Introduction

Secure drinking water sources and well-managed water treatment plants are indispensable to provide safe drinking water for human consumption. However, illegal industrial waste disposal practices stressed or poorly maintained distribution systems, aging pipes, absence of or ineffective filtration and disinfection facilities result in the deterioration of piped drinking water quality below acceptable levels, posing serious health risks to the community (Blokker et al. 2016). In addition, lack of proper sanitation and pollution control measures cause an increase in natural organic matter (NOM) in open water channels and distribution network (Sillanpaa et al. 2018; Abdullah 2014).

The intermixing of sewage with drinking water due to cross-connections is leading to chemical and biological/microbial contamination of drinking water, which in turn results in waterborne diseases within communities in Pakistan (Daud et al. 2017; Prest et al. 2016; Nabeela et al. 2014). An average of approximately 95% of drinking water samples were reported to be bacteriologically contaminated (Azizullah et al. 2011; Raza et al. 2017). In Punjab province, Pakistan, 90% of the people suffer from major waterborne diseases, e.g., cholera and diarrhea with cramps (Hannan et al. 2010).

In Pakistan, chlorine is added to the water in the treatment plant to inactivate pathogenic bacteria that have not been removed during the previous treatment steps and it is referred to as primary disinfection. In addition to the primary disinfection, to reduce the risk of a pathogen’s re-growth and to maintain biological stability of the drinking water, residual chlorine of 0.2 to 0.5 mg/L in the distribution network is permitted by the World Health Organization (WHO) (WHO 2011). However, the chlorination of drinking water containing NOM generates chlorinated by-products (CBPs) mostly trihalomethanes (THMs) (Fattahi and Shariati-Rad 2020; Ramavandi et al. 2015; Rezaee et al. 2014; Ibrahim and Abdul Aziz 2014). Poor management practices in Pakistan result in excessive THM production within the distribution networks most of the time. Various studies reported the presence of THM in the distribution networks of different cities in Pakistan (Abbas et al. 2015; Qaiser et al. 2014; Karim et al. 2011 and Amjad et al. (2013). These THMs are reported to be carcinogenic, as reported by several epidemiological studies (Wang et al. 2011; Brown et al. 2011).

The THM formation in the distribution network has been reported to be a function of the quality of water being chlorinated (Bourdon and Linares 2014), especially in relation to the concentration and characteristics of NOM and the chlorination conditions (Ramavandi 2015). In general, treated surface waters have higher THM concentrations than treated ground waters because of the potential high organic matter level caused by vegetation and warm temperatures (Pagano et al. 2014). In addition, treatment processes such as CT (chlorine initial concentration × contact time) may have a pivotal role in THM formation downstream along with other environmental conditions like water pH and temperature (Ramavandi 2015; Navalon et al. 2008; Sadiq and Rodriguez 2004). On the other hand, water distribution networks, due to extended contact time between chlorine residuals and THM precursors in the system, provide suitable environments for THM formation (Al- Omari et al. 2014). Higher chlorine residuals and the presence of NOM may augment the formation of THMs (Singh et al. 2012; Lee et al. 2010).

It has also been reported that THM concentrations in water increase significantly from the distribution network to the consumer’s tap, however, being higher at the extremities of the distribution network corresponding to the longest retention time (Hien et al. 2015). It is further documented that the levels of THMs increased as the distance from the treatment plant increased (Valdivia-Garcia et al. 2016; Le Bel et al. 1997). Since THMs pose a serious health risk to humans, the US Environmental Protection Agency (US-EPA) has regulated the maximum contaminant level (MCL) for THMs of 80 µg/L (Selvam et al. 2018; Fooland 2011; Valdivia-Garcia et al. 2016) which require water utilities to provide drinking water of high quality (Hua et al. 2016). The formation of THM could be used directly to assess the efficiency of chlorination for THM formation and the efficiency of the disinfection process (Fakour and Lo 2018). Nevertheless, monitoring and minimization of THM formation during chlorination of surface water sources, rich in NOM, pose a real challenge for the authorities in Pakistan (Valdivia-Garcia et al. 2016). In addition, THMs in the drinking water supply system should be monitored periodically to minimize or eliminate their presence whenever the concentration approaches threshold levels (Poleneni and Inniss 2013). Thus, the use of modern available tools, like the Response Surface Methodology–Central Composite Design (RSM-CCRD) for GC–MS process variables optimization and modeling, may result in a better understanding of the interactive effect of these factors and in turn a better prediction of the response through the suggested model.

