter disk.

The porosity of each disk was also determined using a modified Archimedes method by using the American Society for Testing and Materials (ASTM) standard C373-88 (ASTM 2006). Porosity data for each clay to sawdust ratio by mass are presented in Table 1. Porosities of each filter ranged from 31.2% to 46.3%, with an average of 37.2%. The manufacturing process was not altered during production of each filter to prevent any changes in porosity that is not a function of the clay to sawdust ratio by mass.

Table 1

Porosity of filters tested

Clay to sawdust ratio (by mass)Number of samplesAverage porosity (%)
5:1 36.5% 
6:1 38.6% 
7:1 36.2% 
Clay to sawdust ratio (by mass)Number of samplesAverage porosity (%)
5:1 36.5% 
6:1 38.6% 
7:1 36.2% 

Prior to testing, filters were submerged in tap water for a minimum of 24 h to allow complete saturation. Tap water was then filtered through the filters and effluent water was tested for total chlorine with a Hach Total Chlorine field kit (CN-66T) to make sure that all chlorine was removed from the filter and to not allow residual chlorine to affect microbiological removal efficacy. If a test showed presence of chlorine, a solution of thiosulfate was passed through the filters to remove any excess chlorine until chlorine was at levels not detectable by testing, approximately 0.02 mg/L of free chlorine. In addition, a presence/absence Colilert test analyzed effluent from the filters to show that no coliforms were present at the beginning of testing.

The design of the in-line testing apparatus is presented as Figure 1. This apparatus was constructed using a 15.2 cm (6 inch) inner diameter Schedule 40 polyvinyl chloride (PVC) pipe. Two separate pipe sections were cut to lengths of 25 cm and capped. The caps were threaded to allow for the apparatus to be connected to a pumping system. A rubber coupling was used to connect the two separate pipe sections together and the disk filter was placed in the middle of the rubber coupling and between the two PVC pipe pieces.

Figure 1

Laboratory setup.

Figure 1

Laboratory setup.

To ensure that filters were tightly sealed to the rubber coupling of the apparatus, a wooden disk of the same diameter of the filters was placed in the apparatus and tightened with hose clamps until no water flowed out of the filter. The hose clamps were tightened using a torque wrench to measure the required torque necessary to prevent bypass. Testing results indicated this required torque was approximately 2.83 newton-metres (N-m). All filters tested were tightened to this torque using the same torque wrench. The top pipe was glued to the rubber coupling to allow it to stay in place when under higher pressures. A pressure gauge was attached to the upper pipe to measure pressure coming into the disk filter. The arrows indicate the direction of flow of the system. A recycle line was attached to regulate pressure entering the system.

Tests were conducted using a 750 Watt centrifugal pump. The pump was connected to the filter apparatus using 2.5 cm diameter Schedule 40 PVC pipe. Flowrate tests were conducted at 60 min increments. A ball valve was connected at the top of the pipe system to remove any air from the system during testing and to take influent samples from the apparatus.

Microbiological testing was conducted during one hour long tests, initially at pressures of 34 kPa. Filters that proved to effectively remove bacteria under this pressure were tested at 70 kPa on a different date during a different test. Challenge water was created using an approximately 3% mixture of raw influent wastewater from the local publicly owned treatment works wastewater treatment plant and tap water to try and cause failure in the filters. By causing failure in the filters, a logarithmic reduction value (LRV) can be calculated for each filter. LRV is a typically used value to measure bacterial removal efficacy in POU water treatment systems (van Halem et al. 2007; Clark & Elmore 2011). The formula for LRV is presented in Equation (1). Raw water microbiological levels were measured before testing as well as after filtering through the filters. Filters were allowed to filter water for an hour prior to testing. Microbiological testing was conducted using Colilert and the Colilert Quani-Tray 2000 to determine the presence of total coliforms and fecal coliforms in the filter effluent.  
formula
(1)
where A is the number of viable microorganisms before treatment and B is the number of viable microorganisms after treatment.

