Pilot studies for the application of triple media ﬁ ltration at Morton Jaffray Water Works, Harare, Zimbabwe

Lake Chivero, Harare ’ s main source of raw water for drinking water, is algae infested thus affecting water treatment processes at Morton Jaffray Water Treatment Works (MJWTW). Consequently, several challenges have been encountered including frequent ﬁ lter backwashing, leading to a reduced plant output. In this study, the potential of substituting single-media ﬁ lters (currently used at MJWTW) with triple-media ﬁ lters was investigated. This was done using pilot ﬁ lters of a single-media ﬁ lter (SMF) and a triple-media ﬁ lter (TMF) and piloted using clari ﬁ ed water from MJWTW. Electrical conductivity (EC), pH, total dissolved solids (TDS), turbidity, and temperature were the water quality parameters that were monitored. Headloss and ﬂ ow rate were the ﬁ lter operation parameters that were monitored. Both water quality and operational parameters were monitored at hourly intervals until one ﬁ lter reached the maximum allowable headloss or turbidity limit. The variation in ef ﬂ uent turbidity, TDS, EC, and pH between the ﬁ lters was insigni ﬁ cant. Nevertheless, the headloss development for the TMF was signi ﬁ cantly slower than that of the SMF resulting in the SMF reaching the maximum headloss earlier. Therefore, the TMF performed better overall. Therefore, the adoption of TMF could result in a longer ﬁ lter run and improve ﬁ ltration and water production at MJWTW.