THMs are usually analyzed through various chromatographic techniques such as gas chromatography (GC) with electron capture detection (GC-ECD), GC mass spectrometry (GC–MS), high-performance liquid chromatography with mass spectrometry (HPLC–MS), ultrahigh-performance liquid chromatography with mass spectrometry (Up-LC–MS), ion chromatography (IC) and capillary electrophoresis (Liu et al. 2013a, b). Because of the volatile nature of THMs, GC–MS is preferred over HPLC due to its high sensitivity for volatile organic compounds (VOCs) (Niri et al. 2008), its detection limits in the μg/L range, and its good linearity and reproducibility (Valencia et al. 2013).

In GC–MS analysis, although different sample preparation techniques (e.g., extraction and concentration using salting out effect) and operational parameters (e.g., initial temperature and column separation by applying the temperature gradient, i.e., ramp) were evaluated individually by many researchers (Yuan et al. 2018; Valente et al. 2013), these two categorical sets of parameters have not been optimized together; nor was there any statistical optimization ever applied to model the effect of both on THM detection and quantification. Thus, for simultaneous detection and quantification of THM, the GC–MS process variables need to be optimized.

Therefore, the present study was conducted to (1) optimize and model the GC–MS operating conditions for THM detection through liquid–liquid extraction (LLE) using RSM-CCRD and (2) apply on-site the optimized conditions to determine and quantify THMs in drinking water from Ratta Amral, Rawalpindi. This is the first study optimizing and modeling the GC–MS operating and extraction conditions simultaneously. It is expected that this study will set a baseline for the optimization and modeling of the critical GC–MS operating condition necessary for accurate detection and quantification of THMs in drinking water samples. The results of this study will further be helpful to water treatment authorities, regulatory bodies and stakeholders regarding formulation and implementation of the regulation regarding occurrence of THMs in drinking water distribution networks, and their monitoring and removal strategies. This will raise awareness about the potential presence of THMs and the risk linked to consumption of THM containing water in Rawalpindi and Islambad, Pakistan.

Materials and methods

Sample collection

Ratta Amral, under UC-01, Dhoke Ratta, Rawal Town, Rawalpindi is situated at 33° and 36.29" North and 73° and 2.45" East (Fig. 1). Rawal dam and Khanpur dam supply water to Rawal and Sangjani treatment plants, respectively, where water is treated by coagulation, flocculation, sedimentation, filtration and chlorination before distribution to the densely populated area of Ratta Amral and adjacent areas of Rawalpindi. Water samples were collected in glass bottles (1 L each) from the Water Treatment Plant (WTP) outlet, overhead reservoir (OHR) and from consumer’s taps as per standard methods (APHA 2012). For water sample collection, the faucets were turned on for about 5 min before samples were collected to make sure the water was coming directly from the mains and not from the building’s piping system. On site measurement for residual chlorine was performed immediately using portable multimeter (Spectroquant Picco). For THM extraction, amber glass bottles were used with 0.01 N sodium thiosulfate (Na2S2O3) as a quenching reagent for residual chlorine to prevent additional TTHM formation during transportation. Bottles with Teflon-coated rubber septa were filled to zero head space to prevent TTHM volatilization, sealed and stored at 4 °C. Duplicate samples for THM measurement were collected once a month from each sampling location. Samples were analyzed for THMs within 24 h of collection for chloroform (CF), bromodichloromethane (BDCM), dibromochloromethane (DBCM) and bromoform (BF).

