Ceramic pot filters lifetime study in coastal Guatemala

Lack of safe water, sanitation, and hygiene remains a serious world health issue. According to the World Health Organization & United Nations Children's Fund (WHO & UNICEF) (2015), 663 million people still lacked access to an improved drinking water source in 2015. Most of these people (79%) live in rural areas, and almost a quarter rely on untreated surface water. According to UNICEF (2008), water quality interventions in developing countries have a greater impact when applied at the household level. Work by Hunter (2009) also defined ceramic filters as the most effective long-term household water treatment systems. Moreover, several laboratory studies have been conducted to assess the effectiveness of ceramic pot filters (CPFs), including van Halem et al. (2007, 2009), Oyanedel-Craver & Smith (2008), Sobsey et al. (2008), Lantagne et al. (2010), Mwabi et al. (2013) and others. Although field investigations are optimum in understanding filter use, acceptance, and behavior by users in real household conditions, they are more difficult logistically to execute and are not as abundant in the literature as laboratory studies. According to Brown & Sobsey (2006), knowledge of CPF effectiveness over long periods in the field is an essential condition for successful scale-up and responsible investment, but it has not been sufficiently studied.

Lantagne (2001) described a three-week field investigation conducted in Nicaragua about the performance of CPFs distributed as an emergency response after Hurricane Mitch in October 1998. The study included water quality monitoring and a survey of filter users. It was concluded that less than 53% of the filters removed Escherichia coli and post-treatment contamination as a result of water storage in unclean receptacles was a major issue. It was also observed that monitoring visits to families using the filters are strongly correlated with continued use of the filters. A field study in Cambodia (Brown et al. 2009) documented a relatively low rate of abandonment of approximately 2% per month. The study also found that the log₁₀ reduction value (LRV) of E. coli did not appear to have a strong correlation with time in use. Another field study on the effectiveness of CPFs in Cambodia (Roberts 2004) included water quality testing and user surveys and concluded that 99% of CPFs produced water meeting the WHO ‘low risk’ requirements and users experienced a reduction in the rate of waterborne diseases. A retrospective study of filters distributed in Cambodia described by Brown & Sobsey (2007) found that the geometric mean reduction of E. coli and total coliforms in filtered water was 98% and 94%, respectively. The study also found a 46% reduction of diarrheal disease incidence by CPF users. Additional work has shown E. coli reduction by a mean of 96% and diarrheal disease incidence reductions of 49% during the course of an 18-week field study in Cambodia (Brown et al. 2008). Lastly, a four-month field study in Sri Lanka (Casanova et al. 2013) found widely variable flow rates and concluded that water production is a limiting factor of CPFs; however, this did not seem to be negatively perceived by the users.

The main CPF manufacturing company in Guatemala, which is located near the city of Antigua, has been in operation for more than 20 years, and the company owners report that more than 250,000 filters have been distributed throughout the country through 2015. The filters made by this company are the ICAITI/PFP type, described by Lantagne et al. (2010), and consist of a frustum-shaped (i.e., flower-pot shaped) porous clay filtering unit placed in a plastic bucket equipped with a lid and a spigot. The filtering unit is made of a mixture of clay, sawdust, and water. The mixture is pressed into pot-shaped molds, air-cured, fired in a kiln, and finally coated with colloidal silver. Untreated water is poured into the filtering unit, and treated water is collected in the bucket where it is available for use via the spigot. According to the manufacturer's instructions, the filtering unit should be scrubbed every three months with a brush and treated water, and should be replaced every two years. Most of the CPFs were distributed in the central highlands, the Pacific Coast area, and around Guatemala City. In the Atlantic Coast department of Izabal, the CPFs were first distributed in several rural villages by non-governmental organizations (NGOs) as a part of an emergency response following a major earthquake in May 2009. Although it was reported that the CPFs were well-received by the local families over the short term, this new technology was not permanently adopted by the inhabitants. A second effort to introduce CPFs by several NGOs in the region included follow-up visits to characterize the CPF's performance and to increase consumer acceptance of the technology.

This study describes a two-year CPF field monitoring program started in January 2014 as part of a collaboration between the Missouri University of Science and Technology (Missouri S&T), CPF users from four rural villages of the department of Izabal, Guatemala, and the following local NGOs: Alianza de Derecho Ambiental y Agua (ADA2), Asociación Programas de Gestión Ambiental Local (ASOPROGAL), Asociación Maya Pro Bienestar Rural del Área Sarstun (APROSARSTUN), and Red Cross Santo Tomas de Castilla. The goals of the study were to assess CPF-user acceptance and characterize lifetime in terms of disinfection effectiveness and water volume production under real use conditions.

At the beginning of the study period, families from villages 1 and 2 were already familiar with CPFs and were changing the filtering units according to the manufacturer's instructions, while families from villages 3 and 4 did not have CPF experience. Five families in each village volunteered to participate in the study for a total of 20 families. All the CPFs were made in the manufacturing plant near Antigua, Guatemala, and belong to different production batches. Households in Villages 1, 2, and 4 received CPFs produced in 2014 identified as Production Year 2104 (PY14), and Village 3 received CPFs produced in 2015 (PY15).

