The reliability and process capability assessment of suspended growth sewage treatment plant in Melaka, Malaysia

Sewage is the term used for wastewater generated from domestic, commercial, institutional, and industrial sectors. It contains 99.9% of liquid waste and 0.1% solid waste, mostly classified as grey water and black water, and is characterized by volume, physical conditions, chemical and toxic constituents, and its bacterial content. Naturally, sewage physical characteristics can be detected by physical senses, i.e., temperature, colour, odour, and solids. Meanwhile, the chemical characteristics of sewage indicate the strength of decomposition for treatment requirements such as pH, biological oxygen demand (BOD), chemical oxygen demand (COD), total suspended solid (TSS), oil and grease (O&G) and ammoniacal nitrogen (NH₃-N). The biological characteristics of microorganisms, i.e., algae, bacteria, fungi, and protozoa, cause water-borne diseases such as cholera, polio, dysentery, and typhoid, as well as environmental deterioration such as low dissolved oxygen (DO) caused by the eutrophication phenomenon.

Sewage volume increases owing to extensive population growth and urbanization, escalating the number of sewage treatment plants (STPs) significantly. However, increasingly stringent effluent quality standards and the inability to meet the stipulated limits remain the major challenges in the wastewater treatment sector (Hamza et al. 2022). This has raised various environmental health issues and major concerns about proper treatment to remove the hazardous substances, biodegradable organics and pathogenic organisms to ensure the effluent discharge concentrations are within the limits established (Dehghani et al. 2018). Improving the quality of the treated effluent and reducing the operational costs of STPs are very complex due to the process variations, including shock loads, mechanical equipment failures, climates constraints, and human competency (Oliveira & Von Sperling 2008).

Traditional STP performance assessment in Malaysia was based on the experience of the operators and engineers and required more diagnostic analysis than descriptive analysis. To overcome these circumstances, another promising method of reliability assessment can be proposed to gauge their current performances. Reliability of the system can be defined as the probability of achieving the desired functions or adequate performance for a specified time period under specific conditions (Chorafas 1960; Modarres et al. 2016; Padalkar et al. 2019), specifically, the ability of STPs to operate for the specified period of time without failure or the percentage of time that the effluent concentration meets the specified permit requirements (Metcalf & Eddy 2003). The overall long-term reliability in STPs is based on the resilience to effluent variability (Alderson et al. 2015) and limited discrepancies in the plant performance to set the acceptable magnitude of the violation (Oliveira & Von Sperling 2008). When reliability is considered for an STP operation, the analysis results will allow the engineer to exploit the statistical structure of influent and effluent data in order to predict the probability of undesirable events. The inherent variables of both influent quality and quantity during the design phase of the treatment facility may result in deviations from predicted efficiencies (Djeddou et al. 2014).

In mathematical terms, an STP is highly reliable if there is no failure in process performance, operation, or maintenance, including discharge requirement violations at all times. However, high reliability means an increase in operational cost, i.e., utilizing highly efficient equipment, skilled manpower, sophisticated laboratories, sampling apparatus, data acquisition systems, and additional maintenance costs, which is practically unfeasible. The probability of the standard being exceeded by the effluent depends on the distribution function describing the quality of parameter concentrations. The characteristics of effluent were right-skewed probabilistic distributions with low arithmetic mean data and a large number of variables with positive values (Kottegoda & Rosso 1997; Montgomery 2020; Von Sperling et al. 2020) making symmetrical distribution methods difficult to analyze treatment performance. Subsequently, the need to continuously provide an effective and efficient treatment while STPs’ infrastructure is ageing calls for a strategic and tactical approach at the operational level of planning.

To compare the differences in STP performance, the stability measure by standard deviation (σ) is being used as the most appropriate stability indicator. Various distributions, i.e., Weibull or Gamma, were applied in other probabilistic models (Bugajski et al. 2016; Zawadzka et al. 2021). However, taking into account the simplicity of Niku et al. (1979), they apply the log-normal distribution in order to develop the coefficient of reliability (COR), an index that relates the arithmetic mean concentration of parameters to comply with quality standards with a required reliability level focused on the activated sludge process (Niku & Schroeder 1981) and trickling filters (Niku et al. 1982). Those COR also recommended by USEPA were developed based on the TSS and BOD data from suspended and attached growth biological treatment systems with a high degree of mechanization and operational flexibility which were recognized in the previous technical literature and widely used for reliability analysis (Metcalf & Eddy 2003; Oliveira & Von Sperling 2008; Andraka 2019; Von Sperling et al. 2020). A comprehensive study comparing the reliability of suspended growth biological systems of STPs is still lacking and requires significant attention, especially in developing countries, particularly Malaysia.

