Removal of river water turbidity and total dissolved solids using natural coagulants derived from banana peel and Moringa stenopetala seed

The ability of water to natural purification and purifying itself through sedimentation and flocculation processes, which allow pollutants to settle out, is an existing trend (Kazi et al. 2013). However, this natural process is insufficient when harmful contaminants are present in excessive amounts (Debora et al. 2013). Such water is called non-potable water and is unsuitable for drinking due to being a cause of death for human in many countries (Choubey et al. 2012). To ensure the quality of drinking water, the World Health Organization (WHO) provides recommended values for physical, chemical, and biological parameters (Choubey et al. 2012). However, inadequate sanitation and polluted water are still responsible for up to 80% of all diseases worldwide (Janna 2016). Despite two-thirds of the Earth being covered with water (Rajasulochana 2016), the availability of clean water remains uncertain due to pollution and inadequate treatment (Kazi et al. 2013).

Total dissolved solids (TDS) are an important indicator of water quality as they measure the amount of dissolved inorganic and organic substances present in water. These substances can include minerals, salts, and other compounds that can impact the taste, odor, and overall quality of water (Town et al. 2013; U.S. EPA 2019). High TDS levels can also cause scaling and corrosion in pipes and equipment, leading to increased maintenance costs and decreased efficiency (U.S. EPA 2019). To address TDS levels in water, various treatment processes such as reverse osmosis, distillation, or ion exchange are commonly used. However, these methods can be costly and energy-intensive. Natural coagulants have been studied as a potential low-cost and environmentally friendly solution for TDS removal (Balamurugan & Shunmugapriya 2019). Turbidity is a crucial factor in determining the quality of water for human consumption. It measures the clarity of water and indicates the presence of suspended particles like sediment, algae, and other organic and inorganic matter (Kazi et al. 2013). High levels of turbidity can negatively impact the safety and quality of drinking water by reducing the effectiveness of disinfection and providing a breeding ground for microorganisms that can cause diseases (U.S. EPA 2019). The high turbidity of water is due to the presence of colloidal materials, which can absorb harmful chemicals and cause unpleasant tastes and odors (Kazi et al. 2013).

The coagulation–flocculation process is a widely used method for producing potable water and treating wastewater (Debora et al. 2013). Various factors, including pH, coagulant dosage, stirring speed, and settling time, can influence the efficiency of this process. Coagulants can be either natural or chemical in nature (Janna 2016). Chemical coagulants, such as aluminum and ferric salts, have been used for water treatment for centuries (Kazi et al. 2013; Ghernaout 2020). However, these chemical coagulants can release harmful substances into the environment, negatively impacting human health (Choubey et al. 2012; Diver et al. 2023). They are also expensive, produce large amounts of sludge, significantly affect the pH of treated water, and can cause diseases like Alzheimer's (Thakur & Choubey 2014). Moreover, advanced oxidation processes (AOPs) are a viable solution to address environmental concerns caused by organic materials that cannot be treated conventionally (Liu et al. 2019; Gallo-cordova et al. 2021). Different natural coagulants are believed to provide a sustainable and cost-effective solution for water treatment in rural areas, where access to clean water is often limited (Koul et al. 2022). Natural coagulants derived from plant sources are a more sustainable alternative (Saritha et al. 2017). While the effectiveness of these natural coagulants in treating river water has been investigated, there is still a significant gap in exploring the operating conditions to optimize the river water treatment process, the interaction effects of the factors, and the understanding of the morphological and chemical composition of the coagulants themselves. According to Liu et al. (2019), four-factor, five-level central composite design (CCD) was chosen to optimize degradation conditions using ozone-based AOP.

So, it is necessary to develop inexpensive and efficient water treatment accesses that are derived from natural coagulant materials that are readily available in large quantities and are economically feasible. In that context, there is a possibility of using natural coagulants derived from banana peel and Moringa stenopetala seed, which are abundantly available in the southern part of Ethiopia, at a very low cost.

This study aimed to investigate the coagulation capacity of two natural coagulants, namely banana peel powder and M. stenopetala seed powder, in the removal of TDS and turbidity of river water.