INTRODUCTION
Water is an essential element for every living organism proper functioning and plays an important role in health prevention (Lasocka-Gomuła & S ́wietlik 2022).Access to safe drinking water is a basic right for all individuals (Mulamattathil et al. 2014;Libanio 2022;Pace et al. 2022).Globally, the level of poverty can be reduced through the provision of adequate potable water (Feitelson & Chenoweth 2002;Badejo et al. 2015).Access to clean drinking water is a huge challenge globally despite it being an essential component of the sustainable development goals (SDGs) of the United Nations (Alam et al. 2022).Sustainable access to water of sufficient quality is essential for good health (Moe & Rheingans 2006;Inabo & Arshed 2019).Castiglioni et al. (2022) highlighted that water must be of sufficient quantity and quality to protect human health.Therefore, water resources must be kept uncontaminated to ensure that water is clean and safe (Crow & Sultana 2002;Nkwonta et al. 2010).According to Brame et al. (2011), water pollution has been on the rise worldwide in recent years, and the problem is even worse in developing countries.Anthropogenic activities that are mentioned by Said et al. (2022) are one of the major causes of pollution that affects the quality of water resources.Ghernaout et al. (2010) highlighted that the pollution of water sources causes many challenges, including the development of algae, which causes numerous challenges in water treatment, including turbidity.Other challenges caused by algae include the production of taste, odour, and natural toxins (Abouzied & Hassan 2022).According to Magana-Arachchi & Wanigatunge (2020), contaminated water (which is common in developing countries) is linked to the transmission of waterborne diseases such as cholera, dysentery, hepatitis A, and typhoid which are also known to cause death.
Conventional water treatment processes (CWTPs), which are the most widely used technology in developing countries, are now unable to effectively treat water as a result of high levels of pollutants, including algae, as highlighted by Treacy (2019).According to El-Bassuoni et al. (2005), CWTP consists of screening, coagulation, flocculation, sedimentation, filtration, and disinfection.Sedimentation and filtration (the physical processes of CWTP) play an important role in water treatment, as sedimentation plays an intermediate step while filtration is the last step in turbidity removal ( Joh et al. 2011;Rahman & Praseetha 2016).Filtration is a fundamental process for drinking water treatment, being effective towards the removal of colour and microorganisms (Cescon et al. 2016).The use of filtration to improve water quality as a purification measure has been used for many decades (Jaeel & Abdulkathum 2018;Zarezadeh et al. 2018).Bradford et al. (2002) and Braun et al. (2022) highlighted that the filtration process works mainly by physically retaining particles.According to EPA (2006), filtration must be efficient as it serves as a barrier when all other preceding water treatment processes have failed.Joh et al. (2011).Nhongo et al. (2018) stated that poor filtration leads to the carryover of pollutants in potable water, including algae that cause odour and taste in treated water.
Rapid sand filtration is commonly used worldwide in full-scale drinking water treatment plants, mainly to reduce turbidity (Shirakawa et al. 2022).It is the most widely used form of filtration compared to slow sand filtration due to its higher loading rate (120 m 3 /m 2 ·day) than slow sand filters (3-8 m 3 /m 2 ·day) as stated by Davis (2010).Rapid sand filtration is also used for iron (Fe) manganese (Mn) and ammonium (NH 4 þ -N) removal due to its cost-effective, environmentally friendly, and easy-to-operate characteristics (Harun et al. 2022;Yang et al. 2022).According to Berk (2018) and Cescon & Jiang (2020), the most common filters used during water treatment are single-media filters (SMF), although multi-media filters, including triple-media filters (TMF), are considered more effective.Brandt et al. (2017) and García-Ávila et al. (2020) highlighted that the use of multi-media filters improves filter loading rates and extends filter run periods, thus lowering the frequency of backwashing.Other advantages of multi-media filters include a low headloss build-up, large solids storage capacity in the coarser layers and good protection against the breakthrough of impurities in the finer media (Sanyaolu 2010).SUEZ (2021b) highlighted that popular media used in multi-media filters include anthracite, sand, garnet, and magnetite.Sand filters are the most common type of filters in developing countries (Maiyo et al. 2023).Pollution which is a global issue as highlighted by Verma et al. (2017) has been on the rise and sand filters appear to fail to deal with highly polluted water due to reduced filter run (El-Taweel & Ali 2000).Multi-media filters offer an alternative to treat polluted water due to a longer filter run and a large storage capacity of pollutants (Ali et al. 2021).
Multi-media filters have been used in a number of developed countries for water treatment which include England and the United States (TUM .n.d; Zouboulis et al. 2007;Hartshorn et al. 2015;Kazemi Noredinvand et al. 2021).However, the uptake or application of multi-media filters in developing countries including in sub-Saharan Africa especially in areas with drinking water sources having polluted raw water quality has been very low possibly due to limited evidence or awareness on the performance and advantages of multi-media filters.Since the development of the technology, multi-media filters have only been used recently and in a few countries such as South Africa and Ghana as highlighted by Thompson et al. (2016) and PA (2017), respectively.Thus, studies that provide evidence and demonstrate the advantages of multi-media media filters within the sub-Saharan context may promote the uptake of multi-media filters.
Water service delivery is very poor in Harare (Gambe 2015;Pahwaringira et al. 2015;Takavada et al. 2022).This has been the trend for several years as many studies have confirmed this (Manzungu & Mabiza 2004;Manzungu & Machiridza 2005;Gambe 2013;Matamanda et al. 2020).Inadequacies in water treatment processes at Morton Jaffray Water Treatment Works (MJWTW), Harare's main water treatment works, to deal with Lake Chivero's increased pollution have led to water quantity and quality problems in Harare (Dandadzi et al. 2019).As a consequence of the water problems, there have been outbreaks of cholera in Harare linked to potable water shortages and contaminated drinking water (Fernandez et al. 2012;Chirisa et al. 2015;WHO 2018;Winstead et al. 2020).Harare's main source of drinking water, Lake Chivero, is reported to be eutrophic (Utete et al. 2018;Mwaisowa et al. 2023).This has resulted in the lake being algae infested, as highlighted by many studies (Mhlanga et al. 2006;Zengeya & Marshall 2007;Chawira et al. 2013;Nyarumbu & Magadza 2016;Makaya 2022).Hoko et al. (2021b) indicated that excessive amounts of algae in raw water have seriously impacted water treatment at MJWTW.Thus, there has been an increase in the backwash frequency of 4-8 h daily, as opposed to the recommended 24-48 h (Hoko & Makado 2011).As such, plant output has declined as filters are operated in a discontinuous manner due to the need to backwash as highlighted by Dandadzi et al. (2019).This has partly contributed to Harare's potable water shortages (Manzungu & Chioreso 2012;Pahwaringira et al. 2015;Takavada et al. 2022).Water shortages in Harare have been reported to be associated with cholera outbreaks in the city (Kone-Coulibaly et al. 2010;Chirisa et al. 2015;WHO 2018).Therefore, there is a need to investigate options for improving water treatment efficiency and also increase plant output at MJWTW including alternative filter types to ensure that the residents of Harare receive adequate water.This study was conducted during the winter period and compared the performance of an SMF to a TMF in a laboratory.Water quality parameters including electrical conductivity (EC), pH, total dissolved solids (TDS), turbidity, and temperature and filter headloss development were used to assess filter performance.

