Abstract

This paper analyses factors associated with bulking in 195 small scale wastewater treatment plants (WWTPs) in Estonia. Operational data from each plant were collected and analysed statistically. The key factors associated with bulking were infiltration into sewage pipes, the type and purpose of process reactor, operational practices and influent characteristics. Both anaerobic fraction and volumetric fraction of the anaerobic reactor compared to the aerobic reactor resulted in a positive correlation with sludge volume index (SVI) <150 ml/g values. Good operation and maintenance practice as well as an operator's competence play a crucial role in bulking prevention. Using the 30 minute settling test (V30) as the single process control parameter can mislead an operator's judgement in process control strategies and cause effluent violations. Misjudgements in process control decisions can lead to unwanted conditions in small WWTPs (e.g. excessive chemical addition favoured bulking). Use of instrumentation, control and automation helped to keep the process conditions more stable and reduce the probability of bulking. Analyses of variance showed that the factors associated with Microthrix parvicella growth were long solids retention time (SRT), low food-to-microorganism ratio (F/M) and lack of carbon content compared against nitrogen and phosphorus contents.

INTRODUCTION

The main goal of wastewater treatment is to reduce the concentration of pollutants in the effluent below the admissible level in order to eliminate the threat to the environment or to public health. Regardless of a large variety of wastewater treatment technologies (e.g. activated sludge (AS), biofilm or constructed wetlands), the performance of wastewater treatment plants (WWTPs) depends on various technical and non-technical factors such as characteristics of influent wastewater and how well these factors are in accordance with the designed treatment process, operational and management practices, reliability of equipment and flexibility of the process. In many cases, the process parameters in a WWTP can be modelled in detail based on designed values, characteristics of influent or even on the basis of metabolic reactions of specific groups of microorganisms (Henze et al. 2008). However, the actual situation at the plant can differ dramatically from the modelled result due to incorrect design parameters, inevitably fluctuating input parameters, equipment failure, maintenance requirements or sludge bulking and foaming.

Operators in small WWTPs often do not have resources for analysing mixed liquor suspended solids (MLSS). The determination of the 30 minute settling test (V30) is the most commonly used operational parameter for AS process control in small scale (less than 50,000 PE) WWTPs in Estonia (Kõrgmaa et al. 2016). Unfortunately, if MLSS is not analysed, decisions based on the settling test could lead to inadequate sludge wasting and cause deviation from the targeted solids retention time (SRT). Sludge volume index (SVI), calculated as a quotient between V30 and TSS, has an impact on the good performance of the final clarifier (Jenkins et al. 2004) and might cause deterioration from effluent quality limits. According to Bitton (2005), high SVI values (SVI >150 ml/g) can be associated with bulking sludge. The performance of an AS system for biological wastewater treatment is often deteriorated due to sludge separation problems caused by sludge bulking (Guo et al. 2014).

The common practice in describing reasons for bulking has been towards isolating a single cause, evaluating its impact and finding a solution for its removal. Unfortunately, bulking is often a result of several favourable factors happening at the same time (Guo et al. 2012). Bulking consists of filamentous bulking due to excess proliferation of filamentous bacteria (Eikelboom 2000) and non-filamentous bulking resulting from certain microbes that produce large amounts of extracellular material (Jenkins et al. 2004). There are several reasons for filamentous bulking, such as factors like low dissolved oxygen (DO) concentrations (Jenkins et al. 2004; Martins et al. 2004; Gerardi 2008), nutrient deficiency (Vaiopoulou et al. 2007; Gerardi 2008), low temperatures (Rosetti et al. 2005), low pH (Glymph 2005; Gerardi 2008) and septic wastewater (Glymph 2005; Gerardi 2008). Operational parameters like long SRT and low food-to-microorganism (F/M) values favour bulking (Guo et al. 2012; Li et al. 2016) even in AS systems where anaerobic selectors are present. In order to obtain well-settling sludge, the ratio of biochemical oxygen demand (BOD) to nitrogen (N) to phosphorus (P) in influent should generally satisfy 100:5:1 (Eikelboom 2000; Guo et al. 2014).

Research of AS bulking and foaming problems has a long history, even though the operational conditions under which bulking sludge occurs are usually only marginally documented (Martins et al. 2004). Although poor quality of influent creates prerequisites for bulking, it is important not to underestimate the operating conditions. The performance of WWTPs depends on various technical and non-technical factors such as characteristics of influent wastewater and how well these are in accordance with the designed treatment process, operational and management practices, reliability of equipment and flexibility of the process (Hegg et al. 1979; Olsson 2012; Hao et al. 2013). According to a survey in US municipal WWTPs, the average facility had 15 performance-limiting factors (e.g. infiltration, process controllability, equipment accessibility for maintenance) and at no facility was one single factor observed to be limiting the performance (Hegg et al. 1979).

This paper focuses on performance-limiting factors associated with bulking sludge. This study is based on the national survey of 245 small and medium size WWTPs that were studied during 2014–2015 and evaluated according to a novel method for rapid assessment of the performance and complexity of small WWTPs. This paper focuses only on the findings from AS systems (n = 195).

MATERIALS AND METHODS

Method for evaluating performance

The overall assessment of the treatment performance of WWTPs was based on the performance of individual treatment phases, but unlike that described by Chen et al. (2015), the individual treatment steps had to be evaluated separately due to the great variance in technologies used and environmental objectives set for the evaluated WWTPs. For the purpose of ensuring comparability of individual WWTPs, the following prerequisites were set:

  • The steps of treatment processes (primary, secondary and tertiary treatment) are characteristic for all WWTPs.