Fig. 1
figure 1

Location of the sampling area of Ratta Amral, Rawalpindi (Courtesy: Google 3D maps). In an effort to assess water quality changes after each treatment operation and evaluate treatment efficacy relevant to the potential cause for high TTHM in the finished water of Ratta Amral DWDN, water samples were collected on a monthly basis for a period of six months. Drinking water samples were collected as per standard methods (APHA 2012) from the water treatment

Full size image

Design of experiment (DoE) using RSM-CCRD

RSM is useful for scheming experiments, structure models and investigating the effects of independent variables on dependent variables (Ramyadevi et al. 2012). In this study, RSM coupled with CCRD was used to optimize the GC–MS analytical conditions for THM detection and quantification. The optimization technique in RSM includes displayable and interactive three-dimensional (3D) plots and two-dimensional (2D) contour graphs. In addition, a graphical representation of the regression equation is used to visualize the relationship between the response and experimental points of each factor (Younis et al. 2014). RSM-CCRD quantifies the relationship between the controllable input parameters and resultant response surfaces. Despite having other more effective statistical and mathematical tools for process optimization such as the Box–Behnken design (BBD), CCRD was selected because it predicts response based on a small number of experimental data set, in which all parameters vary within a chosen range (Czyrski and Jarzębski 2020; Almeida et al. 2017). Also, the degree of freedom offered by CCRD helps the creation of more reliable models, especially in situations when some experiments can be affected by experimental error (Rakic et al. 2014). The development of statistical models may be valuable for forecasting and understanding the results of experimental factors. The main advantage of RSM-CCRD is its optimization capacity for multiple operational variables simultaneously with a small number of experiments, saving time and labor.

In liquid–liquid extraction GC–MS (LLE–GC–MS), salt is added to enhance the THM extraction from the aqueous phase to the solvent phase as a sample preparation parameter (Budziak and Carasek 2007). The addition of salt increases the ionic strength of the solution, changing the vapor pressure, solubility and surface tension of the analytes, resulting in the change of liquid/vapor equilibrium of the system, therefore making it easy to be extracted and separated from the aqueous phase (Niri et al. 2008). Maximum THM extraction was observed at low concentration of the salt as described by Santos et al. (2013); therefore, low salt concentration between 0.25 and 1 g was selected for optimization studies. Parkinson et al. (2016) described a 10% increase in extraction by adding an additional 10% of salt. As operational parameters, initial temperature and ramp functions were selected (Uppeegadoo et al. 1999). To overcome challenges, the optimization and modeling of the GC–MS process variables are highly required through the modern available modeling and statistical tools like response surface methodology coupled with a central composite rotatable design (RSM-CCRD).

Three independent variables, i.e., initial temperature (temp, ºC), ramp (ºC/min) and salt addition (g), were selected for the optimization process in this study. The variables with the ranges are shown in Table 1, applying RSM-CCRD (Design-Expert: version 9). The selection of these variables with defined experimental ranges was carefully chosen based on a previous screening of these variables determined in earlier preliminary experiments using the classical one-variable-at-a-time approach and from earlier documented literature (Elfghi and Amin. 2013; Niri et al. 2008). But single-factor optimization is time-consuming and laborious. Furthermore, it hardly ever promises the determination of optimal conditions for effective and efficient production of the target product (Shafi et al. 2018). Based on these facts, the lowest, center and the highest levels of selected variables, including axial star points of (− ά and + ά), are shown in Table 1.

Table 1 Independent variables with their ranges for optimization of GC conditions
Full size table

The independent variables were used as input variables by RSM-CCRD, which resulted in a composite factorial design with the lowest value (on extreme left) of – α, − 1, 0, + 1 and + α (the highest value on extreme right) (Table 2). The design of experiments (DoE), with twenty experimental runs, was carried out randomly to guarantee the independence of the results, reducing the effects of uncontrolled factors. After the assessment of the responses, the model was predicted, evaluated and analyzed as described by Teglia et al. (2015) and Kohli and Singh (2011).