The microbiological disinfection effectiveness was annually assessed by Missouri S&T geological engineering undergraduate and graduate students by counting the number of colony forming units (CFU) of E. coli and total coliforms in the raw (influent) and treated (effluent) water. The influent sample was collected from the container used by the individual household to fill the CPF and the effluent sample was collected directly from each CPF's plastic container via the spigot. Samples were analyzed using a Coliscan Plus Easygel Kit (Micrology Laboratories LLC, Goshen, IN, USA) following the manufacturer's instructions. Coliform growth media was mixed with water samples (1, 3, or 5 mL depending on a pretesting estimate of the influent water quality), placed in a Petri dish, sealed, and incubated at 35.5°C for 24 hours. The E. coli and other coliform colonies were counted and reported as CFU/100 mL. Samples were duplicated, and the results were averaged for analytical purposes. Influent and effluent samples were also tested for turbidity using a Hach 2100P portable turbidimeter (Loveland, CO, USA).

The water production capacity of CPFs was characterized approximately every two months through falling head flow rate tests by employees of local NGOs who received training during S&T field visits. The plastic container was emptied, the filtering unit was filled, and after 1 hour the treated water was measured using a 2 L graduated cylinder. In addition, the date, time, and volume of water poured in the filter were recorded on a log sheet by the users.

In order to evaluate the potential adoption and acceptance of the filters, evaluations of the filter performance were obtained by interviewing CPF users during Missouri S&T visits at the beginning and end of the program. Free and informed consent of the participants or their legal representatives was obtained, and the study protocol was approved by the Campus Institutional Review Board at Missouri S&T, MO, USA, on April 3, 2014. A community meeting was initially organized to obtain a general idea of the users' satisfaction, and then the families participating in the study were interviewed individually. The data collection team determined the state of each CPF (that is, were all parts intact and functional) and whether the filter was in current use (was the ceramic filtering unit completely saturated). Then a questionnaire was administered to the primary caregiver for the household who was usually an adult female. Data on basic household demographics, water handling and use, CPF use and cleaning practices, advantages and disadvantages of using the filter, and perceived changes in the family health conditions were collected. All survey instruments were prepared in English and Spanish before use in the study. When necessary, the questions were translated to Q'eqchi’ by a native speaker with experience in community work. The content of the English version of the survey is shown in Table 2.

The total number of monitoring activities conducted in each village is shown in Table 3.

As seen in Figure 2, the CPF use resulted in a reduction of all the measured parameters. E. coli was not found in any of the effluent samples, and there were no coliforms (total) detected in 57% of the effluent samples. In addition, more than 75% of the effluent samples present turbidity levels lower than one nephelometric turbidity unit (NTU).

Inspection of the time series makes evident that the bacterial removal capacity of CPFs decreases over time and the average LRVs follow a linear negative trend with an R² value of 0.938. Brown & Sobsey (2007) found that LRV is a valuable measure of technology performance; however, reduction is a function of influent water characteristics and low LRVs do not necessarily indicate poor performance. Lantagne et al. (2010) established a LRV of bacteria equal to two as the criterion for effective removal when it is not possible to spike samples with bacteria and the water is not contaminated enough to document a higher LRV. The performance requirements for small-scale and household drinking-water treatment defined by WHO (2011) indicate that two is the minimum LRV of bacteria for a technology to be considered protective. The dashed line in Figure 2 at LRV = 2 shows that on average the CPFs used during the study were able to achieve the recommended bacterial reduction during the first 14 months of usage.

According to The Ceramics Manufacturing Working Group (CMWG) (2011), the flow rate test can be used as an indicator of production consistency, pathogen and suspended solids removal efficacy, and water production capacity. Uniform flow rates are expected from a standardized manufacturing process while high flow rates could compromise the quality of treated water and low flow rates may not result in water quantities sufficient for consumer needs. The flow rate test is the most common quality control test performed by the manufacturers, but there are differences between the factory-established acceptable flow rates. Historic work by Rayner et al. (2013) found that the quality control flow rate ranges vary from a minimum of 1 to 3 L h⁻¹ to a maximum of 2 to 5 L h⁻¹. Lantagne et al. (2010) stated that a production process should be considered reliable if the quality control flow rate range at the factory is 1 to 2 L h⁻¹, while CMWG (2011) established an acceptable range between 1 and 3.5 L h⁻¹. The company that manufactured the CPFs used in this study accepts filters with flow rates between 1 and 2 L h⁻¹. The PY14 CPFs initially met the manufacturer's requirements, exceeded the 2 L h⁻¹ threshold around the second month of use, and then returned to the expected range prior to dropping below 1 L h⁻¹ at the end of the study. The PY15 filters presented more consistent values and maintained a flow rate between 1 and 2 L h⁻¹ during the study. On average, the water production of CPFs decreased below 1 L h⁻¹ after approximately 10 months of use.