Location of the study area
The study was carried out in a laboratory at the University of Zimbabwe, Faculty of Engineering and the Built Environment, Department of Construction and Civil Engineering.Clarified water from MJWTW, the main water treatment plant that supplies Harare with drinking water, as stated by Hoko et al. (2021b), was used for the experiments.Figure 1 shows the location of MJWTW.Nhapi (2009) highlighted that Harare is the capital city of Zimbabwe and according to ZIMSTAT (2022), Harare urban has an estimated population of 1,491,740.Harare is the largest city and capital of Zimbabwe (Tsiko & Togarepi 2012;Bandauko et al. 2018).Thus, it is a city of great significance for Zimbabwe.

Background on MJWTW
The Water and Sanitation Department of the City of Harare (CoH) is responsible for the water supply to the City of Harare (Hamudi 2022).The Water and Sanitation Department of the CoH is also responsible for water supply to the surrounding Water Supply Vol 00 No 0, 3 towns of Harare and is responsible for water treatment at MJWTW (Ndunguru & Hoko 2016).According to Dandadzi et al. (2019), MJWTW abstracts raw water from Lake Chivero and Lake Manyame.Lake Chivero is capable of supplying approximately 421,000 m 3 /day to MJWTW without drawing down the Lake (Nhapi & Hoko 2004), thus Lake Chivero is the main water source.Muisa et al. (2011) stated that MJWTW has three water treatment units, Unit 1, Unit 2, and Unit 3 with treatment capacities of 160,000, 227,000, and 227,000 m 3 /day, respectively.CWTP at MJWTW includes aeration, coagulation, flocculation, sedimentation, filtration, disinfection, and stabilization by adding lime, as highlighted by Hoko et al. (2021a).Figure 2 shows a schematic of MJWTW.
According to Nhongo et al. (2018) and Hoko et al. (2021a), MJWTW has challenges in treating water to an acceptable standard due to the high level of pollution of Lake Chivero which has resulted in high chemical demand and frequent filter backwashing.Also, there has been filter clogging due to the high amount of algae in raw water that leading to poor quality treated water, as highlighted by Hoko et al. (2021b).Furthermore, there has been a carryover of algae and organics from MJWTW that has resulted in algae regrowth and the formation of trihalomethanes (THMs) as reported by Nhongo et al.Water Supply Vol 00 No 0, 4 (2018) and Dandadzi (2019).The poorly treated water in Harare has also been reported to have taste and odour problems, triggering user complaints and rejection, according to a study by Dandadzi et al. (2019).Dandadzi et al. (2020) highlighted that the pH, turbidity, free residual chlorine and algae in the finished water from MJWTW ranged from 6.95 to 7.07, 0.79 to 1.24, 0.05 to 0.49 mg/L, and 45-232 cells/mL, respectively.Dandadzi et al. (2019) stated that water collected in the finished water from MJWTW had an average pH of 6.95, free residual chlorine ranging from 0.08 to 0.60 mg/L, chlorophyll a concentration ranging from 0.13 to 0.30 μg/L.According to Nhongo et al. (2018), the total solids (TS), turbidity, COD, free residual chlorine, and THMs in the final treated water from MJWTW ranged from 276 to 1,193 mg/L, 0.83 to 4.34 NTU; 1.56 to 6.04 mg/L; 0.05 to 0.30 mg/L; and 7.60 to 24.00 μg/L.