The process capability of STP is dependent on many factors described above, but the major contributor of variance is from the biological treatment process and results in unstable treatment, which negatively affects the effluent quality. Various efficient techniques were developed to solve the instability, most of the well-known Six Sigma philosophy was applied to overcome problems, mainly in the manufacturing sector. The Six Sigma structure helps to optimize the company's performance, increase quality, maintain customer satisfaction, reduce costs, as well as benefit the financial, social, and environmental by eliminating waste (Mohamad et al. 2019). This technique was first implemented in the 1980s by Motorola as a systematic framework to address quality variation challenges faced in manufacturing. Contemporarily, the application of Six Sigma has been expanded into numerous non-manufacturing sectors such as clinical care, healthcare, environmental pollution control programmes, and wastewater treatment field as a complementary framework to decrease waste, improve resource optimisation, increase customer satisfaction after offering a consistent and reliable process (Ishak et al. 2022). Generally, STP operations undergo a routine performance monitoring programme with expensive data collection activities. However, dozens of complex datasets become useless when these datasets remain unutilised to investigate the treatment performance. This could help in the early detection and timely response to the inefficiencies, process failures and operational abnormalities in the treatment process. The benefit of using process capability analysis, i.e., Process Stability (Pp and Cp) determines the potential ability of a process, whereas the Process Capacity Index (Ppk and Cpk) calculates the actual capacity of the treatment process. The process capability assessment was expressed as the ratio of the process specification interval variations in 6σ. This straightforward technique will utilize the existing probabilistic COR assessment to evaluate the routine STPs’ performance without requiring highly skilled data scientists by using available performance monitoring datasets.

Previous literature regarding process capability assessment demonstrates positive results for controlling pharmaceutical wastewater quality (Rimantho & Nugraha 2020), while an examination of energy and reliability analysis in WWTPs in Ontario (Hamza et al. 2022) revealed a high reliability level of over 95%. A recent study to investigate the effectiveness combination of COR and process capability to assess 23 Portuguese WWTPs’ annual reliability for BOD₅, COD, and TSS (Silva & Rosa 2020), as well as Malaysian poultry wastewater by using the Weibull probabilistic coefficient (Ishak et al. 2022). Hence, the current work attempts to use a simple and consolidated reliability-based and process capability method to trigger decision making on the STPs’ operational improvement and discharge compliance to maintain the treatment effectiveness and efficiency. The aim of the study is to (i) diagnose the STPs’ reliability, (ii) determine the process capability of the STPs, (iii) estimate the different performance of various suspended growth biological systems of the STPs; and (iv) derive the benefits of COR and process capability assessment for future research. This study is being conducted for 15 Malaysian STPs with various suspended growth biological processes located in historical Melaka City's suburbs. Despite the high potential of this simple and consolidated framework, to our knowledge, it has not been fully explored as a support tool in STPs’ effectiveness and efficiency at the routine performance monitoring level, particularly in Malaysia.

Over the last half of the 20th century, Malaysians have made tremendous economic progress and have invested in quality living infrastructure. Therefore, the rapid urbanization throughout the country requires dramatic improvement in sanitation and sewerage management. In the 1960s, the five-year Malaysia Plan was launched to construct appropriate sanitation facilities across the nation to replace septic tanks as the primary sewage treatment system. According to the National Sewerage Development Program, the installation of sewerage facilities in major cities with the aim of introducing modern sewerage systems in urban areas is underway. The policy introduced during the 1980s that obliged housing developers to build sewerage systems for regions comprising more than 30 households (150 population equivalent (PE)) resulted in the diffusion of many small-scale STPs all over the country, which are gradually connected to large-scale sewerage systems to complete the public sewerage connection. Although recently, this large-scale sewerage system compromises the multipoint and regional STPs have increased in number. Since 1994, the management of wastewater in Peninsular Malaysia and the Federal Territory of Labuan has been under the purview of the Water Services Industry Act, 2006 (WSIA 2006) and is undertaken by a private company named Indah Water Konsortium Sdn Bhd (IWK). According to the most recent Water and Sewerage Fact Book, 2020, more than 7,101 multipoint STPs covering 18,321,827 PE, 102 regional STPs catering 9,203,835 PE from the public sector, and 4,307 private sector STPs covering 4,724,242 PE were in operation. The services also maintain 20,868 km of sewer pipelines and 1,312 network pumping stations across the country (Malaysia National Water Commision 2020). Under the national agenda to the transform water and wastewater sectors in Malaysia, the long-term Water Sector Transformation 2040 (WST 2040) will be restructured to be a significant contributor to national growth aligned with Sustainable Development Goals 2030 (SDGs). However, the main challenges regarding existing plant operations are that there is still flaws in terms of highlighting the level of compliance in Malaysian STPs utilising the probabilistic reliability and process capability methodology to assess their effluent discharges.