The non-parametric two-sample Wilcoxon rank sum test, also known as the Mann-Whitney test, was used to evaluate the relationships between flowrate, porosity, thickness, and clay to sawdust ratio to LRV. The Mann-Whitney test was performed since none of the data fit a normal distribution based on using Minitab's distribution analysis. A p-value that is less than one minus the confidence interval results in rejecting the null hypothesis. The confidence interval used for this testing was 95%, meaning a p-value of greater than 0.05 indicates a relationship between the two variables tested.

RESULTS AND DISCUSSION

The flowrate and LRV from each test for each individual filter were determined and a summary of these values for each filter based on thickness and clay to sawdust ratio by mass can be found in Table 2. LRV was calculated by taking the log10 of the influent bacteria concentration (most probable number [MPN]/100 mL) divided by the log10 of the effluent bacteria concentration from the filter (MPN/100 mL). According to WHO (2011), the performance target for a household water treatment system for bacteria is a LRV of at least 2. Therefore, filters with LRV of 2 or greater were considered effective at removing bacteria. Microbiological testing results are presented in Table 3. Filters with thicknesses of 3.2 cm appear to be the most effective at removing bacteria from influent water, especially in the filters with a 5:1 and 6:1 clay to sawdust ratio. Based on this statistical summary, further testing was performed based on different characteristics of the filters.

Table 2

Summary statistics of filters grouped by thickness and clay to sawdust ratio by mass

Clay to sawdust ratio (by mass)Thickness (cm)Mean
Standard deviation
Coefficient of variation
Q (L/hr)LRVQ (L/hr)LRVQ (L/hr)LRV
5:1 2.5 3.8 1.3 2.09 0.5 0.55 0.39 
6:1 2.5 2.44 2.1 1.27 0.48 0.521 0.23 
7:1 2.5 3.18 1.49 0.97 0.469 0.49 
5:1 3.2 4.06 2.5 0.98 0.615 0.48 
6:1 3.2 2.17 2.1 1.48 0.66 0.682 0.31 
7:1 3.2 7.59 1.3 9.13 0.44 1.2 0.34 
5:1 3.8 9.04 1.1 11.9 0.66 1.31 0.61 
6:1 3.8 2.85 1.4 0.82 0.56 0.287 0.42 
7:1 3.8 7.55 1.1 2.77 0.24 0.366 0.22 
Clay to sawdust ratio (by mass)Thickness (cm)Mean
Standard deviation
Coefficient of variation
Q (L/hr)LRVQ (L/hr)LRVQ (L/hr)LRV
5:1 2.5 3.8 1.3 2.09 0.5 0.55 0.39 
6:1 2.5 2.44 2.1 1.27 0.48 0.521 0.23 
7:1 2.5 3.18 1.49 0.97 0.469 0.49 
5:1 3.2 4.06 2.5 0.98 0.615 0.48 
6:1 3.2 2.17 2.1 1.48 0.66 0.682 0.31 
7:1 3.2 7.59 1.3 9.13 0.44 1.2 0.34 
5:1 3.8 9.04 1.1 11.9 0.66 1.31 0.61 
6:1 3.8 2.85 1.4 0.82 0.56 0.287 0.42 
7:1 3.8 7.55 1.1 2.77 0.24 0.366 0.22 
Table 3

Average microbiological testing results grouped by thickness and clay to sawdust ratio