Experimental design
The filtration experiments were carried out in May during the winter season that is critical for filtration.According to Ramli & Bolong (2016) and Brandt et al. (2017), the viscosity of water is inversely proportional to temperature; therefore, when temperatures are low, water becomes more viscous, resulting in greater filtration resistance.Figure 3 shows the experimental setup consisting of two pilot filters made of 2 m high perspex cylindrical columns with an internal diameter of 0.2 m.The diameter of the filter columns was comparable to the 0.15 m diameter filters used by Davies & Wheatley (2012) in a similar study.The experimental setup also had a backwash tank which was used for cleaning the filters by pumping clean water through the filter as recommended by Davis (2010).
Filter media should have good hydraulic properties and a high surface area to retain organic matter and other contaminants (Hägg and Pott 2022).Granular filter media used in water treatment are usually selected on the basis of size, but filter bed behaviour is also affected by the density and voidage of the media, particularly during backwashing (Fitzpatrick 1998).The filter media used in this study included garnet, silica sand, and anthracite, as recommended by Tchobanoglous et al. (2004).The filter bed was 900 mm deep and was supported by a 150 mm gravel layer which was comparable in height to the 200 mm used by Templeton et al. (2007) in a similar study.Table 1 shows the properties of the filter media used in the two pilot filters.

Selection of filter operational parameters
The filtration rates for the SMF and TMF range from about 4 to 7.2 m 3 /m 2 ·h and 7.2 to 12 m 3 /m 2 ·h, respectively (Tchobanoglous et al. 2004;Cescon & Jiang 2020;WHO 2021).For this study, a filtration rate of 7.5 m 3 /m 2 ·h was selected to match the available pump with capacity to ensure a filtration rate in the range for both filter types.According to WHO (2011) and Elfaki et al. (2015), the headloss of a filter is one of the most important operational parameters and is regarded as the most important constraint when terminating filtration to allow backwash.A maximum headloss of 2.4 m and 2.4-3.0 m per filter run has been recommended by Camp (1961) and Davis (2010), respectively.For this study, the maximum headloss was set at 2.4 m.The supernatant water (depth of water above the filter media) ranges from about 1.2 m according to SUEZ (2021a) to 2.0 m according to Smethurst (1988).For this study, 1.5 m was chosen.

Operation of pilot filters
The clarified effluent, which is water collected from the sedimentation basins just before the rapid sand filters at MJWTW, was stored in four 5,000 L plastic tanks.These tanks served as the sump for the pump that fed the two filters.A 0.5 Horsepower centrifugal pump operating at a constant flow rate of 0.236 m 3 /h was used to achieve a constant surface loading rate of 7.5 m 3 /m 2 ·h to the filters.The inlet to the filters was 1.5 m above the top of the filter bed to achieve a level of supernatant water of 1.5 m.Both filters were run concurrently.The filters were run until one of the filters reached the maximum head loss of 2.4 m suggested by Camp (1961) or exceeded the SAZ (1997) turbidity limit of 1 NTU.The experiment was carried out for two filter runs for each filter column.

Selection of water quality parameters
Parameters that were measured in this study include EC, pH, TDS, turbidity, and temperature.According to Sawyer et al. (2003) and Rusydi (2018), EC is used to evaluate the performance of water treatment processes, as it is used to determine the residue of chemicals used in water treatment and is also an indirect measure of TDS.The pH was considered as a study parameter as it affects the solubility of heavy metals in water, thus affecting the level of TDS and therefore, the filtration process (Zhang et al. 2018).Sawyer et al. (2003) also stated that the pH of the water after filtration has an impact on the disinfection process, particularly when chlorine is used as the disinfectant.Chlorine is used as a disinfectant at MJWTW (Hoko & Makado 2011;Nhongo et al. 2018).The turbidity of water is an essential parameter that affects filtration, the aesthetics of water, and chlorine demand, as highlighted by Sawyer et al. (2003).The temperature of the water was also monitored hourly as it affects the viscosity of the water, affecting filtration as highlighted by Ramli & Bolong (2016).