Table 2 Design of experiment (DoE) with independent variables and their low and high levels
Full size table

TTHM extraction

Prepared THM standards were purchased from Supelco in a stock concentration of 5000 µg/mL dilution in methanol as solvent, and a known working solution of fluorobenzene (FB) (2000 µg/mL) as an internal standard was prepared. The USEPA method for THM detection 551.1 via GC–MS was applied using LLE with methyl-ter butyl ether (MtBE) as an extraction solvent. For extraction of THMs, 7 mL of the water sample and various masses of sodium sulfate (anhydrous) were mixed briskly to observe the salting out effect. One mL of the MtBE was vortexed in the same test tube and left undisturbed for 2 min from which 1 µL of organic layer was analyzed through GC–MS. The GC–MS conditions are described elsewhere (Rasheed et al. 2016).

Statistical analysis

Analysis of variance (ANOVA) was applied to observe the significance of interactive effect and strength between each independent variable as determined by the F and p value, respectively. A p value less than 0.05 shows a constructive influence on overall response (i.e., THM extraction and quantification) for each factor (Behin and Farhadian 2016).

Multifactorial optimization

Multifactorial optimization is the arrangement of different optimized parameters, giving a specific reaction concurrently (Ramyadevi et al. 2012). The simultaneous optimization of all responses is only possible by combining input variables into a single-objective function or desirability function, denoted by (D), which basically represents the relationship of all responses that are to be optimized (Ackey and Anagun 2013; Kohli et al. 2011; Mohan et al. 2012). A value of D closer to 1 is considered most appropriate (Hegazy et al. 2013).

Results and discussion

TTHM quantification by GC–MS and optimization of critical variables

Analytical method development and validation procedures are vital in the determination and quantification of any contaminant in water samples. Experimental design was used to find the optimal analytical conditions for the chromatographic separation. For THM extraction, and its subsequent detection from water samples through GC, it is reported to be dependent on salt concentration during LLE, while for individual component separation and detection, it is reported to be column initial temperature and ramp function dependent. Therefore, the optimum performance of GC–MS analytical conditions was optimized using RSM-CCRD and data analysis was performed using Design-Expert (DX version 9). RSM-CCRD establishes the effects of the input variables on the dependent response and their interactions (Karimi et al. 2011). The result would be helpful to optimize extraction and analytical conditions which are required to isolate these by-products from water, thus allowing their detection through GC–MS.

The design matrix consisted of 20 experimental runs incorporating the independent variables (salt concentration, initial temperature and ramp function) and the responses (THMs) as displayed in Table 3. The TTHMs were extracted under various experimental conditions defined by DoE, and THM concentrations were recorded as a response on the extreme right of Table 3. The ANOVA was calculated through CCRD. The suggested two-factor interaction (2FI) model was significant (p = 0.005) at a 95% confidence level with an F value of 5.35 (Table 4). A huge F value, while with a very small p value (p = 0.005), indicated a significance of the derived model as described earlier by Appavoo et al. (2014), Chen et al. (2022) and Liu et al. (2015). Furthermore, the higher the F value, the higher the probability that the variance contributed by the RSM model will be considerably larger than random error (Alman-Abed et al. 2020). The determination coefficient R2 value was 0.85, while adjusted R2 (Adjust. R2) was 0.68, authorizing that the model elaborated the experimental statistics well. Adequate precision (Adeq. Pre.˃ 4) suggests that the model could be applicable for describing the effect of variables on THM detection and quantification. It is evident that the effect of initial temperature was insignificant (p = 0.64), while ramp function (p < 0.0043) and salt addition (p < 0.04 were observed to be significant.

Table 3 DoE with respective results for GC optimization conditions
Full size table
Table 4 Analysis of variance (ANOVA) for TTHM extraction conditions
Full size table

Furthermore, product of initial temperature and ramp (A*B; p = 0.0081), and ramp and salt addition (B*C; p = 0.046) were observed significant. The interactive influences of the studied factors were plotted as 2D and 3D graphs using DX-9. These graphs described the precise elucidation of an interactive effect studying independent variables as suggested by Asadzadeh et al. (2018). The effect of the different variables is discussed one by one below.