Clay to sawdust ratio (by mass)Thickness (cm)Number of testsInfluent total (MPN)Influent E. coli (MPN)Effluent total (MPN)Percent removal totalEffluent E. coli (MPN)Percent removal E. coli
5:1 2.5 5,793 97.4 278.7 93% 6.6 93% 
5:1 3.2 43,983 3,411 563.7 99% 268.6 91% 
5:1 3.8 26,198 2,249.6 959.0 92% 425.6 86% 
6:1 2.5 33,266 2,795.3 597.4 99% 238.3 89% 
6:1 3.2 25,633 1,833.6 1,000.0 99% 101.4 95% 
6:1 3.8 34,109 9,241.9 1,718.2 92% 786.0 84% 
7:1 2.5 22,913 1,557.4 863.1 94% 415.2 84% 
7:1 3.2 27,557 1,642.7 1,230.6 93% 193.9 89% 
7:1 3.8 33,310 9,882.9 2,029.0 91% 1,094.3 86% 
Clay to sawdust ratio (by mass)Thickness (cm)Number of testsInfluent total (MPN)Influent E. coli (MPN)Effluent total (MPN)Percent removal totalEffluent E. coli (MPN)Percent removal E. coli
5:1 2.5 5,793 97.4 278.7 93% 6.6 93% 
5:1 3.2 43,983 3,411 563.7 99% 268.6 91% 
5:1 3.8 26,198 2,249.6 959.0 92% 425.6 86% 
6:1 2.5 33,266 2,795.3 597.4 99% 238.3 89% 
6:1 3.2 25,633 1,833.6 1,000.0 99% 101.4 95% 
6:1 3.8 34,109 9,241.9 1,718.2 92% 786.0 84% 
7:1 2.5 22,913 1,557.4 863.1 94% 415.2 84% 
7:1 3.2 27,557 1,642.7 1,230.6 93% 193.9 89% 
7:1 3.8 33,310 9,882.9 2,029.0 91% 1,094.3 86% 

LRV as a function of porosity

A comparison was performed on the filters tested, comparing the effect that flowrate, porosity, and thickness have on the calculated LRV values. Figure 2 shows LRV plotted against the measured porosity for each filter for each filter thickness. From this graph, no correlation between bacterial removal and porosity was observed for any filter thickness. This phenomenon has been seen in other research conducted on CPFs by Soppe et al. (2015) and White et al. (2015), both of which showed little correlation between porosity and LRV. Soppe et al. (2015) concluded that bacterial removal will only be compromised by the size of the burnout material, not necessarily the amount.

Figure 2

LRV vs. measured porosity for each thickness group.

Figure 2

LRV vs. measured porosity for each thickness group.

The Mann-Whitney test was performed to test if LRV was related to porosity, with the null hypothesis being that there was a relationship between the two variables. The results from these tests showed a p-value of 0.000, meaning the null hypothesis is rejected and that there is no relationship between LRV and porosity for this testing.

LRV as a function of thickness

Figure 3 shows the relationship between LRV and thickness of the filter. From this graph, it can be observed that the 3.2 cm thick filters performed the best in regards to LRV. The mean LRV of the filters with 3.2 cm thickness was the greatest of all filters tested, followed closely by the 1 inch thick filters. From this figure, it can be seen that the range of LRVs is quite large, with most ranges being 2 LRV or greater. It also shows that the 3.8 cm thick filters were the most ineffective filters used. This does not follow the hypothesis that the thicker the filters, the more effective the filters will be when put under pressure. This could be due to the higher flowrates that were seen in the filters with thicknesses of 3.8 cm. Filters with 3.8 cm thicknesses also appeared to not fire completely through the filter. This caused filters to be more fragile and chip around the edges of the filter, reducing the effective thickness of the filters.

Figure 3

Box and whisker plot LRV vs. thickness based on clay to sawdust ratio by mass.

Figure 3

Box and whisker plot LRV vs. thickness based on clay to sawdust ratio by mass.

The Mann-Whitney test was performed on all filters used during testing, with the null hypothesis being that LRV were related to thickness. The p-value for the LRV to thickness test was 0.0031, rejecting the null hypothesis. This suggests that thickness of the filter is not related to LRV. This analysis may have been affected by the 3.8 cm filters since these filters showed higher flowrate and consistently low LRV.