The secondary treatment of STP consists of various types of biological treatment processes in the form of fixed film, suspension, or a combination of both (Metcalf & Eddy 2003). Hence, the biological treatment process for STP can be classified under one of the following; (a) attached growth; (b) suspended growth; and (c) combined process (hybrid). The active microorganisms grow and attach to the mobile or immobile medium such as a rock or plastic in the attached growth process to utilise the food in the sewage. The surface of biomass is used as the practical measure of the total organism's activity (Niku et al. 1982). The trickling filter (TF), rotating biological contactor (RBC), submerged biological contactor (SBC), fluidised bed, and packed bed reactor are the common attached growth processes found in Malaysia. Concurrently, in a suspended growth process, the active microorganism remains in suspension in the sewage and their concentration is usually related to mixed liquor suspended solid (MLSS), or mixed liquor volatile suspended solid (MLVSS) which requires a mechanical aeration system to remove organic substances in the sewage by the active microorganism grow during the process (Von Sperling et al. 2020). Popular suspended growth processes in Malaysia include the waste stabilisation pond (WSP), deep shaft (DS), oxidation ditch (OD), AL, CAS, EAAS, and SBR. Modernization in sewage treatment technology includes the combination of various aforementioned attached growth and suspended growth processes to obtain the best performance and most economical treatment. As an upgrade alternative to the old STP, these hybrid processes offer advantages in stability and shock load resistance.

To verify the reliability and capability of the suspended growth process, data from 15 STPs in Melaka City, the central part of Peninsular Malaysia, was used to develop this study. Five-year (2017–2021) data was collected from monthly performance monitoring records provided by the plant operator. The spatial distribution of the various systems integrates homogenous climatic conditions within the state. From 1968 to 2019, the annual average 24-h mean temperature was 27.4 °C, with the highest mean 24-h temperature recorded in the month of May at 28.1 °C, and the lowest in the month of December at 26.8 °C. The annual mean precipitation is about 2,056.7 mm, with an annual mean rainy day of 170 days. The month of June recorded a mean value of 235.7 mm as the wettest month, whereas the driest month occurs in January with an average rainfall of 80.5 mm. Melaka City has a tropical climate with annual 24-h mean relative humidity around 81.3%. The lowest and highest months were February and November, at 76.8% and 84.3%, respectively. As presented in Table 1, the STPs analysed cover different design capacities, treated volumes, duration of operation, and effluent compliance standards.

Prior to the detailed descriptive analysis of the preliminary dataset, the missing, extreme, or unusual values that could affect the analysis must be resolved. Data cleaning and pre-processing involving missing values and outliers’ determination were performed. More than 10,000 initial performance monitoring records were obtained and identified to exclude the missing values. The outliers assessment recommended by empirical rule based on the interquartile range of datasets was initiated (Von Sperling et al. 2020). The outlier values below the lower limit or above the upper limit were excluded from the dataset; remaining sampling records were used for the assessment. The graphical box plot will determine the current performance of various treatment systems compared to prescribe limit compliance.