  • Different equipment and processes have the same function in the same treatment step (e.g. the bar screen and screw screen are both devices for removing particles from wastewater).

  • All the processes and equipment having the same purpose at the same treatment step have to be comparable by setting specific critical control points (CCPs) for each treatment step.

A questionnaire was developed to assess WWTPs in situ. The questionnaire was divided into five main categories and 21 subcategories. For each subcategory (treatment step), a minimum of four CCPs were set. Depending on the complexity of the wastewater treatment process, the number of CCPs to be evaluated varied between 37 (two treatment steps − septic tank and oxidation ponds) and 170 (nine treatment steps − primarily treated wastewater was divided into two parallel treatment lines using an AS process or sequencing batch reactor (SBR), and the sewage sludge was stabilized on site). To overcome the problem of assessment subjectivity, all the CCPs were formulated as questions containing the choice of answers, which was set between two to five variables, mainly in the form of ‘yes’ or ‘no’.

Evaluations of performance and complexity in each treatment step were established on the basis of evaluation of the CCPs in the same step (Kõrgmaa et al. 2019). In this paper, CCPs are defined as factors that (a) influence performance of the treatment step (e.g. growth of filamentous microorganisms), (b) describe operational conditions (e.g. the surface of the final clarifier is kept clean, pumps are in working order) or (c) describe the complexity of the treatment step (e.g. screenings are pressed and washed). For each wastewater treatment step, a minimum of two CCPs were defined. All CCPs were chosen by a group of experts according to the literature (Kuusik et al. 2001; Jenkins et al. 2004; Maastik et al. 2011; Baumann et al. 2012).

Data collection, sampling and laboratory analyses

Most of Estonia is sparsely populated and, as a result, there are 664 municipal WWTPs for 1.35 million people (VEKA 2017). In total, 195 small scale (less than 50,000 PE) AS WWTPs were assessed in Estonia according to a uniform method over a period of 6 months between October 2014 and March 2015 (Figure 1). In total, 479 grab samples were collected, of which 193 samples were taken from the effluent of the secondary treatment and 94 from the effluent of tertiary treatment units. One hundred and ninety-two samples were collected for determination of MLSS. Effluent samples were not collected from two WWTPs, as there was no outflow during the plant visit. Analysis results from a national monitoring program were used for these two WWTPs. In 15 WWTPs, composite samples from influent and effluent were collected and the microscopic examination of AS was carried out.

During the investigation, the following actions were performed: (a) design parameters (e.g. flow rate, SRT) were collected; (b) the actual situation of the plant (e.g. flow rate, SRT, SVI, equipment failures) was documented (taking photos, filling Excel sheets of the model); (c) samples from effluent and process reactors were collected to determine biological oxygen demand (BOD7), chemical oxygen demand (COD), total suspended solids (TSS), total nitrogen (TN) and total phosphorus (TP). The following parameters: pH, conductivity (K), DO and water temperature (To) were determined in situ. In addition to the data needed for evaluation of the WWTPs' performance, some additional data was collected during the plant inspection: (a) the operators' knowledge about the process and evaluation of the operators' competence and (b) additional data that was not used in any assessment, but was expected to be relevant in the interpretation of results.

All the wastewater samples were collected according to ISO 5667-10. Wastewater samples were stored and transported to the accredited laboratory according to ISO 5667-3 and immediately analyzed according to the standard methods in the laboratory. Due to the financial limits, influent samples were not collected from all WWTPs and the data on influent quality was gathered from water companies using their analytical results. Statistical analyses do not include these results, as influent analyses did not often reflect an actual situation in the plant during the visit.

Statistical analyses

For studying the relationship between the plant performance, SVI and WWTP complexity, tools of correlation and regression analysis were applied. Together with the Pearson coefficient, the Spearman correlation coefficient was also applied in those cases, where the dependence between study variables was of the monotonic type instead of linear. For categorical variables, analysis of variance (ANOVA) was used for comparing the population means in different groups.

In order to find the frequency of coexisting problems in studied WWTPs (see Figure 2), matrix A was formed in such a way that findings of performance limiting factors in each WWTP were marked as ‘1’ (the problem was recorded) and ‘0’ (the problem was not recorded). Matrix B was formed as a multiplication of matrices A and its transposed matrix AT as follows: 
formula
Figure 1

Location and size of evaluated wastewater treatment plants in Estonia.

Figure 1

Location and size of evaluated wastewater treatment plants in Estonia.

Figure 2

Frequency of co-existence of selected problems in Estonian WWTPs (n = 190).

Figure 2

Frequency of co-existence of selected problems in Estonian WWTPs (n = 190).

Matrix B describes the co-existence of two problems (e.g. in total there were 63 WWTPs where SVI >150 ml/g and design loading was not based on actual measurements in 88 WWTPs; there were 25 cases where both problems co-existed). To find the frequency of co-existence of selected issues (matrix C), as described in Figure 2, all values (b) in each column in matrix B were divided by the total number of problems (x) that have been described in said column (e.g. SVI >150 ml/g was observed in 28% of WWTPs, where the design loading was not based on actual measurements, but measurements were absent in 40% of WWTPs, where SVI >150 ml/g was observed) as follows: 
formula

RESULTS AND DISCUSSION

Most common problems in Estonian activated sludge plants

A wastewater treatment plant can produce good quality effluent even if there are several shortcomings. All the plants (n = 195) visited during this study had some kind of problems, but the severity of these problems varied to a great magnitude. Most common issues were associated with effluent quality (56.4% of WWTPs had high TSS in the effluent), aeration problems (57.9%), foaming (38.1%), bulking (32.3%) and ensuring anaerobic (8.2%) and anoxic (15.4%) conditions in said reactors.