Except for the raw banana peel and M. stenopetala seed, all the reagents used in this analysis were analytical grade.

The functional groups involved in the coagulation process were identified by Fourier transform infrared (FTIR) analysis (spectrum 65 FTIR, Perkin Elmer) in the range 4,000–400 cm⁻¹ using KBr pellets. Origin Pro 2023 software (Version 10.0.5) was used to plot and smooth the graphs of absorbance against wavelength for every sample.

An FTIR chart (Nandiyanto et al. 2019) was used to determine the functional groups present in the respective samples through the identification of the IR absorption peaks at different wavelengths on the graphs. The surface morphology of banana peel powder and M. stenopetala seed powder (before and after coagulation) was investigated using scanning electron microscopy (SEM) images with different magnifications obtained from a Quanta 200 scanning electron microscope (FEI Company, USA).

In this study, portable pH meter (model pH3310), digital turbidity meter (LA-34) Digital Nephelometer, Digital TDS Meter (SI-187), jar test apparatus (LCD display digital flocculator), stirrers and 1,000-ml beakers, oven, and mass balance were used. The samples that were gathered were moved to the laboratory while adhering to the usual operating procedures using the APHA standard (APHA 1998).

The pH of raw water samples and coagulated water was determined using the portable pH meter (model pH3310). Calibration of the pH meter probes was performed using standard solutions. The probes were then carefully inserted into the samples, ensuring complete submersion and contact with the sensing edge. The pH values were recorded once the meter's display showed stable readings. This procedure allowed for the accurate measurement of pH levels in the water samples.

In order to assess the turbidity of both raw water samples and coagulated water after the coagulation experiments, a digital turbidity meter (LA-34) Digital Nephelometer was employed. The meter was calibrated using distilled water and a formazine standard solution to ensure accurate readings. The raw water samples and coagulated water were then measured for turbidity using the calibrated meter. The measurements were taken when the display on the meter reached a stable state, ensuring reliable and consistent results.

TDS measurements were conducted using a digital TDS meter, specifically the Digital TDS Meter (SI-187). The meter was calibrated using a standard solution. Both raw water samples and coagulated water after the coagulation experiments were then measured for TDS using the calibrated meter. The TDS values were recorded once the meter's display became stable.

The experimental design for this study was implemented using Design Expert software version 13.0.5.0, which utilizes statistical and mathematical methods to model the relationship between the input variables and the output variables, which were considered factors and responses, respectively. To assess and optimize the interactions among the independent variables (pH, coagulant dosage, stirring speed, and settling time) and their effects on the dependent variables (response) TDS and turbidity removals, a response surface methodology (RSM)-based face-centered CCD was employed. The RSM approach allowed for obtaining maximum information with the minimum number of experiments, as it reduced the required number of runs (Nor & Wan 2020). The levels of the experimental factors for the independent variables are presented in Table 1.

Here, k represents the number of factors, and n represents the number of center points. In this study, there were four independent variables, so a 2⁴ full factorial CCD was utilized. It consisted of 16 factorial points, 8 axial points, and 6 replicates at the center points, resulting in a total of 30 experiments.

To evaluate the adequacy of the model and the effects of the input parameters on the response variable, an analysis of variance (ANOVA) was conducted. A statistical evaluation of the p-value and F-value of the regression coefficient at a 95% confidence interval was performed. Additionally, the coefficient of determination (R²), adjusted coefficient of determination (R² adj), adequate precision (AP), and coefficient of variation (CV) were used to assess the quality of fit of the developed model. Furthermore, 3D response surface plots were generated to visualize the interaction between the independent factors and their respective effects on the response variable.

The TDS and turbidity removal efficiency of natural coagulants can be affected by several factors. Some coagulants choose specific pH conditions, while others work in an extensive pH range. This kind of challenge is also very common with stirring speed, settling time, and coagulant doses for different coagulant types. Therefore, it is imperative to assess the optimum conditions for the TDS and turbidity removal efficiency of banana peel powder and M. stenopetala seed powder coagulant. In this regard, different dose levels 0.3, 0.65, and 1 g/L; different pH levels such as 3, 6.5, and 10; different settling times 20, 40, and 60 min; and stirring speeds including 30, 60, and 90 rpm are used for optimization purposes.