The USL – upper specification limit, LSL – lower specification limit, Cpu – upper one-sided capability ratio (within), Ppu – upper one-sided capability ratio (overall), Cpl – lower one-sided capability ratio (within), Ppl – lower one-sided capability ratio (overall) are used to determine the Cpk and Ppk, the process capability indices for within subgroup and overall treatment process, respectively, in STP. Since the performance monitoring sampling from STP was not performed rapidly in a short period of time and was a very expensive procedure, the subgroup of the sampling is normally equal to 1. Additionally, the lognormal data distribution of study parameters concentration deals with positive data; the process capability focuses on the upper one-sided capability ratio (Cpu and Ppu), which is value (maximum standard limit) and the value of Cp, Pp, Cpk, and Ppk to identify the Capability Process Index (CPI) and the range between Cpk and Ppk for this assessment. The interpretation of the capability index of Cpk and Ppk will follow the normal possible value should be 1.00, which indicates the STP process is adequate to fulfil the treatment tolerances and performance specifications during the normal condition (Lant & Steffens 1998; Montgomery 2020). The Cpk and Ppk ranges were used to determine whether the treatment process was in statistical control. The asymmetry of data distribution was fixed by Box-Cox transformation before performing the process capability assessment. Finally, the probabilistic reliability and process capability input were analysed by Minitab 18 software using one-way ANOVA and a post-hoc Tukey simultaneous test to compare the mean values of best performance and their differences within the group of suspended growth systems.

The low mechanical flexibility in the wide surface hydraulic loading rate (HLR) of OP treatment results in the lowest capability to treat the sewage influent in this study. The OP system indicates the higher mean and median values for BOD₅, COD, TSS, and O&G, especially for OP3. In the effluent of the OP system, the BOD₅ and TSS are closely related in the treatment With a low mixing rate by surface aerator, the possibility of these results is also associated with algae growth in the ponds (Oliveira & Von Sperling 2008). However, in the AL system, excessive aeration mixed the suspended microorganisms in the optimal dimensions (Metcalf & Eddy 2003). The EAAS outperforms the CAS by a smaller margin, resulting from different operational settings such as MLSS, MLVSS, food to microbe ratio (F/M), hydraulic retention time (HRT), and sludge age (Niku & Schroeder 1981). Despite this, the EAAS3 performed marginally worse than the other EAAS plants in this study. The higher hydraulic plant is capable of imbibing the inherent variability of the system as a consequence of the hydraulic load differences in the design capacity. From the graphical assessment, the SBR process marks the most efficient system, followed by EAAS, CAS, AL, and OP.

Overall, the observed poor performance in terms of nutrient removal was expected in conventional systems for AL, CAS, and OP since those systems are not designed for nitrogen removal. These findings suggest the conventional suspended growth process is acceptable for non-potable effluent reuse such as agriculture with limited human contact, because nutrient availability is a positive element for plantation (Alderson et al. 2015). However, the low NH₃-N performance in EAAS and SBR was very interesting to probe the current treatment procedure of each plant. The observed differences in the final effluent quality highlight the need to design the suspended growth process to produce an average effluent concentration substantially below discharge standards. However, by comparing discharge parameter values through descriptive analysis alone, the plant operator might be unable to identify the treatment's performance and the level of variability in the system. Consequently, the COR presents an attractive methodological alternative achieve an acceptable risk with process capability (Ppk and Cpk) to identify the real picture of the operating conditions.

Annual BOD₅, COD, TSS, O&G, and NH₃-N reliability and process capability studies were performed for the 15 STPs whose yearly values fit the normal and log-normal distributions (2017–2021). The detailed results obtained are presented in Appendix 1, which also includes the arithmetic mean value; ⁠, reliability at 95% of ⁠, and Ppk and Cpk for the effluent parameters. The annual dataset for parameters will be transformed for assessment if it does not fit a normal distribution.

As mentioned in the previous chapter, this study attempts to investigate the relationship of treatment reliability and process capability assessment to highlight the benefits of that combined method. The Anderson-Darling (AD) test was employed to verify if the parameter concentration from the 15 STPs analysed in each year of the five-year period fit a lognormal distribution at a 5% significant level. More than 97% of the data fit a lognormal distribution, allowing the COR method based on the log-normality of the data to be utilised to evaluate the reliability of the plant. In addition, the process capability assessment by using Minitab 18 was also performed in the log-normality transformation. The Kruskal-Wallis test concluded for all discharge parameters; the reliability, Ppk, and Cpk values were statistically different amongst the types of STPs, allowing for detailed comparison of those values within the suspended growth system in this study. For the 95% reliability level, the COR values decrease with the CV increase. Low COR values do not always indicate the malfunctioning of the treatment plant, but rather suggest less stable operational circumstances due to the higher variability of data reflected by the CV (Oliveira & Von Sperling 2008). The one-way ANOVA was performed to compare the mean values of reliability and process capability for the systems. The p-value at 95% significance level, marked at 0.000, indicates to reject the H_o, those mean values are not statistically equal. The detail values of respective parameters are shown in Table 3.