Interactive effect of ramp and initial temperature

Column temperature was documented as a critical factor for separation of the target analytes by Liu et al. (2015). While observing the interactive effect of initial temperature (X1: A) and ramp function (X2: B) in Fig. 2, where TTHM concentration (µg/L) is shown on z-axis, the color gradient from blue to red showed the increased concentration of TTHM species, and the same color representation is depicted in the 2D contour graph (Fig. 2, X1: A: initial temperature, X2: B: ramp function). When discussing the initial temperature, a reasonable TTHM detection was observed at 45 ºC, while it is apparent that by increasing the ramp from 4*150 (4 degree rise in temperature per minute to 6*180 (6 ºC rise in temperature per minute to 180 ºC), TTHM detection and quantification increased from 300 to 700 µg/L, respectively (Fig. 2). This may be due to an increased temperature which increased the desorption rate of the TTHM species from the column; therefore, reaching the detector resulted in good peak signal production. It is further reported that these compounds spend more time in the mobile phase at higher temperatures, helping them elute faster and reducing band-broadening (Bloomberg and Klee 2001). Allard et al. (2012) also documented an increased response from 160 to 200 ºC for CHBrI2 and CHI3.

Fig. 2
figure 2

3D Surface plot and 2D contour of ramp and initial temperature interaction

Full size image

Effect of salt addition and ramp function

Shi and Adams (2012) discovered that the use of sodium sulfate improved the sensitivity of THM detection in their study. To enhance the separation of the organic layer from water sample, various salt concentrations were applied. The mutual effect of salt concentration and ramp function (X1: B: ramp, X2: C: salt addition) showed enhanced TTHM separation in the organic layer at higher salt concentration (Fig. 3) which resulted in a significant increase in the peak area of TTHMs, depicting higher concentration of TTHMs in the same water sample than when a small amount of salt was used. Figure 3 shows that the salt addition of 0.5 g was sufficient to cause maximum separation of the organic layer from the solvent layer, while maximum TTHM separation as peaks and their detection was observed at a ramp function of 6*180. The same effect can be observed in a 2D contour graph (Fig. 3: X1: B: ramp, X2: C: salt addition). This may be due to the fact that an increase in salt concentration could enhance the ionic strength of the aqueous phase and thus drive the target analytes into the organic phase (Liu et al. 2013a, b). Therefore, more extraction of TTHMs from the water sample was observed with increasing salt concentration from 0.25 to 0.5 mg/L. According to Santos et al. (2013), the addition of salt resulted in the change of liquid/vapor equilibrium of the system. These results agreed with those already described by Budziak and Carasek (2007).

Fig. 3
figure 3

3D Surface plot and 2D contour of ramp and salt addition influencing TTHM extraction

Full size image

Modeling THM extraction and separation conditions

The Pareto analysis compares the relative contribution of each studied variable on the response in a graphical form (Asadzadeh et al. 2018). So, to assess the contribution of most significant factors toward effective extraction of TTHMs from water samples, the interactive plot of significant factors was taken into consideration (Table 4). A Pareto chart of effect was plotted for each factor taken as an individual, as well as a mutual interaction, in various combinations, based on percentage share in the overall process (Fig. 4). The bar lengths, in Fig. 4, described the percentage contribution of each variable factor in TTHM extraction and detection by GC–MS as mentioned by Tuncel and Topal (2011). The most significant variable was determined to be the product of A and B (initial temp*ramp), which contributed almost 40% to the process. This may be due to the fact that temperature plays a vital role in the separation of each and every constituent of the mixture by interplaying with the stationary and mobile phase as described by Blumberg and Klee (2001). The analysis indicated that the order of effective variables was B (ramp) > a product of B*C (ramp*salt addition) > C (salt addition) > A (initial temperature) > and a product of A and C (initial temperature*salt addition).

Fig. 4
figure 4

Pareto chart of effects

Full size image

Model development for the THM extraction and detection

Based on the ANOVA for the interactive effect of the studied factors (Table 1), a relationship between the input variables and the resulting response was attained and expressed by a reduced model.

$${\text{THMs }}\left( {{\mu g}/{\text{L}}} \right) \, = { 491 } + { 67}.{\text{42 B }} - { 45}.{\text{76 C}} + {79}.{\text{75 AB}} - {56}.{\text{25 BC}}{.}$$

The proposed statistical model provides a critical analysis of individual and simultaneous interactive impacts of the selected independent variables.