LRV as a function of flowrate

Figure 4 shows the relationship between LRV and flowrate grouped by filter thicknesses. Flowrate is a commonly used parameter for quality control in CPF factories to determine the effectiveness of filters to remove bacteria (Ceramics Manufacturing Group 2011). This would indicate that there would typically be a relationship between flowrate and LRV (namely, that filters with a higher flowrate would allow more bacteria to pass through the filter and vice versa), as shown by White et al. (2015). A scatter plot of the data in these tests did not show a direct relationship between flowrate and LRV, especially in the filters that were 2.5 cm thick. The data did show that any flowrate above 5 L/h would unlikely give an LRV value of 2 or greater.

Figure 4

Flowrate vs. LRV based on filter thickness.

Figure 4

Flowrate vs. LRV based on filter thickness.

The Mann-Whitney test was performed on each clay to sawdust ratio groups based on thicknesses, testing to see if flowrate and LRV were related. The null hypothesis was that flowrate and LRV were related. The results from this testing can be seen in Table 4. From this statistical test, the filters that showed a relationship between flowrate and LRV were the 5:1 3.2 cm, 6:1 2.5 cm, 6:1 3.2 cm, and 7:1 2.5 cm. These filters are filters that had a mean LRV value greater than or equal to 2. The other five filter groups did not show relationships between flowrate and LRV and also had mean LRV values of less than 2. Analysis of these five filter groups indicated the reasons why these filters did not show a relationship between LRV and flowrate included having filters with higher flowrates than the necessary 5 L/h and that filters in these groups experienced more chipping than filters in the groups that did show a relationship. The 5:1 clay to sawdust ratio by mass, 3.8 cm thickness filters were especially ineffective due to most of the filters not completing firing in the kiln. This caused filters to be more fragile and this group had higher flowrates than most filter groups.

Table 4

Mann-Whitney test results for flowrate and LRV for each filter group

Clay to sawdust ratio (by mass)Thickness (cm)p-value
5:1 2.5 0.001 
5:1 3.2 0.204 
5:1 3.8 0.004 
6:1 2.5 0.453 
6:1 3.2 0.791 
6:1 3.8 0.001 
7:1 2.5 0.158 
7:1 3.2 0.014 
7:1 3.8 0.000 
Clay to sawdust ratio (by mass)Thickness (cm)p-value
5:1 2.5 0.001 
5:1 3.2 0.204 
5:1 3.8 0.004 
6:1 2.5 0.453 
6:1 3.2 0.791 
6:1 3.8 0.001 
7:1 2.5 0.158 
7:1 3.2 0.014 
7:1 3.8 0.000 

The Mann-Whitney test was also performed testing the relationship between LRV and porosity, thickness, and flowrate while taking out the 3.8 cm filters. From this testing, there was still no relationship found between LRV and porosity, thickness, and clay to sawdust ratio. There was no relationship found from literature review between these relationships as it does not appear to have been tested yet.

Testing conducted at 10 psi

Table 5 shows data collected during testing at 70 kPa. The letter designates the filter name used during testing. Not all filters that did not have a 2 LRV at 34 kPa were tested at 70 kPa since it was shown that if filters were not effective at 34 kPa they would not be effective at 70 kPa, as evident by filters 5:1 2.5 cm C, 6:1 2.5 cm C, and 7:1 3.2 cm B. Results indicate that the LRV is typically lower than at lower pressures, but acceptable LRVs can be obtained.

Table 5

Results of tests conducted at 70 kPa

Clay to sawdust ratio (by mass)Thickness (cm) and specimen IDLRV 1Q 1 (L/hr)LRV 2Q 2 (L/hr)
5:1 2.5 C 0.78 20.2   
5:1 3.2 B 1.9 6.74 2.7 4.98 
5:1 3.2 C 1.9 4.25 2.6 3.61 
6:1 2.5 A 1.9 8.98 1.4 7.63 
6:1 2.5 C 1.01 7.88   
6:1 3.2 A 1.7 3.76 1.0 4.75 
6:1 3.2 C 1.9 3.32 1.4 5.57 
7:1 2.5 B 2.1 2.12 1.7 3.56 
7:1 2.5 C 2.2 4.55 1.7 5.94 
7:1 3.2 B 0.761 18.2   
Clay to sawdust ratio (by mass)Thickness (cm) and specimen IDLRV 1Q 1 (L/hr)LRV 2Q 2 (L/hr)
5:1 2.5 C 0.78 20.2   
5:1 3.2 B 1.9 6.74 2.7 4.98 
5:1 3.2 C 1.9 4.25 2.6 3.61 
6:1 2.5 A 1.9 8.98 1.4 7.63 
6:1 2.5 C 1.01 7.88   
6:1 3.2 A 1.7 3.76 1.0 4.75 
6:1 3.2 C 1.9 3.32 1.4 5.57 
7:1 2.5 B 2.1 2.12 1.7 3.56 
7:1 2.5 C 2.2 4.55 1.7 5.94 
7:1 3.2 B 0.761 18.2   