As shown in Figure 8, banana peel coagulant shows a continuous removal of TDS and turbidity with increases in coagulant doses up to 0.65 g/L. At this dosage, a maximum TDS and turbidity removal efficacy of 66.8 and 82.06%, respectively, were found. This increment in removal of TDS and turbidity is due to an increase in the active site of the banana peel powder, and also negatively charged colloidal particles are adsorbed onto positively charged functional groups of the natural polymer, which causes particles in the river water to be destabilized and flocculated. Then, the removal efficiency of TDS and turbidity decreased further, increasing from 0.65 to 1 g/L, respectively, which might be attributed to the over dosage of the flocculants in the river water sample that led to electrostatic repulsive forces and poor removal efficiency.

The study on M. stenopetala seed powder found that as the dosage of the coagulant increased, the clarity of the water also increased up to 0.65 g/L. This improvement in clarity may be due to the poly-cationic nature of the coagulant, which induces a charge neutralization mechanism, causing destabilization and flocculation of the negatively charged colloidal particles. At a dosage of 0.65 g/L, the study found a maximum TDS and turbidity removal efficiency of 69.72 and 93.42%, respectively (Figure 9).

From the experimental results of this study, the increment from a lower dose to a higher dose increased the TDS and turbidity removal efficiency of both coagulants up to the 0.6 g/L dosage. This is due to the increase in the coagulant's active site (Mohammed & Shakir 2018). But further dosage increments show a decrease in water quality. This may be due to the fact that when over-dosage occurs, the water has the color of Moringa powder and turbid.

The pH of the sample water was 5.8, but for this experimental study, it was adjusted to three ranges: 3, 6.5, and 10 in order to check the coagulant effectiveness in acidic, near-neutral, and basic conditions. So as the experimental result shows, when treating river water using banana peel coagulant, coagulant turbidity and TDS removal increased up to the near-neutral state and decreased at the basic condition.

As per the experimental details of this study, stirring speed plays a very important role in the coagulation and flocculation processes of the river water treatment. So since this study tries to check the effect of stirring speed on the coagulation process at 30, 60, and 90 rpm, at 30 rpm some suspended small flocs were dispersed on the full surface of the sample water in the beaker and took time for sedimentation at this slow stirring speed. But at 60 rpm, the stirrer was highly rotated, and suspended small flocs were collected at the center of the sample water surface and showed the formation of an agglomeration of those small flocs, which quickly and easily settled.

Response surface methodology was employed to determine the association between the dependent variable and experimental factors. In this model, 30 experiments were carried out as shown in the Supplementary Appendices. The quadratic model was suggested to associate the response variable with the four independent variables. Because there is a close agreement between adjusted R² and predicted R² values, it has the highest coefficient of determination (R²) approaching 1, and it also has the lowest standard deviation and p-value (Dawood et al. 2013).

As shownin Tables 2 and 3, the Model F-value of 36.50 and 21.79 implies the model is significant. There is only a 0.01% chance that an F-value this large could occur due to noise. The lack of fit F-values of 3.02 and 0.78, respectively, for TDS and turbidity removal imply that the lack of fit is not significant relative to the pure error. There is an 11.71 and 65.31% chance that a lack of fit F-value this large could occur due to noise. A non-significant lack of fit is good.

The significance of model terms is determined by their p-values. p-values less than 0.05 indicate that the model terms are significant, while values greater than 0.1 indicate that they are not significant. In the context of TDS and turbidity removal, the significant model terms include A, B, C, D, AB, AC, AD, BC, CD, A², B², and C² for TDS removal and A, B, C, D, BC, CD, and B² for turbidity removal. If there are many insignificant model terms, reducing the model may improve its effectiveness.