Evaluation of the model

The adequacy of the model was verified and validated by correlation between normal plots of residuals (difference between the experimental and the predicted values) as shown in Fig. 5 (x-axis: X1: Externally studentized residuals, y-axis: X2: normal percentage probability) as described by Singh et al. (2012). Graphical data close to a straight line showed that the model sufficiently predicted the effect of the studied factors which is further confirmed by R2 = 0.85 and p ≤ 0.005) (Rezaee et al. 2014) with no serious violation of the assumptions underlying the analyses and confirming residuals independence.

Fig. 5
figure 5

Normal plot of residuals

Full size image

Numerical optimization

The optimum conditions for all the studied factors were optimized by attaining the maximized objective function (D) by numerical optimization technique using RSM-CCRD. For maximum TTHM extraction and its quantification, the initial temperature at 50 ºC, ramp at 6 ºC rise/min to 180 ºC and 0.5 g salt were optimized at D value approximately 0.97 as a multifactorial optimization process (Fig. 6).

Fig. 6
figure 6

Ramp function for optimum conditions for maximum TTHM detection in GC at D = 0996

Full size image

Graphical optimization

Graphical optimization was used to trace the optimum levels for maximum response, which could be visualized in the form of a graph as shown in Fig. 7; x-axis: A: initial temperature, y-axis: B: ramp). The maximum predicted value is indicated by the surface confined in the smallest eclipse in the yellow contour known as the overlay plots (Trinh and Kang 2010). These overlay plots allow a visual selection of the optimum conditions according to a certain criterion which is in full agreement with the set limits in this yellow region (Almeida et al. 2017).

Fig. 7
figure 7

Overlay plot with optimal conditions region of TTHM detection and quantification

Full size image

The CCRD determined and modeled the TTHM extraction and quantification process efficiently. It is a technical and economic method to gain the maximum amount of information within a small period of time and with few experiments. As a conclusion, significant variables affecting the extraction efficiency were determined, including the amount of salt for solvent separation in the LLE extraction process as a sample preparation step. Oven temperature and ramp function were optimized as GC–MS operational conditions for maximum THM detection and quantification. The optimum experimental conditions obtained from this statistical evaluation included a 0.5 g salt during the LLE process resulted in the maximum TTHM extraction at an initial temperature of 50 ºC, ramp at 6 ºC rise/min to 180 ºC at D = 0.97.

These proposed optimized conditions were applied in the second phase of the experiment to determine and quantify TTHMs from water samples of a local populated area, and good agreement was observed between modeled and actual data.

Application of optimized parameters on real water samples for the determination of THMs in the water distribution network of Ratta Amral, Rawalpindi

An assessment of the influent and effluent water in WTP can determine the relationships between WTP performance, water quality and DBP formation (Cook and Drikas 2011). As discussed earlier, the constitution of various THM compounds in chlorinated water depends on the chlorine dose, the nature and concentration of organic precursors, pH, temperature, presence of iodide ions and residual chlorine concentration (Hamed et al. 2017; Ramavandi et al.2015). After optimizing the analytical conditions for the THM extractions and analysis, drinking water samples for TTHMs from Ratta Amral, Rawalpindi, were analyzed. Residual chlorine, including free and total chlorine, were also analyzed to determine their effect on THM formation in the area (Table 5). The proposed optimized conditions were applied to determine and quantify TTHMs from water samples of a local populated area, and good agreement was observed between the modeled and actual data.