Table 3

Mean values for reliability, Ppk and Cpk for overall STPs

Parameter .	.	AL .	CAS .	OP .	SBR .	EAAS .
BOD5		0.8979	0.9571	0.9010	0.9997	0.9899
	Ppk	0.5740	0.6047	0.4480	1.4590	1.9310
	Cpk	0.6830	0.7007	0.4767	1.7390	2.3240
COD		0.9740	0.9840	0.9760	0.9999	0.9985
	Ppk	0.8047	0.8993	0.6813	1.7390	1.4580
	Cpk	1.0260	1.049	0.7367	1.9190	1.7450
TSS		0.9722	0.9676	0.9115	0.9996	0.9987
	Ppk	0.7013	0.8870	0.6020	1.5210	1.1770
	Cpk	0.9950	0.9930	0.6253	1.6040	1.3450
O&G		0.9377	0.9471	0.9364	0.9992	0.9881
	Ppk	0.6360	0.6220	0.5960	1.1960	0.8347
	Cpk	0.7047	0.6400	0.5940	1.3060	0.9100
NH3-N		0.4231	0.6412	0.6866	0.9048	0.7922
	Ppk	−0.0647	0.1253	0.1327	0.4100	0.2667
	Cpk	−0.0527	0.1247	0.1600	0.7050	0.3047

Parameter .	.	AL .	CAS .	OP .	SBR .	EAAS .
BOD5		0.8979	0.9571	0.9010	0.9997	0.9899
	Ppk	0.5740	0.6047	0.4480	1.4590	1.9310
	Cpk	0.6830	0.7007	0.4767	1.7390	2.3240
COD		0.9740	0.9840	0.9760	0.9999	0.9985
	Ppk	0.8047	0.8993	0.6813	1.7390	1.4580
	Cpk	1.0260	1.049	0.7367	1.9190	1.7450
TSS		0.9722	0.9676	0.9115	0.9996	0.9987
	Ppk	0.7013	0.8870	0.6020	1.5210	1.1770
	Cpk	0.9950	0.9930	0.6253	1.6040	1.3450
O&G		0.9377	0.9471	0.9364	0.9992	0.9881
	Ppk	0.6360	0.6220	0.5960	1.1960	0.8347
	Cpk	0.7047	0.6400	0.5940	1.3060	0.9100
NH3-N		0.4231	0.6412	0.6866	0.9048	0.7922
	Ppk	−0.0647	0.1253	0.1327	0.4100	0.2667
	Cpk	−0.0527	0.1247	0.1600	0.7050	0.3047

Note: Significant p-values < 0.05 indicates the means are not equal.

View Large

The COD is associated with and dependant on all the other parameters such as BOD₅, TSS, and O&G, which together contribute to the COD value along with other pollutants (Padalkar et al. 2019). Additionally, the COD is the most important parameter to assess the reliability and process capability of STPs due to the simplicity of their analytical procedures. The one-way ANOVA assessment for the COD parameter is more than 95% of values in Figure 3(B1) reveals the high reliability in overall treatment systems. As shown in Table 3, the COD reliability mean values for SBR, EAAS, CAS, OP, and AL were 0.9997, 0.9985, 0.9840, 0.9760, and 0.9740, respectively. The Tukey simultaneous test at 95% of the confident interval (CI) in Figure 4(B1) shows there are statistically no differences in mean value between the reliability of SBR and EAAS when p-value > α (0.999). However, they are statistically significant for differences of mean values between OP-EAAS, OP-SBR, EAAS-AL, and SBR-AL with p-values of 0.006, 0.003, 0.002, and 0.001, which indicates that the highly mechanical systems such as SBR and EAAS performed better reliability compared to the conventional systems (Metcalf & Eddy 2003). The differences in reliability mean value (Figure 4(B1)) in CAS system were not statistically different compared to the highly mechanical system and conventional ponding system. However, the Ppk and Cpk values for COD parameters were shown in Figures 3(B2), 3(B3), 4(B2), and 4(B3) compared with different types of treatment systems. There was a statistically significant difference between the mean values of Ppk and Cpk for both in highly mechanical system (SBR & EAAS), when the mean value was >1.00 which indicates the treatment process has meet the objectives and the capability to meet discharge limits with inherent random variability is small enough to be acceptable (Lant & Steffens 1998). The SBR, with a Ppk mean value of 1.739 and a Cpk mean value of 1.919, performs the highest process capability compared to EAAS, with a 1.458 and 1.745 Ppk and Cpk mean value respectively in this study. Consequently, the Cpk mean value for CAS and AL were >1.00 but the Ppk mean value were <1.00. The Cpk > Ppk indicates the treatment process is not performing up to its capability. Finally, the Ppk and Cpk mean values for OP were 0.6813 and 0.7367, respectively, making this ponding system the highest variability and lowest COD treatment reliability at 0.9740.