During the assessment it was observed that five WWTPs were at the stage of start-up of the process due to the loss of AS by serious hydraulic overloading. These plants were excluded from further analyses. The main approach for finding reasons for bulking is to identify the specific filamentous bacterium in the bulking sludge (Martins et al. 2004). Microscopic examination of AS was performed only in 15 WWTPs (Microthrix parvicella was the dominant filamentous organism in 61.5% of samples), but as the reasons for foaming and bulking in the rest of the WWTPs were not known, statistical analyses in this paper focused mainly on CCPs.

Some of the factors triggering bulking are not easily traced (e.g. low DO concentrations could be the result of sudden load of readily biodegradable organic matter as well as lack of aeration capacity or poor maintenance of the aeration system or a combination of said factors), good design and construction quality as well as professional O&M practice can minimize the probability of bulking or its impact on effluent quality. In order to understand the complexity of the actual situation in the WWTP that could trigger the disturbances in the normal wastewater treatment process and favour bulking, most common problems and their coexistence in each WWTP were assessed.

Figure 2 gives an overview of the frequency of coexistence of the selected problems (n = 20) in investigated AS plants. In total, 40 problems were initially analysed and similarly to the situation reported by Hegg et al. (1979), an average Estonian AS WWTP had 14.6 ± 3.9 issues. It can be seen from Figure 2 that if certain problems exist in the WWTP, the probability for the connected issue could be high. For example, column A15 shows that on 67% of occasions when the WAS was not removed, SVI >150 ml/g was observed, meanwhile the event of bulking could have several other initiators (only in 13% of WWTPs, where SVI >150 ml/g, was WAS not removed).

Although only 10 factors give significant (p-value <0.05) correlation with ‘bad’ SVI values (Table 1), the coexistence of multiple factors, as shown in Figure 2, does not necessarily mean the occurrence of bulking. For example, in 98 WWTPs the balancing tank was absent and 112 WWTPs had extreme peak flows, 78 WWTPs had F/M greater than 0.15 g BOD7/g MLSS. There were 28 plants with combinations of all said problems. In all these 28 plants, SRT was less than needed, but only eight of them had SVI >150 ml/g. Figure 2 allows assessment of the possibility of bulking. For example, 42.4% of plants that had infiltration to the sewer system had SVI >150 ml/g and 66.7% of WWTPs, where waste AS was not removed, had the same issue.

Table 1

Factors that had a statistically significant (p-value <0.05) or important (p-value <0.10) correlation with ‘good’ SVI

Parameter Pearson
 
Spearman
 
p-value Ρ p-value 
Performance of biological reactor (on the 10p scale) 0.247 0.001 0.234 0.003 
30 minute settling test (V30), ml −0.244 0.002 −0.226 0.004 
Is the real mass surface loading rate in the final clarifier less than 500 kg/(m2·d)? −0.215 0.012 −0.215 0.012 
Industrial sources in the influent cause problems to the WWTP −0.380 0.022 −0.380 0.022 
MLSS, g/l −0.307 0.021 0.168 0.031 
Effluent TSS, mg/l 0.096 0.226 0.163 0.038 
Does an operator measure TP for process control? −0.161 0.039 −0.161 0.039 
Effluent TN, mgN/l 0.185 0.018 0.162 0.039 
Is effluent quality within limits? −0.146 0.061 −0.146 0.061 
Are there any hydraulic problems in the final clarifier? 0.160 0.062 0.160 0.062 
Is biological P-removal possible? −0.148 0.064 −0.148 0.064 
SRT(real)/SRT(designed), d/d −0.088 0.358 0.172 0.071 
Real SRT, d −0.117 0.204 0.172 0.071 
Performance of final clarifier (on the 10p scale) −0.103 0.212 −0.148 0.083 
Are WAS pumps working? 0.087 0.293 0.145 0.091 
Is there any infiltration to the sewer system? −0.132 0.093 −0.132 0.093 
Volumetric fraction of anaerobic reactor (AN/OX)a 0.304 0.080 0.282 0.106 
Anaerobic fraction (fAN)a 0.313 0.071 0.264 0.132 
Effluent TP, mgP/l 0.139 0.077 0.051 0.496 
Parameter Pearson
 