Table 5 THM and residual chlorine concentration in DWDN of Ratta Amral, Rawalpindi
Full size table

THMs were observed in all sampling stations ranging from 248 to 305 µg/L as shown in Fig. 8 far above the WHO permissible limit of 80 µg/L. These results were in accordance with those reported earlier by Amjad et al. (2013). Meanwhile, the free chlorine concentration at the outlet of the water treatment plant was 0.85 mg/L which reduced by more than a factor of 3 to 0.25 mg/L, i.e., 0.25 mg/L, when water reached the OHR. After that, no free chlorine was detected at any sampling point. The presence of TTHMs in samples from the WTP could be attributed to the presence of organic matter compounds, which are difficult to remove using conventional water treatment technologies and the continuous presence of residual chorine in the distribution network. These results were found in accordance with Toroz and Uyak (2005), who observed seasonal changes in the concentration of THM in a distribution system. According to El Shafey et al. (2000), the formation of THMs in the treatment plant only represented about 45% of the THMs found at the end of the pipelines. The provision of adequate residual chlorine to protect the drinking water from bacterial contamination/re-growth is recommended by USEPA (2006). But this practice could lead to additional THMs in drinking water under improper organic removal (Chaudhary et al. 2008). CF was found to have the highest concentration (approximately 90% of total THM) among the THMs, followed by BDCM and DBCM with BFM having concentrations below the detection limit (Fig. 8). This THM species distribution may be attributed to the fact that bromine–carbon bonds are more tolerant to dissociation, compared to chlorine, as a result of lower dissociation energies (Abusallout et al. 2017). These results were in accordance with Clayton et al. (2019), Hua et al. (2016) and Elsheikh and Basiouny (2011). This high concentration of TTHMs is of significant concern, requiring the use of an alternative water source or additional treatment such as activated carbon to remove DBPs prior to water distribution.

Fig. 8
figure 8

THM concentration from residential area of Ratta Amral, Rawalpindi

Full size image

On the other hand, it was further observed that with increasing distance from the WTP, the formation of THMs increased. Chang et al. (2010) reported most of the THM formation occurred during the initial contact hours of treatment. However, the results contradicted Kolla (2004) and Chang et al. (2001) who observed no noteworthy rise in THM concentration beyond 48 h of disinfection. In a study by Le Bel et al. (1997), levels of THMs increased as the distance from the treatment plant increased. Generally, under favorable environmental conditions, the presence of NOM and sufficient residual chlorine in the distribution network, THMs continue to be formed (Sun et al. 2009).

Conclusion

The proposed optimized conditions were applied to determine and quantify TTHMs from water samples of a local populated area, and good agreement was observed in proposed and actual data. The effect of key parameters (initial temperature, ramp function and salt addition) on TTHM extraction and quantification through GC–MS was optimized using RSM-CCRD. This investigation shows that RSM-CCRD is a suitable method to optimize the operating conditions in order to maximize the TTHM response. The use of RSM-CCRD in optimizing the separation, as well as operational conditions, turned out to be a significant innovation.

The main conclusions drawn from this work are:

  • A ramp function and salt addition were observed to be significant with a p value of 0.004 and 0.04, respectively. On the other hand, a product of initial temperature and ramp (A*B; p = 0.008) and ramp and salt addition (B*C; p = 0.04) were observed to be significant as two-factor interactions.

  • An initial temperature of 50 ºC, ramp at 6 ºC rise/min to 180 ºC and 0.5 g salt resulted in the maximum TTHM extraction at D = 0.97.

  • TTHMs were observed in all sampling stations ranging from 248 to 305 µg/L, which were far above the USEPA allowable limit of 80 µg/L. This high TTHM concentration could be linked to the presence of organic matter and residual chlorine in the distribution network.

  • CF had the highest concentration (approximately 90% of total THM) among the THMs, followed by BDCM, while DBCM and BFM concentrations were below the detection limit.

  • The free chlorine concentration exiting the water treatment plant was found to be approximately 0.85 mg/L, which reduced to more than one third of 0.85 mg/L, i.e., 0.25 mg/L, when water reached the OHR. After this point, no free chlorine was detected at any sampling point.

  • Despite the significant findings of the study, sometimes it is not easy to achieve the adjusted values of the chosen factors in the RSM-CCRD design. There is also a predictable inability of the tool to estimate individual interaction terms.

Based upon the above conclusions, the following recommendations are made:

  • Continuous monitoring of THM species should be carried out to check the stability of the system.

  • The entry of the total organic carbon causing THM formation in the system could be minimized by taking control measures.

  • Additional treatment options like granular activated carbon adsorption may be used before the water reached the end user.