In the case of M. stenopetata seed coagulants, the R² values for TDS and turbidity removal were 0.9715 and 0.9531, respectively, indicating a high correlation between the response's actual and predicted values and a good fit of the model (Sharma et al. 2009). The summary of fit statistics shown in Table 4 demonstrated that the adjusted R² values, which are 0.9449 and 0.9094, are closer to the predicted R², and the difference is less than 0.2, which indicates that the experimental data is considered satisfactory (Sharma et al. 2009).

Furthermore, the values of the CV that measure the reproducibility of the model are 4.85 and 1.93% for TDS and turbidity removals, respectively. A CV value of less than 10% is considered appropriate for the reproducibility of any model (Salehi et al. 2010). The signal-to-noise ratio of the adequate precision measures for the response models was 23.4022 and 19.5762 for TDS and turbidity removals, respectively. The adequate precision value higher than 4 is appropriate and shows that the regression model equation can be employed within the range of factors in the design space (Dawood et al. 2013). These two indices, CV and adequate precision, suggested that the models are reproducible and have a high degree of reliability and accuracy in the experiments.

However, the p-values (0.8845) in Table 3 and (0.3467, 0.5033, 0.9328, and 0.1656) in Table 4 indicate that the interaction effect between settling time and pH on TDS removal, dose and settling time, dose and stirring speed, dose and pH, as well as pH and settling time, has an insignificant impact on the turbidity removal efficiency of M. stenopetala seed powder. The p-value of lack of fit exceeding 0.05 signifies an insignificant p-value and the ability of the model to fit the experimental data accurately (Kalsido et al. 2021). The 3D plot showing the interaction effect of the operating parameters is indicated in Figures 17–20.

As shown in Tables 5 and 6, the Model F-value of 26.14 and 21.91 implies the model is significant. There is only a 0.01% chance that an F-value this large could occur due to noise. The lack of fit F-values of 2.75 and 0.80, respectively, for TDS and turbidity removal implies that the lack of fit is not significant relative to the pure error. There is a 13.78 and 64.21% chance that a lack of fit F-value this large could occur due to noise. A non-significant lack of fit is good.

The significance of model terms is determined by their p-values. p-values less than 0.05 indicate that model terms are significant, while values greater than 0.1 indicate that they are not significant. In the context of TDS and turbidity removal, the significant model terms include A, B, C, D, AB, AC, AD, CD, B², C², and D² for TDS removal and A, B, C, D, BD, and A² for turbidity removal. If there are many insignificant model terms, reducing the model may improve its effectiveness.

In the case of banana peel coagulants, the R² values for TDS and turbidity removal were 0.9606 and 0.9534, respectively, indicating a high correlation between the response's actual and predicted values and a good fit of the model (Sharma et al. 2009). The summary of fit statistics shown in Table 7 demonstrated that the adjusted R² value, which is 0.9239 and 0.9099, is closer to the predicted R², and the difference is less than 0.2, which implies that the experimental data is considered satisfactory (Wantala et al. 2012).

Additionally, the values of the CV that measure the reproducibility of the model are 4.64 and 2.22% for TDS and turbidity removals, respectively. A CV value of less than 10% is considered appropriate for the reproducibility of any model (Salehi et al. 2010). The signal-to-noise ratio of the adequate precision measures for the response models was 22.5021 and 21.0647 for TDS and turbidity removals, respectively. The adequate precision value higher than 4 is appropriate and shows that the regression model equation can be employed within the range of factors in the design space (Dawood et al. 2013).

However, the p-values (0.8370 and 0.4482) in Table 5 and (0.4214, 0.2042, 0.3207, and 0.1140) in Table 6 indicate that the interaction effect between dose and stirring speed and also dose and settling time on TDS removal, pH and dose, pH and stirring speed, pH and settling time, as well as dose and stirring speed, has an insignificant impact on the turbidity removal efficiency of banana peel powder. The probability value of lack of fit exceeding 0.05 signifies an insignificant p-value and the ability of the model to fit the experimental data accurately (Kalsido et al. 2021).