Spearman
 
p-value Ρ p-value 
Performance of biological reactor (on the 10p scale) 0.247 0.001 0.234 0.003 
30 minute settling test (V30), ml −0.244 0.002 −0.226 0.004 
Is the real mass surface loading rate in the final clarifier less than 500 kg/(m2·d)? −0.215 0.012 −0.215 0.012 
Industrial sources in the influent cause problems to the WWTP −0.380 0.022 −0.380 0.022 
MLSS, g/l −0.307 0.021 0.168 0.031 
Effluent TSS, mg/l 0.096 0.226 0.163 0.038 
Does an operator measure TP for process control? −0.161 0.039 −0.161 0.039 
Effluent TN, mgN/l 0.185 0.018 0.162 0.039 
Is effluent quality within limits? −0.146 0.061 −0.146 0.061 
Are there any hydraulic problems in the final clarifier? 0.160 0.062 0.160 0.062 
Is biological P-removal possible? −0.148 0.064 −0.148 0.064 
SRT(real)/SRT(designed), d/d −0.088 0.358 0.172 0.071 
Real SRT, d −0.117 0.204 0.172 0.071 
Performance of final clarifier (on the 10p scale) −0.103 0.212 −0.148 0.083 
Are WAS pumps working? 0.087 0.293 0.145 0.091 
Is there any infiltration to the sewer system? −0.132 0.093 −0.132 0.093 
Volumetric fraction of anaerobic reactor (AN/OX)a 0.304 0.080 0.282 0.106 
Anaerobic fraction (fAN)a 0.313 0.071 0.264 0.132 
Effluent TP, mgP/l 0.139 0.077 0.051 0.496 

aCorrelations calculated without certain package plants (n = 10). The reason for excluding these WWTPs was that these plants had a very big anaerobic reactor (fAN = 0.5) that was not suitable for enhanced biological phosphorus removal (Gašparikova et al. 2005).

Factors correlating with SVI

SVI is calculated as a quotient between V30 and MLSS, but statistical analyses revealed that there were statistically significant, but weak relationships between SVI and V30 (Pearson's r = 0.215, p-value = 0.004) and between SVI and MLSS (Pearson's r = −0.301, p-value = 4.11e−5), if looked at separately. This suggests that operational decisions based only on the measurement of sludge settleability can mislead the operator's judgement in process control strategies.

For statistical analyses, the SVI was divided into two groups with ‘good’ (SVI <150 ml/g) and ‘bad’ settleability (SVI >150 ml/g) and was evaluated with ‘1’ or ‘0’ respectively. All questions regarding CCPs were formed to be answered as ‘yes’ or ‘no’ and were evaluated with ‘1’ or ‘0’ respectively. Further analyses showed correlations as described in Table 1. Settleability was influenced by biological phosphorus removal, performance of process parts and influent sources.

Table 1 shows that the factors correlating with SVI could be divided into two groups: (a) factors that affect SVI (e.g. WAS removal, usage of remote control) and (b) factors that are affected by SVI (e.g. performance of the final clarifier, effluent quality). A negative correlation shows that an occurrence of the factor induces the bulking (e.g. if there are industrial sources in the influent, the bulking is more likely to occur).

The reasons for bulking could be divided into three groups: (a) influent characteristics, (b) design and construction (D&C) and (c) operation and maintenance (O&M). While influent characteristics involve factors like nutrient deficiency, low temperature and pH that have been reported to be the reasons for bulking (Eikelboom 2000; Jenkins et al. 2004; Gerardi 2008), the factors resulting from misgivings in the fields of D&C or O&M are sparsely reported.

Design and construction

Design and construction (D&C) shortcomings are not always easily found and need an inspection to be assessed. Even if expertise has been used, the correlation between D&C issues and bulking is not often reported. It can be also debated whether the simplicity of the process as a consequence of minimal investment possibilities could be the main reason for bulking (e.g. the absence of bypasses or grease separators). Absence of grease separators is a common problem in small WWTPs. Nielsen et al. (2002) reported that Microthrix parvicella took up oleic acid under both anaerobic and aerobic conditions, while only a few floc formers were able to take it up under anaerobic conditions.

One-way ANOVA showed that the statistically relevant (p-value <0.05) possible causes for bulking were (a) the type of biological reactor used, (b) infiltration to the sewer system and (c) use of phosphorus removal (bulking was observed at 45.9% of AS plants with bio-P). The choice of reactor type was important (p-value 0.07) if compared against ‘good’ SVI with one-way ANOVA. 69.8% of plug-flow and SBR systems had SVI <150 ml/g. Meanwhile, only 56.3% of continuous stirred-tank reactors (CSTR) had similar SVI values. The balancing tank is usually needed to reduce variations in wastewater flow and concentrations. In many cases it helps to reduce the effect of peak flows and the risk of hydraulic overloading. In small WWTPs, where the balancing tank was absent, the average SVI value was 134.7 ml/g (n = 90), whereas in WWTPs with a balancing tank the average SVI value was 173.0 ml/g (n = 57). The effect was significant (p-value 0.05), but could be misleading as an average HRT in balancing tanks was 1.6 d−1. This period could be long enough for wastewater septicity to develop.

Selectors are usually considered to be an effective way of bulking control, but they do not work for all filamentous microorganisms (Martins et al. 2004). Due to the high requirements of effluent quality, 65.7% of Estonian AS WWTPs are designed with the possibility for biological nitrogen removal and 20.0% of plants have enhanced biological phosphorus removal. In total, there were 33 WWTPs included in this study with AAO configuration. Although one-way ANOVA showed that the volume fraction of an anaerobic reactor compared to volumes of aerobic (AN/OX) and/or anoxic reactors (AN/AX or AN/(AX + OX)) could play a crucial part in the probability of sludge bulking (Table 2), the data was influenced by certain package plants that have big anaerobic tanks which are not suitable for enhanced biological phosphorus removal (Gašparikova et al. 2005). When these plants were excluded from the ANOVA analyses, the p-value of an anaerobic volume fraction was 0.14 for AN/OX and 0.07 for AN/(AX + OX). An anaerobic fraction and volumetric fraction of anaerobic reactor compared to the aerobic reactor both give positive correlation with SVI <150 ml/g values, even without the mentioned package plants (for fAN Pearson's r 0.313, p-value 0.07 and for AN/OX Pearson's r 0.304, p-value 0.08). Analyses of hydraulic retention time in anaerobic and anoxic reactors revealed that in small WWTPs the contact time was much higher than recommended by Henze et al. (2008) with average values for anaerobic reactors of 10.5 ± 6.8 h and for anoxic reactors of 27.9 ± 24.0 h respectively.