As shown in Figure 3(C1), the TSS reliability mean values for SBR, EAAS, AL, and CAS remain good with 0.9996, 0.9987, 0.9722, and 0.9676, respectively, while OP remains the poorest with 0.9115. Figure 4(C1) shows the differences in reliability mean value were statistically significant for OP-EAAS and OP-SBR at p-value = 0.029 and 0.027, which indicates highly mechanical system performs better than OP as a conventional ponding system (Alderson et al. 2015). In Figure 3(C2) and 3(C3), TSS Ppk and Cpk value for SBR and EAAS were >1.00, while the value of TSS Ppk and Cpk for AL, CAS and OP remains <1.00 require better process control to remove TSS by controlling the incoming flowrate, optimal hydraulic retention time and periodically preventive maintenance at settling chamber to avoid suspended solid carry over (Metcalf & Eddy 2003; Zawadzka et al. 2021). Figure 4(C1)–4(C3) shows the TSS reliability to mean value, Ppk and Cpk for SBR were statistically significant differences with OP when p-value < 0.05, which indicates the SBR system performs at the highest efficiency while the OP is at the worst performance.

To assess the efficiency of primary treatment from various suspended growth systems in this study, the TSS and O&G parameters were determined to identify which system performs the best treatment to reduce those pollutants. An O&G removal chamber is present as a part of preliminary treatment, and TSS is majorly removed by gravitational force in the screening chamber and primary settling tank (Metcalf & Eddy 2003). There is no physical-chemical unit process for clari-flocculation applied in the studied plants. The one-way ANOVA test at α = 0.05 indicates the p-value is 0.033, indicating there are statistically significant differences in the reliability mean value for the comparable systems. Again, as shown in Figure 3(D1), the highly mechanical system performs at a good reliability level of >95% for SBR and EAAS at 0.9991 and 0.9881, respectively. The CAS, AL and OP remain below the acceptable level (95%) at 0.9471, 0.9377, and 0.9364 accordingly. However, in Figure 4(D1), the Tukey simultaneous test at 99.34% of individual confidence level indicates there are no statistically significant differences in reliability mean values between various systems when all group comparisons of p-value > 0.05. Figure 4(D2) and 4(D3) reveal the SBR Ppk and Cpk mean values are significantly superior to EAAS at p-value = 0.001 and 0.004, respectively, which indicates the system performs better O&G removal with optimal hydraulic retention time (Hamza et al. 2022). Figure 3(D2) and 3(D3) show the Ppk and Cpk for EAAS, AL, CAS and OP were <1.00, which indicates the system is incapable of meeting the specification, even under good control. There is excessive random variation in the systems, which must be eliminated by changing the treatment procedure (Ishak et al. 2022).