The use of ceramic filter disks under pressure has a potential to be an efficient and economical way to filter water, especially in a pressurized system. The filters that showed the most consistent removal were filters with a thickness of 3.2 cm. The porosity of the filter does not seem to be a primary factor in the LRV of filters. This indicates that porosity of the filters can be a value that can be as high as the manufacturer wants as long as it does not affect the strength of the filter. The maximum porosity of a filter that did not break that was completely fired during our testing was 46.5%. More testing is necessary to determine the strength of the filter based on porosity.

One limiting factor during testing was the fragility of the filters when under pressure. When the filters are put into the apparatus, the hose clamps are tightened and the filter is compressed within the rubber coupling. During testing, filters were taken in and out of the apparatus multiple times. This led to filters chipping and breaking. If filters were left in place during operation, the filters should have less chipping than the filters tested during this study. This compression caused some of the filters to chip around the sides of the disk, making it much more difficult to keep the filter tightly sealed to the rubber coupling. Some of the filters also cracked when pressure was induced on them. Most filters remained working under 34 kPa conditions although some cracked down the middle of the filter, making them ineffective. This phenomenon was increased for filters that were put under a pressure of 70 kPa. Filters should be completely fired and have complete combustion to increase strength in the filters.

Tests indicate that the filter disks were unable to effectively remove bacteria when flowrate through the filter was above 5 L/h. Therefore, if a filter has a flowrate above 5 L/h, the filter will not effectively remove bacteria under pressures equal to or greater than 34 kPa. This is greater than the recommended 1–2 L/h, but Soppe et al. (2015) have stated that gravity fed filters could be effective at flowrates of up to 10 L/h. Testing also indicated that LRV and flowrate are related, according from the results of the Mann-Whitney testing. Porosity and thickness do not seem to have a relationship to LRV from this testing. Further analysis should be performed to determine a better correlation between flowrate and LRV.

CONCLUSIONS

Ceramic disk technology has proven to be an effective method of treating water under pressure, and reduces microbiological contamination at a much higher flowrate than traditional CPF technologies. However, LRV values of traditional CPF technology were not consistently achieved. Therefore, this technology would prove beneficial as a polishing filter at locations with minimal concentrations of microbiological contaminants. The ability to reduce the quantity of materials while increasing flowrate would reduce costs significantly. This would also allow ceramic disk technology to be affordable in markets currently devoid of filter technology such as Africa. The ability to adapt this technology under pressure would allow the system to be adapted to pumping wells in target communities, which would allow a community-based remedial approach. This would prevent the need for treatment at individual households, and allow for entire communities to gain access to clean water.

A long-term study on the use of filter disks under pressure would allow better characterization of the lifetime of the ceramic disk filter and the ability of the filter to consistently remove bacteria at acceptable flowrates. More tests also need to be performed to improve the filters ability to stay consistently tight to the rubber coupling. Thicker filter disks seemed to have a higher chance of being ineffective. This seems counter intuitive since the thicker filters should remove more bacteria than a thinner filter but our testing did not show this result. A complete firing of the filter also appears to be important in removal when filters are put under pressure. Filters with a zone that did not completely fired never reached LRVs of 2 during testing. A dye test could be performed on the filters to test if bypass was occurring. This would help validate if a filter was performing as intended and a more ideal filter design could be implemented for use in the field.