Table 2

The design parameters of anaerobic reactors showed significant impact on the bulking according to the one-way ANOVA

Parameter Unit SVI <150 ml/g Variance SVI >150 ml/l Variance Number of WWTPs Mean square F-value P-value 
fOX – 0.772 0.038 0.739 0.032 188 0.045 1.250 0.265 
fAN – 0.320 0.052 0.312 0.063 37 0.001 0.011 0.916 
fAX – 0.355 0.019 0.363 0.019 122 0.002 0.101 0.751 
AN/OX m3/m3 0,400 0.135 0.187 0.011 37 0.428 5.352 0.027 
AN/OXa m3/m3 0.259 0.030 0.187 0.011 35 0.048 2.243 0.143 
AX/OX m3/m3 0.954 0.871 0.521 0.086 122 1.571 2.979 0.094 
AN/AX m3/m3 1.054 0.955 0.521 0.086 32 2.199 4.107 0.052 
AN/(AX + OX) m3/m3 0.291 0.066 0.135 0.005 32 0.223 5.820 0.021 
AN/(AX + OX)a m3/m3 0.195 0.013 0.135 0.005 29 0.031 3.392 0.074 
HRTAN d−1 0.399 0.051 0.369 0.058 32 0.007 0.131 0.719 
HRTAX d−1 1.240 1.277 1.091 0.624 113 0.596 0.582 0.447 
HRTOX d−1 3.717 14.186 2.478 3.174 165 58.671 5.752 0.018 
Parameter Unit SVI <150 ml/g Variance SVI >150 ml/l Variance Number of WWTPs Mean square F-value P-value 
fOX – 0.772 0.038 0.739 0.032 188 0.045 1.250 0.265 
fAN – 0.320 0.052 0.312 0.063 37 0.001 0.011 0.916 
fAX – 0.355 0.019 0.363 0.019 122 0.002 0.101 0.751 
AN/OX m3/m3 0,400 0.135 0.187 0.011 37 0.428 5.352 0.027 
AN/OXa m3/m3 0.259 0.030 0.187 0.011 35 0.048 2.243 0.143 
AX/OX m3/m3 0.954 0.871 0.521 0.086 122 1.571 2.979 0.094 
AN/AX m3/m3 1.054 0.955 0.521 0.086 32 2.199 4.107 0.052 
AN/(AX + OX) m3/m3 0.291 0.066 0.135 0.005 32 0.223 5.820 0.021 
AN/(AX + OX)a m3/m3 0.195 0.013 0.135 0.005 29 0.031 3.392 0.074 
HRTAN d−1 0.399 0.051 0.369 0.058 32 0.007 0.131 0.719 
HRTAX d−1 1.240 1.277 1.091 0.624 113 0.596 0.582 0.447 
HRTOX d−1 3.717 14.186 2.478 3.174 165 58.671 5.752 0.018 

aValues calculated without certain package plants. The reason for excluding these WWTPs was that an anaerobic tank used in these plants was not suitable for enhanced biological phosphorus removal.

While the impacts of reactor type and infiltration to the SVI could be explained by kinetic selection theory (e.g. infiltration causes the dilution of nutrients and therefore gives growth advantage to filamentous organisms with lower Ks values over floc-forming bacteria with higher Ks values), the role of the phosphorus removal process as the cause of bulking remains uncertain. Some of the possible explanations for connections between bulking and phosphorus removal have been presented by Nielsen et al. (2002, 2010) and Wang et al. (2014). Nielsen et al. (2010) suggested that soluble components, either from the wastewater or produced by hydrolysis in the anaerobic tank, are taken up for storage as polyhydroxyalkanoates (PHA)/lipids by the filamentous Microthrix parvicella, by PAOs (Accumulibacter), and by GAOs (Competibacter, Defluviicoccus). Wang et al. (2014) suggested that M. parvicella could take part in enhanced biological phosphorus removal. M. parvicella is the most dominant organism in Estonian WWTPs that is causing bulking.

Fan et al. (2017) reported that M. parvicella favoured lower temperature, alteration between anaerobic and aerobic conditions and long chain fatty acids (LCFA) in batch tests. In the anaerobic/aerobic alternation experiment reported by Fan et al. (2017), the AN/OX ratio was 0.5 which seems to be favouring SVI <150 ml/g. As shown in Table 2, their study still showed that this was enough to initiate M. parvicella bulking with the presence of LCFAs as the sole carbon source. Their experiments with different water temperatures (13 °C and 20 °C) combined with anaerobic-aerobic conditions and LCFA feed also showed great differences in M. parvicella abundance, favouring colder temperatures. In real conditions, the number of factors occurring simultaneously could be unlimited (Figure 2 and Table 1) and the event of bulking could be initiated or even suppressed. Still, the role of the volume fraction of an anaerobic reactor could be significant and needs further investigation.