Methods of data collection
Grab samples of clarified feed water collected from MJWTW (influent) and outlet (effluent) points from each filter column were collected at hourly intervals for water quality tests.The headloss values were read from manometers which were installed on multiple positions along each of the filter columns.

Methods of water quality analysis
Water samples were analysed at the University of Zimbabwe, Department of Construction and Civil Engineering Water Laboratory.The analysis was carried out according to the standard methods suggested by APHA (2012) for all parameters that were studied.Table 2 shows the summary of the analytical methods, the method number, and the brand of equipment used to measure each parameter.

Methods of data analysis
The t-test was used to compare the filter influent and effluent water quality to establish the efficiency and significance of the differences between the performance of the filters as recommended by Xu et al. (2017).The Mann-Kendall test was also used to test for the temporal trend of headloss during filtration as Hussain and Mahmud (2019) highlighted that it is a widely used non-parametric test to detect significant trends in a time series.

RESULTS
Table 3 shows the summary of the results of the filter performance regarding water quality parameters for the SMF and TMF.

EC and TDS variation during filter runs
Figure 4 shows the variation of the influent and effluent EC results for the SMF and TMF.The average SMF effluent and the TMF effluent were comparable as they were 326.3 and 329.5 μS/cm, respectively (Table 3).The box and whisker plots and the coefficient of variation (CV) results in Table 3 show a low spread of the effluent EC values during the experiments.The variation of TDS for the influent and effluent for the SMF and TMF is shown in Figure 5.The SMF effluent and the TMF effluent TDS values were comparable as they were 119.2 and 118.7 mg/L, respectively.There was a low variation in the effluent TDS for the SMF and the TMF as they had a CV of 0.9 and 0.7%, respectively.

pH variation during filter runs
The variation of the influent and effluent pH results for the SMF and TMF are shown in Figure 6.The average pH results for the effluent of the SMF and the TMF were comparable (7.7 and 7.6, respectively).There was a low CV of pH in the effluent of the SMF and TMF (1.9 and 1.4%, respectively) which is also seen on the box and whisker plots.

Turbidity variation during filter runs
Figure 7 shows the variation of the influent and effluent turbidity results for the SMF and TMF.The average turbidity of the SMF and TMF effluent were 1.0 and 0.9 NTU, respectively.There was high variability of the effluent turbidity for SMF and TMF (CV of 51.0 and 74.0%, respectively) showing that the filters removed colloidal matter.The variability can also be seen on the box and whisker plots.