Numerical optimization allows the selection of a desirable value in the form of a range, target, minimum, or maximum value for each input variable factor and response. A desirability function is a technical approach in a statistical design that simultaneously measures the optimal settings of input parameters to produce optimum performance levels of one or more output variables (Mourabet et al. 2017). In this study, the input variables were given specific ranged values, whereas the output variables were designed to achieve a maximum, as shown in Tables 8 and 9.

In order to verify the optimization results, an experiment was performed under the predicted conditions of the developed model. The response surface method (RSM) in its optimum condition was used in an experiment (Liu et al. 2019).

For verification of the optimization results, an experiment was performed under the predicted conditions of the developed model, which resulted in the TDS and turbidity removal (83.64 and 95.13%), respectively, as shown in Table 11.

As illustrated in Figures 25 and 26 and Tables 10 and 11, respectively, the TDS and turbidity removal using banana peel powder and M. stenopetala seed powder as coagulants demonstrate that the predicted and experimental values agreed well with a small deviation. This suggests that for the expected value, the model is thought to be accurate and dependable (Vera et al. 2014).

In various regions around the world, locally available natural coagulants are utilized to reduce water turbidity. According to research conducted by Beyene et al. (2016), a dosage of 3.5 g/L of cactus powder was found to eliminate 54.80% of turbidity from an initial turbidity level of 41.38 NTU. Another study by Asrafuzzaman et al. (2011) revealed that a dosage of 0.1 g/L of Moringa oleifera successfully removed 94.1% of the turbidity from a water sample containing 100 NTU. Additionally, Birhanu & Leta (2021) study discovered that 5 g/L of odarcha soil eliminated 88.46% of turbidity from an initial turbidity level of 800 NTU.

Generally, when we compare the turbidity removal efficiency of banana peel and M. stenopetala seed coagulants with other natural coagulants mentioned in Table 12, both coagulants show good efficiency.

According to research conducted by Jeje (2021), a dosage of 0.5 g/L of cactus powder was found to eliminate 45.1% of TDS from an initial TDS level of 255 mg/L. Another study by Gali Aba Lulesa et al. (2022) revealed that a dosage of 0.5 g/L of M. oleifera successfully removed 84% of the TDS from a water sample containing 70.1 mg/L. Additionally, Jacob's (2023) study discovered that 0.02 g/L of orange peel powder eliminated 20% of TDS from an initial TDS level of 1,500 mg/L.

Generally, when we compare the TDS removal efficiency of banana peel and M. stenopetala seed coagulants with other natural coagulants mentioned in Table 13, both coagulants show good efficiency next to M. oleifera.

This study investigated the effectiveness of banana peel powder and M. stenopetala seed powder in removing TDS and turbidity from Batena river water. The coagulation capacity of these coagulants was confirmed through Fourier transform infrared and SEM characterization analysis, revealing the presence of pores, void spaces, and polymeric substances like carbohydrates and proteins. The coagulation and flocculation capacity of both coagulants were found to be influenced by pH, coagulant dosage, stirring speed, and settling time.

The optimum conditions for both coagulants were determined using numerical optimization-based desirability function and compared with predicted values from the second-order quadratic model of CCD. The results showed that banana peel powder had the highest removal efficiency for TDS and turbidity under these optimum conditions, while M. stenopetala seed powder had the highest removal efficiency for TDS and turbidity under these conditions.

The analysis of variance evaluation of the p-value and F-value of the regression coefficient at a 95% confidence level showed that all individual factors and interaction effects had a significant effect on the removal efficiency of banana peel powder coagulant. In contrast, M. stenopetala seed powder coagulant had the highest removal potential for TDS and turbidity.

These findings suggest that natural coagulants, derived from readily available materials, can effectively reduce TDS and turbidity in water treatment processes, particularly M. stenopetala seed powder. This finding highlights the potential of natural coagulants as an alternative to chemical coagulants in water treatment, providing affordable and efficient water treatment solutions for communities.

The authors are grateful to the Hydraulics and Water Resources Engineering department for allowing us to use laboratory equipment.

The authors appreciate Wachemo University post-graduate schools for allowing us to work on this research at laboratories.

This research work complies with the research project's ethical standards.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.