Operation and maintenance

O&M is dependent on the human factor. The importance of the human factor in wastewater treatment process has been described very briefly in literature (Hegg et al. 1979; Olsson 2012) and in many cases it is considered to be the main reason for poor process performance (Hegg et al. 1979). Hegg et al. (1979) listed improper operator application of concepts and testing to process control as well as inadequate understanding of sewage treatment as the two highest ranking factors contributing to poor plant performance. The competence of the operator has been outlined as one of the key factors for successful plant control (Hegg et al. 1979; Muga & Michelic 2008; Olsson 2012). The competence of the operators was evaluated during the data collection in the form of a hidden test on a scale of 10 points. The average result was 7.31 and the minimum result 1.5. Although statistical analyses revealed that operators' competence did not influence the bulking directly, it could have severe consequences for the WWTPs' performance. For example, in one SBR that was treating wastewater from a dairy factory, an operator observed that the AS had poor settling properties and added external activated sludge with good settling properties into his WWTP. The settling did not improve as the operator had not removed any WAS and as a result MLSS was 12 mg/l during the visit.

Procedures for operation and maintenance constitute the key factors for successful pollution removal. Table 1 shows that O&M factors also have low correlation with the event of bulking; these factors are statistically important (p-value >0.10), but might also be misleading. Some of the statistically important factors (e.g. control of chemical precipitation, use of back-up generators for instrumentation, control and automation (ICA)) show negative correlations where the desired effect should be positive. In some cases, correlations shown in Table 1 can be also misleading without further analysis.

The chemical precipitation of phosphorus thickens and compacts the AS (Lind 1998), but it might also initiate toxic effects and control over chemical addition is crucial (Suresh et al. 2018). Table 1 shows negative weak correlation (Pearson's r = −0.161, p-value 0.039) between bulking and the operators' claim regarding adjusting chemicals for phosphorus removal. Further analyses showed that contradiction between the desired effect of chemical precipitation (less bulking) and the operators' claim was driven by the operators' tendency to add too much of chemicals. In WWTPs where operator claimed to be making adjustments, according to the real measurements the ratio of chemicals added to the amount actually needed was 1.6 ± 1.5, but in the other group the same ratio was 1.3 ± 1.2. Although the difference between groups was not statistically relevant (p-value 0.29), it shows that while operators make decisions on chemical addition based on effluent results (according to ANOVA, average TP was 2.4 mgP/l in the group that made adjustments against 4.3 mgP/l in the group that did not, p-value 0.02), they tend to add too much of iron salts and as a result it affects AS properties negatively.

ICA plays an important role in reaching operational goals (Olsson & Jeppsson 2006). 66.7% of studied WWTPs used ICA for process control. ANOVA showed that in WWTPs where ICA was used the average SVI was 136.0 ml/g, while in the other group it was 173.8 ml/g (p-value 0.02).

According to the ANOVA, there was no difference in SRT values if compared against plants with or without bulking, even though Spearman's P was 0.172 with a p-value of 0.071. Some of the filamentous organisms grow on a wide range of SRT (Jenkins et al. 2004; Martins et al. 2004). As microscopic examination was performed only in 13 WWTPs, it remains open how much bulking was influenced by SRT values.

Influent characteristics

Infiltration to the sewer system had a significant impact (p-value 0.03) on the SVI value. Average SVI value in WWTPs that had infiltration was 177.9 ml/g, while in the plants that did not have any infiltration it was 133.1 ml/g. Industrial sources caused problems in 35 WWTPs and the impact on the SVI value was important (p-value 0.07), causing bulking (an average SVI value of 180.4 ml/g). WWTPs without industrial sources had an average SVI value of 139.9 ml/g. Average wastewater temperature during the sampling session was 8.9 ± 3.3 °C and the impact of cold water temperature could not be evaluated as there was no adjacent group available.

It was not possible to analyse influent quality in all WWTPs due to the financial limitations, but 24 h composite samples were collected from 15 AS plants. Microscopic analyses of AS were performed in 13 WWTPs. Analyses showed that Microthrix parvicella was dominant in eight WWTPs, and three plants had foaming problems caused by Nocardioforms. In these 15 plants, no correlation between SVI and influent parameters was observed, but significant correlations were found with the presence of M. parvicella (Table 3). One-way ANOVA showed that M. parvicella favoured long SRT and low F/M values, as described earlier by Fan et al. (2018) and Jenkins et al. (2004). Table 3 shows that M. parvicella favoured wastewaters with lower BOD7, COD and TSS content and a lack of carbon content compared against nitrogen and phosphorus contents.

Table 3

One-way ANOVA showed significant (p-value <0.05) difference in process control parameters and influent characteristics in WWTPs where Microthrix parvicella was found