Figure 3(E1)–3(E3) depicts the nutrient removal from secondary biological treatment and discloses the NH₃-N parameter mean values for reliability and process capability for the whole systems were below expectations. The reliability mean value for SBR was marked at 0.9048, while EAAS, OP, CAS, and AL are at 0.7922, 0.6866, 0.6412, and 0.4231, respectively, below 0.95 as shown in Table 3. The post-hoc Tukey test shown in Figure 4(E1) also reveals the AL marked the worst system compared to others, SBR was the most efficient but the mean value was not statistically significant between EAAS at p-value = 0.255. The same results in Figures 4(E2) and 4(E3) pertain to reliability and process capability values for EAAS-OP-CAS systems when p-value > 0.05, which indicates those system reliability mean values were not statistically different. The negative values of Ppk and Cpk for AL indicated in Table 3 show this type of treatment is inferior to fulfilling the treatment tolerances and performance specifications under typical conditions (Montgomery 2020). Besides, this study specifies that a fully closed loop and control treatment cycle in the reactor like SBR is more capable of removing nutrients such as NH₃-N from the sewage load (Hamza et al. 2022). The flexibility to control the treatment sequence in SBR promotes pre-anoxic conditions in the fill and decant phases, which results in better control of nitrification and denitrification processes for nitrate removal. Process change by promoting the pre-anoxic and post-anoxic intermittent cycles over controlling the DO level in the reactors, proper return activated sludge (RAS) procedure, and additional carbon source for nutrient removal in EAAS, OP, CAS, and AL is very important to promote higher nutrient removal in the systems (Metcalf & Eddy 2003).

Combining the alternative Niku's probabilistic reliability-based method and the process capability of Six Sigma, they herein integrate the holistic assessment of the compliance of 15 sub-urban STPs with Malaysian discharge requirements. This study showed that the log-normal distribution consistently gave a good overall fit to observe effluent parameters as compared to actual data for reliability and process capability assessment. This integrated analysis allows the author to estimate the STPs’ reliability, process capability, treatment stability, and performance of the compliance rate at a given reliability, which is key information to trigger decision-making on operational improvements. In addition, the basic purpose of the reliability and process capability analyses is to generate information that can be used by the authorities, plant designers, and plant operators of STPs in evaluating and predicting the performance of treatment processes. Moreover, these innovative methods help the regulatory agencies in establishing more reasonable, effective and suitable discharge standards in the future. The introduction of stochastic concepts into the design process and selection of treatment systems based on probabilistic considerations would be beneficial and would avoid the authority to establish unrealistic parameter values. The analyses of annual BOD₅, COD, TSS, O&G, and NH₃-N from 15 full-scale STPs in Melaka, Malaysia have resulted in the following general findings.

The highest reliability values were obtained by the SBR system, which presented a high mean value above 95%, followed by the EAAS system for BOD₅, COD, TSS, O&G except for NH₃-N. Despite that, the reliability and process capability for those highly mechanical systems (SBR & EAAS) were not statistically significant, except for the O&G Ppk, O&G Cpk, and NH₃-N Cpk parameters. Thirdly, in the BOD₅, COD, and O&G treatment, the CAS system performed better than OP and, conversely, in the NH₃-N treatments. OP system performed below 95% of its reliability level for TSS and O&G treatment, which indicates the conventional ponding system faces a carry-over problem regarding volumetric and surface hydraulic rates in the operational stage. Periodical maintenance, such as pond desludging activities, becomes critical for those ageing ponds in the system. Finally, the worst results in terms of organic and nutrient removal (BOD₅, COD, and NH₃-N) compliance were obtained by the AL system. The observed poor performance in terms of nutrient treatment was expected, since the AL system has not been designed for nitrogen removal. Proper development planning is required for the Malaysian government to revamp the wastewater industry according to SDGs and domestic WST 2040 targets. Notwithstanding, the poorest organic removal performance for BOD₅, and COD indicates the DO in the reactors is not optimal for the biodegradation process. In conclusion, the aerator capacity and DO performance monitoring have become critical factors for plant operators.

In general, better treatment reliability presented in highly mechanical systems (SBR & EAAS) became the evident of stable process capability. However, the moderate NH₃-N treatment for the whole system should be triggered, to avoid eutrophication problems for the water bodies that receive the effluent. The limitations of this study only enabled the author to investigate a fifteenth Malaysian STP. More samples of real-running plants are required to extend the combination of COR and Six Sigma to establish the current scenario in the country. Proper dataset management is required prior to using data-driven analysis to investigate the plant's reliability. The combination of Niku's method and Six Sigma, principally to enhance the STP's performance, becomes a conceptual framework for wastewater management procedure, since the discharged standards solely depend on the effluent quality. Such efforts will also reduce the treatment plant operating costs, increase the level of environmental sustainability, and even promote social security and health.

The authors are grateful to the Department of Environment Malaysia, Indah Water Konsortium Sdn Bhd and Universiti Teknikal Malaysia Melaka (UTeM) for providing the information and funding the research via grant FRGS/1/2020/TK0/UTEM/02/42.

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

The authors declare there is no conflict.