A total of 25 families with an average of 5.6 people per household participated in the interviews. Other than visually verifying that the CPFs were in use (by examining the brim for complete saturation), there was no mechanism for verifying the veracity of participants' responses. The main water sources were reported to be rainwater collected from rooftops and stored in tanks for 28% of the families, surface water (streams) for 18%, and shallow wells (defined here as ≤10 m in depth) for 54%. All the families participating in the study were using the filter daily at the time of the follow-up and reported filling the filter an average of 2.1 times per day, with a maximum of 4 times per day and a minimum of once a day. According to the manufacturer's instructions, the filtering unit should be cleaned every three months using just-treated water. Most of the families reported to be familiar with the recommended cleaning process, and 20% of them reported cleaning the filtering units using bleach or soap. However, 63% of the families stated that they preferred to clean the filter more frequently than quarterly. According to Salvinelli & Elmore (2015) and van Halem et al. (2009), the water production capacity seems to be the major limiting factor of the CPF's lifetime and sustainability. Nevertheless, 76% of the families reported that the filters produced sufficient drinking water, but it is important to note that 20% of them owned two CPFs and a single CPF would not have produced a sufficient quantity of treated water. A high percentage (88%) of the respondents reported that the use of the filter had positive consequences on the health condition of family members, especially children, by reducing the perceived frequency of disease symptoms including diarrhea, fever, nausea, and vomiting. Most of the positive comments about the CPFs referred to the water quality, which was stated to be clear, fresh, safe, and with a better taste than boiled and chlorinated water. In addition, users who previously drank boiled water were satisfied about saving the time and energy required to gather firewood and prepare a fire. The most common CPF disadvantages were identified as the slow filtration rates and the fact that the CPFs are fragile and can easily break.

Interactions at community meetings also demonstrated satisfaction among the users and the positive impact of the follow-up visits seemed to extend beyond the five volunteer households in each village. In January 2014, 57% of families in village 1 and 63% of families in village 2 changed the filtering units and covered a portion of the costs. Furthermore, in January 2016, these families had achieved up to 3.5 years of continuous CPF use. These results are consistent with Lantagne (2001), who concluded that continued use of the filter was strongly correlated with monthly visits to the household by the local NGO or community leaders.

The daily filling frequencies reported by the users were an average of 2.5 times per day for the PY14 filters and 2.0 times per day for the PY15 filters. As shown in Figure 5, the volume of water predicted by the model (around 20 L day⁻¹ for PY14 filters and 15 L day⁻¹ for PY15 filters) is higher than the actual volume reported by filter users. This finding could possibly be a result of the users not filling the filters at regular time intervals. However, the model suggests that even though the water production capacity depends on the filter flow rate, the amount of available treated water could be significantly increased with frequent and constant filling intervals.

The relatively small number of water quality samples collected during this study and the relatively high detection limits associated with those results mean that the results should be interpreted as general characterizations of CPF performance as opposed to the more definitive results that could be generated by a laboratory study. The field data show that the CPFs used in this study had the ability to provide good quality water by treating highly contaminated waters with total coliform LRVs greater than 2. A negative correlation between filter disinfection efficacy and time in use was observed, and after 14 months disinfection efficacy dropped below the WHO requirements for bacterial removal. This could be a concern because consumers would not readily be aware of the increased risk. However, further studies with a larger number of samples and more accurate microbiological testing procedures should be conducted to confirm these results.

The CPF flow rates were maintained in the recommended range of 1 to 2 L h⁻¹ during the first 10 months of use. The water production capacity was reported to be sufficient for most of the users during the entire 24 months of the study, and modeling results show that the production could be increased by filling the filters more frequently.

In general, the filters were well accepted by users who appreciated the esthetic quality of the treated water, reported lower incidences of health problems especially among children, and expressed their preference for CPFs over other treatments such as boiling or chlorinating drinking water. Unlike previous experiences with CPFs in the region, locals who participated in the field investigation have continued to use the filters for 3.5 years. It is postulated that the presence of an ongoing monitoring program increases the acceptance rate and yields an improvement in use and maintenance practices.

Field data are more difficult to collect from a logistical perspective relative to laboratory data. This study successfully illustrates how a synergistic collaboration between university researchers, local NGOs, and water consumers can generate data that can be used to characterize the real-life performance of CPFs in a field setting. There is potential that this type of collaborative effort could also include governmental organizations and CPF manufacturers in the future, and thus provide even more detailed information regarding the health conditions of target and control groups during future studies.

This work was supported by the Missouri S&T Geological Engineering program. The authors wish to thank the local NGOs ADA2, ASOPROGAL, APROSARSTUN, and Red Cross Santo Tomas de Castilla for their support during the field activities conducted in Guatemala, and the NGOs ASOPROGAL, CISP, and Fondazione SIPEC for the purchase of the CPFs. They also wish to thank the inhabitants of the villages of La Angostura, San Juan, San Francisco del Mar, and Creek Grande for generously participating in the study and sharing their experience and knowledge about CPFs.