Parameter Unit With M. parvicella Variance Without M. parvicella Variance Number of WWTPs Mean square F-value P-value 
SRT 39.10 352.96 14.56 155.91 13 1,852.96 6.59 0.03 
F/M kgBOD7/ kgMLSS 0.033 2.721 × 10−4 0.107 0.005 13 0.017 8.10 0.02 
Filament index – 4.13 1.55 2.00 0.00 13 13.89 14.05 3.22 × 10−3 
BOD7 mgO2/l 350.00 5,314.29 576.00 13,480.00 13 157,156.92 18.97 1.14 × 10−3 
COD mgO2/l 632.50 42,221.43 1,012.00 79,320.00 13 443,139.23 7.95 0.02 
TSS mg/l 263.00 3,609.14 433.20 23,171.20 13 89,132.43 8.31 0.01 
BOD7/N – 4.15 1.41 6.07 2.37 13 11.33 6.44 0.03 
BOD7/P – 28.71 104.46 47.57 147.07 13 1,094.74 9.13 0.01 
Parameter Unit With M. parvicella Variance Without M. parvicella Variance Number of WWTPs Mean square F-value P-value 
SRT 39.10 352.96 14.56 155.91 13 1,852.96 6.59 0.03 
F/M kgBOD7/ kgMLSS 0.033 2.721 × 10−4 0.107 0.005 13 0.017 8.10 0.02 
Filament index – 4.13 1.55 2.00 0.00 13 13.89 14.05 3.22 × 10−3 
BOD7 mgO2/l 350.00 5,314.29 576.00 13,480.00 13 157,156.92 18.97 1.14 × 10−3 
COD mgO2/l 632.50 42,221.43 1,012.00 79,320.00 13 443,139.23 7.95 0.02 
TSS mg/l 263.00 3,609.14 433.20 23,171.20 13 89,132.43 8.31 0.01 
BOD7/N – 4.15 1.41 6.07 2.37 13 11.33 6.44 0.03 
BOD7/P – 28.71 104.46 47.57 147.07 13 1,094.74 9.13 0.01 

CONCLUSIONS

Usually it is not possible to point out only a single reason behind high SVI values. An average Estonian AS WWTP had 14.6 ± 3.9 issues, as described above, and it is common that several problems occur simultaneously. In real conditions, the number of factors occurring simultaneously could be unlimited and the event of bulking could be initiated or even suppressed by a combination of the said issues. The best result is achieved by combining stable influent characteristics with good design solutions, and excellent operation and maintenance practices.

Influent characteristics have a significant influence on filamentous growth. Microthrix parvicella dominated in 65% of WWTPs where microscopic examination was performed. ANOVA showed that factors triggering M. parvicella growth were long SRT, low F/M and lack of carbon sources compared against nitrogen and phosphorus content. Infiltration had significant correlation with bulking in all WWTPs.

The increasing need for nitrogen and phosphorus removal has evoked an even wider use of selectors. In order to avoid bulking, designers should consider the purpose and type of the reactor. The reactor type was important (p-value 0.07) if compared against ‘good’ SVI with one-way ANOVA. 69.8% of plug-flow and SBR systems had SVI <150 ml/g; meanwhile, only 56.3% of continuous stirred-tank reactors (CSTR) had similar SVI values. An anaerobic fraction and volumetric fraction of anaerobic reactor compared to the aerobic reactor resulted both in a positive correlation with SVI <150 ml/g values (for fAN Pearson's r 0.313, p-value 0.07 and for AN/OX Pearson's r 0.304, p-value 0.08). Analyses of hydraulic retention time in anaerobic and anoxic reactors revealed that in small WWTPs the contact time was much higher than recommended by Henze et al. (2008), with average values for an anaerobic reactor of 10.5 ± 6.8 h and for an anoxic reactor of 27.9 ± 24.0 h respectively. The role of the volume fraction of the anaerobic reactor could be significant and needs further investigation.

Good operation and maintenance practice as well as operators' competence plays a crucial role in bulking prevention. Using V30 as the only process control parameter can mislead operators' judgement in process control strategies and cause effluent violations. Misjudgements in process control decisions can lead to unwanted conditions in small WWTPs (e.g. in order to reduce effluent phosphorus too much, chemicals were added and this favoured bulking). Use of ICA helped to keep process conditions more stable and reduce the probability of bulking.

In this study, we demonstrated, that this approach could be used more widely. Statistical analyses of operational conditions (including influent characteristics and identification of filamentous organisms) on the broad range of WWTPs could simplify ascertainment and impact the assessment of the factors that affect bulking.