Water treatment efficiency of filters
The SMF and TMF performed in a comparable manner regarding EC removal (6.1 and 5.2%, respectively) and TDS removal efficiencies (3.6 and 4.0%, respectively).Based on a t-test at a 95% confidence interval, the difference between the influent  Water Supply Vol 00 No 0, 9 Uncorrected Proof Downloaded from http://iwaponline.com/ws/article-pdf/doi/10.2166/ws.2024.086/1404498/ws2024086.pdf by guest and effluent EC and TDS values for the SMF and the TMF was not statistically significant (p .0.05).A similar performance for EC and TDS removal was expected, as Rusydi (2018) stated that EC is an indirect measure of TDS.The influent EC (for both filters) was within the recommended 700.0 μS/cm limit highlighted by SAZ (1997).As a result of the low influent EC, the effluent EC was also below the recommended limit for EC for both filters.WHO (2011) stated that potable water should have a TDS lower than 600 mg/L.The TDS of the influent was already below the recommended limit, and as such the TDS values of the effluent for both filters remained below the recommended limit.The TDS removal efficiency of the TMF was comparable to the 2.7% removal efficiency obtained by Seidmohammadi et al. (2021).However, this was not similar to what Reddy et al. (2021) found that triple-media filters removed metals more efficiently (above 60%) which are said by Rusydi (2018) to be linked to EC.The difference is likely to have been caused by the synthetic stormwater (SSW) solution prepared in a laboratory  with a high concentration of metals that was used in the study by Reddy et al. (2021).The results found in this study were similar to those of Samuel et al. (2012) and Kumar et al. (2012) established that sand filters showed a lower capacity for the retention of contaminants than multi-media filters.Thus, both the SMF and TMF practically performed the same in terms of EC and TDS removal.
The average pH of the SMF effluent increased by 0.1% whilst that of the TMF decreased by 1.0%.Based on a t-test at a 95% confidence interval, the difference between influent pH and effluent pH for both SMF and TMF effluent pH was not statistically significant (p .0.05).The small change in effluent pH of the SMF was comparable to the 0.8% obtained by Gülay et al. (2014) for an SMF whilst the decrease in effluent pH for TMF was comparable to the 2.6% obtained by Chowdhury & Chowdhury (2021) for a TMF.There was a fluctuation in the pH of the influent and effluent from both filters during the filter runs.The fluctuation in pH could have been caused by algae, as studies by Hoko & Makado (2011) and Dandadzi et al. (2019) have reported algae in the raw and treated water of Lake Chivero from MJWTW.The pH of water influences the effectiveness of chlorine disinfection, and disinfection by chlorine gas is more effective at low pH values of less than 8 (WHO 2011).Disinfection follows filtration at MJWTW; therefore, the effluent pH for the SMF and TMF (7.7 and 7.6, respectively) was generally within the recommended pH range for chlorine disinfection.Therefore, both filters did not alter pH that much and performed the same.
The effluent turbidity of both filters continued to decrease and became almost the same after 1 h.After 7 h, SMF effluent turbidity started to increase while the TMF effluent turbidity continued to show a general decrease.After 10 h, the turbidity of the effluent from the SMF was higher than that of the TMF and also above 1 NTU recommended for potable water treated by SAZ (1997).Based on a t-test at a 95% confidence interval, the difference between influent and effluent turbidity for both the SMF and TMF was not statistically significant (p .0.05).This was different from what Noredinvand et al. (2021) established that the difference between the influent and outlet turbidity figures for both SMF and TMF was significant (p , 0.05).Over the filtration period, the TMF showed lower turbidity values than the SMF especially after ripening.Davis (2010) highlighted that ripening occurs when the turbidity of the filter effluent stops decreasing and the turbidity becomes essentially constant with no significant change.The same source stated that the peak is due to residual backwash water being flushed from the media, and from particles in the influent water that are too small to be captured.Although the difference in performance between the two filters may not have been statistically significant (p .0.05), the SMF showed a progressive increase in turbidity after the seventh hour and could therefore lead to a shorter filter run.The effluent turbidity was generally below 1 NTU as recommended by SAZ (1997) after both filters had ripened.However, the effluent turbidity from the SMF went above 1 NTU after the 10th hour.The SMF and the TMF had comparable overall turbidity removal efficiencies (63 and 66%, respectively), although the TMF removal efficiency was slightly higher.This was similar to what Ncube et al. (2018) obtained, that TMFs remove more colloidal particles compared to SMFs.The overall turbidity removal efficiency of the TMF obtained in this current study was less than the 81.3% obtained by Chowdhury & Chowdhury (2021).This was likely to have been due to the fact that clarified effluent was used in this study thus some of the colloidal matter will have been removed during the sedimentation stage.Although both filters had comparable turbidity removal efficiencies, the SMF had a shorter ripening period compared to the TMF, resulting in the effluent turbidity for the SMF being lower than that of the TMF at the beginning of the filtration process.This could be attributed to the large pore spaces in the TMF filter media as compared to sand in the SMF since the media gradation for TMF has more pore volume at the top of the filter and gradually decreases to a minimum at the bottom of the filter (Hamoda et al. 2004).Therefore, the TMF is likely to clog later than the SMF, resulting in the TMF having a longer filter run.However, although there was no statistical difference in performance between both filters, the TMF performed better than the SMF regarding turbidity removal.
Overall, the variation in effluent turbidity, TDS, EC, and pH between the SMF and the TMF was statistically insignificant.However, the TMF performed slightly better in terms of turbidity removal, which could result in the TMF having a longer filter run.