REFERENCES

REFERENCES
American Society for Testing and Materials (ASTM)
2006
Standard Test Method for Water Adsorption, Bulk Density, Apparent Porosity, and Apparent Specific Gravity of Fired Whiteware Products 2006, C373-88
.
West Conshohocken, PA
,
USA
.
Ceramics Manufacturing Working Group
2011
Best Practice Recommendations for Local Manufacturing of Ceramic Pot Filters for Household Water Treatment
, 1st edn.
Centers for Disease Control and Prevention
,
Atlanta, GA
,
USA
.
Clark
,
K. N.
&
Elmore
,
A. C.
2011
Bacteria removal effectiveness of ceramic pot filters not applied with colloidal silver
.
Water Science and Technology: Water Supply
11
(
6
),
765
772
.
Clasen
,
T. F.
&
Bastable
,
A.
2003
Fecal contamination of drinking water during collection and household storage; the need to extend protection to the point of use
.
Journal of Water and Health
1
(
3
),
109
115
.
Hubbel
,
L.
,
Elmore
,
A. C.
&
Reidmeyer
,
M.
2015
Comparison of a native clay soil and an engineered clay used in experimental ceramic pot filter fabrication
.
Water Science and Technology: Water Supply
15
(
3
),
569
577
.
Kalman
,
E. N.
,
Oyanedel-Craver
,
V. A.
&
Smith
,
J. A.
2011
Ceramic filters impregnated with silver nanoparticles for point-of-use water treatment in rural Guatemala
.
Journal of Environmental Engineering
137
(
6
),
407
415
.
Lantagne
,
D.
,
Klarman
,
M.
,
Mayer
,
A.
,
Preston
,
K.
,
Napotnik
,
J.
&
Jellison
,
K.
2010
Effect of production variables on microbiological removal in locally-produced ceramic filters for household water treatment
.
International Journal of Environmental Health Research
20
(
3
),
171
187
.
Oyandel-Craver
,
V. A.
&
Smith
,
J. A.
2007
Sustainable colloidal-silver-impregnated ceramic filter for point-of-use water treatment
.
Environmental Science and Technology
42
,
927
9330
.
Soppe
,
A. I. A.
,
Heijman
,
S. G. J.
,
Gensburger
,
I.
,
Shantz
,
A.
,
van Halem
,
D.
,
Kroesbergen
,
J.
,
Wubbles
,
G. H.
&
Smeets
,
P. W. M. H.
2015
Critical parameters in the production of ceramic pot filters for household water treatment in developing nations
.
Journal of Water and Health
13
(
2
),
587
599
.
United Nations Children's Fund
2008
UNICEF Handbook on Water Quality
.
UNICEF
,
New York, NY
,
USA
.
van Halem
,
D.
,
Heijman
,
S. G. J.
,
Soppe
,
A. I. A.
,
van Dijk
,
J. C.
&
Amy
,
G. L.
2007
Ceramic silver-impregnated pot filters for household drinking water treatment in developing countries: material characterization and performance study
.
Water Science and Technology: Water Supply
7
(
5–6
),
9
17
.
White
,
E. P.
,
Langenfeld
,
J. K.
,
Bradford
,
E. L.
,
Haywood
,
H. G.
,
Salvinelli
,
C.
&
Elmore
,
A. C.
2015
Assessment of Flow Rate as a Quality Control Test for Ceramic Pot Filters. In: World Environmental and Water Resources Congress 2015
, pp.
1378
1387
.
World Health Organization
2011
Evaluating Household Water Treatment Options: Health-Based Targets and Microbiological Performance Specifications
.
World Health Organization Water Sanitation Health
,
Geneva, Switzerland
.
World Health Organization & United Nations Children's Fund
2014
Progress on Sanitation and Drinking-Water – 2014 Update
.
Joint Monitoring Programme for Water Supply and Sanitation, WHO
,
Geneva
,
Switzerland
.