REFERENCES

REFERENCES
Baumann
P.
,
Krauth
K.
,
Maier
W.
&
Roth
M.
2012
Operational Problems in Wastewater Treatment Plants. 1
, Vol.
3
.
DWA Landesverband
,
Stuttgart
,
Germany
.
Bitton
G.
2005
Wastewater Micobiology
.
John Wiley & Sons, Inc.
,
Hoboken
,
USA
.
Chen
Z.
,
Zayed
T.
&
Qasem
A.
2015
An efficiency-centred hierarchical method to assess performance of wastewater treatment plants
.
International Journal of Environmental Research
9
,
7
8
.
Eikelboom
D. H.
2000
Process Control of Activated Sludge Plants by Microscopic Investigation
.
IWA Publishing
,
London
,
UK
.
Fan
N.
,
Qi
R.
,
Rosetti
S.
,
Tandoi
V.
,
Gao
Y.
&
Yang
M.
2017
Factors affecting the growth of Microthrix parvicella: batch tests using bulking sludge as seed sludge
.
Science of the Total Environment
609
,
1192
1199
.
Fan
N.
,
Wang
R.
,
Qi
R.
,
Gao
Y.
,
Rosetti
S.
,
Tandoi
V.
&
Yang
M.
2018
Control strategy for filamentous sludge bulking: bench-scale test and full-scale application
.
Chemosphere
210
,
709
716
.
Gašparikova
E.
,
Kapusta
Š.
,
Bodik
I.
,
Derco
J.
&
Kratochvil
K.
2005
Evaluation of anaerobic-aerobic wastewater treatment plant operations
.
Polish Journal of Environmental Studies
14
,
29
34
.
Gerardi
M. H.
2008
Microscopic Examination of the Activated Sludge Process
.
John Wiley & Sons Inc.
,
Hoboken
,
USA
.
Glymph
T.
2005
Wastewater Microbiology. A Handbook for Operators
.
American Water Works Association
,
Denver
,
USA
.
Hao
R. X.
,
Liu
F.
,
Ren
H. Q.
&
Cheng
S. Y.
2013
Study of comprehensive evaluation method for the assessment of operational efficiency of wastewater treatment plants
.
Stochastic Environmental Research and Risk Assessment
27
,
747
756
.
Hegg
B. A.
,
Rakness
K. L.
&
Schultz
J. R.
1979
Evaluation of Operation and Maintenance Factors Limiting Municipal Wastewater Treatment Plant Performance
.
Report EPA-600/2-79-034
,
US Environmental Protection Agency
,
Springfield, Virginia
,
USA
.
Henze
M.
,
van Loosdrecht
M. C. M.
,
Ekama
G. A.
&
Brdjanovic
D.
, (eds)
2008
Biological Wastewater Treatment: Principles Modelling and Design
.
IWA Publishing
,
London
,
UK
.
Jenkins
D.
,
Richard
M. G.
&
Daigger
G. T.
2004
Manual on the Causes and Control of Activated Sludge Bulking, Foaming and Other Solid Separation Problems
,
3rd edition
.
CRC Press
,
Florida
,
USA
.
Kõrgmaa
V.
,
Tenno
T.
,
Gross
M.
,
Kriipsalu
M.
,
Kivirüüt
A.
,
Tamm
P.
,
Värk
V.
,
Karabelnik
K.
,
Terase
H.
,
Kuusik
S.
,
Leisk
Ü.
,
Sinikas
N.
,
Pitk
P.
,
Tõnisberg
E.
&
Maastik
A.
2016
Evaluation of Treatment Efficiency of Wastewater Treatment Plants, Constructed and Reconstructed in 2004–2014, Using Grants by the EU and Estonian Environmental Investment Centre
.
Report 4-1.1/14/90
,
Estonian Ministry of Environment
,
Tallinn
,
Estonia (in Estonian).
Kõrgmaa
V.
,
Tenno
T.
,
Kivirüüt
A.
,
Kriipsalu
M.
,
Gross
M.
,
Tamm
P.
,
Karabelnik
K.
,
Terase
H.
,
Värk
V.
,
Lepik
N.
,
Pachel
K.
&
Iital
A.
2019
A novel method for rapid assessment of the performance and complexity of small wastewater treatment plants
.
Proceedings of the Estonian Academy of Sciences
68
,
32
42
.
Kuusik
A.
,
Pachel
K.
,
Sokk
O.
,
Suurkask
V.
&
Kuusik
A.
2001
Draft of Technological and Technical Recommendations and Manuals of Small Wastewater Treatment Units for Local Municipalities
.
Tallinn University of Technology
,
Tallinn
,
Estonia (in Estonian)
.
Lind
C. B.
1998
Phosphorous inactivation in wastewater treatment: biological and chemical strategies
.
Water Engineering & Management
145
,
18
21
.
Maastik
A.
,
Danilišina
G.
,
Gross
M.
,
Kriipsalu
M.
,
Tamm
P.
&
Tenno
T.
2011
Maintenance and Instructions for Small (up to 2000 pe) Waste Water Treatment Plants
.
University in Tartu
,
Tartu
,
Estonia
(in Estonian).
Martins
A. M. P.
,
Pagilla
K.
,
Heijnen
J. J.
&
van Loosdrecht
M. C. M.
2004
Filamentous bulking sludge – a critical review
.
Water Research
38
,
793
817
.
Muga
H. E.
&
Michelic
J. R.
2008
Sustainability of wastewater treatment technologies
.
Journal of Environmental Management.
88
,
437
447
.
Nielsen
P. H.
,
Mielczarek
A. T.
,
Kragelund
K.
,
Nielsen
J. L.
,
Saunders
A. M.
,
Kong
Y.
,
Hansen
A. A.
&
Vollersten
J.
2010
A conceptual ecosystem model of microbial communities in enhanced biological phosphorus removal plants
.
Water Research
44
,
5070
5088
.
Nielsen
P. H.
,
Roslev
P.
,
Dueholm
T. E.
&
Nielsen
J. L.
2002
Microthrix parvicella, a specialized lipid consumer in anaerobic-aerobic activated sludge plants
.
Water Science & Technology
46
(
1–2
),
73
80
.
Olsson
G.
2012
ICA and me – a subjective review
.
Water Research
46
,
1585
1624
.
Olsson
G.
&
Jeppsson
U.
2006
Plant-wide control: dream, necessity or reality?
Water Science & Technology
53
,
121
129
.
Suresh
A.
,
Grygolowicz-Pawlak
E.
,
Pathak
S.
,
Poh
L. S.
,
Majid
M. b. A.
,
Dominiak
D.
,
Bugge
T. V.
,
Gao
X.
&
Ng
W. J.
2018
Understanding and optimization of the flocculation process in biological wastewater treatment processes: a review
.
Chemosphere
210
,
401
416
.
Vaiopoulou
E.
,
Melidis
P.
&
Aivasidis
A.
2007
Growth of filamentous bacteria in an enhanced biological phosphorus removal system
.
Desalination
213
,
288
296
.
VEKA
2017
Estonian Water Use Database
. .