Headloss development in filters
The SMF's headloss reached the 2.4 m limit suggested by Camp (1961) in 13 h, while the TMF's did not reach the 2.4 m limit during the experiment and was only around 1.36 m.Based on a t-test at a 95% confidence interval, the difference between headloss for the SMF and the TMF was statistically different (p , 0.05).The headloss in both filters increased with increasing filter run time.This was expected as Davis (2010) stated that filter resistance increases with time due to the accumulation of suspended matter in the interstices of the filter medium, resulting in increased headloss.Hu et al. (2022) stated that the removal of suspended particles by filter media is mainly by sedimentation and interception and that the headloss is affected primarily by the interception of suspended particles and by the media structure.There was an increase in the headloss buildup for both the SMF and the TMF, as depicted by positive Kendall tau values.Also, the increasing trend was statistically significant (p , 0.05) for both filters.However, the total headloss development in the SMF was higher than that in the TMF.This is due to the fact that the SMF has uniform filter media with uniform porosity, which causes the top layer of the bed to be used more than the rest of the bed, resulting in rapid clogging and shorter filer runs (Templeton et al. 2007;Michel et al. 2020).On the other hand, the TMF has the advantage of filtration from coarse to fine, resulting in an even distribution of solids in the filter bed during filtration, allowing for a longer filter run and a lower headloss development (Cescon & Jiang 2020).Ncube et al. (2018) also obtained similar results, as there was a higher headloss development in the SMF compared to a TMF.The TMF also showed a lower rate of headloss development, which could result in a longer filter run time than the SMF.Thus, the rate of headloss development in the SMF was much higher compared to the TMF implying that the TMF performed better and had a potential of a longer filter run and higher water production.

CONCLUSIONS AND RECOMMENDATIONS Conclusions
The SMF and the TMF performed similarly in terms of treatment based on the selected parameters [TDS, EC, pH, and turbidity].Both filters reduced the turbidity to the requirement of 1 NTU of the Standards Association of Zimbabwe (SAZ).When compared to each other, the TMF performed slightly better than the SMF in terms of turbidity removal.However, the rate of headloss build-up in the SMF was higher than in the TMF.The SMF reached the headloss earlier than the TMF resulting in the SMF having a shorter filter run time as compared to the TMF.Overall, the TMF performed better than the SMF regarding water treatment and also headloss build-up.Therefore, the adoption of TMF at MJWTW may result in longer filter runs, an improvement in filtration efficiency, and ultimately increased plant productivity.

Recommendations
It is recommended that the SMF be replaced by the TMF at MJWTW to improve filtration efficiency and increase plant output.Further research should look into other filtration alternatives including the use of nanotechnology and alternative filter media at Morton Jaffray Water Treatment Works.

Figure 1 |
Figure 1 | Location of the study area (adopted from Hoko et al. 2021b).

Figure 2 |
Figure 2 | Flow schematic of Morton Jaffray Water Treatment Works (adopted from Hoko et al. 2021b).

Figure 3 |
Figure 3 | Laboratory set up of the two pilot filters (all dimensions are in mm).

Figure 4 |
Figure 4 | Influent and effluent EC variation: (a) influent and effluent EC for SMF vs TMF and (b) effluent EC for SMF vs TMF.

Figure 5 |
Figure 5 | Influent and effluent TDS variation: (a) influent and effluent TDS for SMF vs TMF and (b) Effluent TDS for SMF vs TMF.

Figure 9
Figure 9 depicts the trend in headloss development for the SMF and TMF.The SMF and TMF total headloss ranged from 0 to 240.7 cm and 0 to 135.5 cm, respectively, during the experiments.

Figure 6 |
Figure 6 | Influent and effluent pH variation: (a) Influent and effluent pH for SMF vs TMF and (b) effluent pH for SMF vs TMF.

Figure 7 |
Figure 7 | Influent and effluent turbidity variation: (a) influent and effluent turbidity for SMF vs TMF and (b) effluent turbidity for SMF vs TMF.

Figure 8 |
Figure 8 | Influent and effluent temperature: (a) influent and effluent temperature for SMF vs TMF and (b) effluent temperature for SMF vs TMF.

Table 1 |
Properties of the filter media used in the pilot filters

Table 2 |
Methods and equipment used for water quality analysis

Table 3 |
Summary of influent and effluent parameters for the SMF and TMF