The Air and Sewage Pollutants from Biological Waste Treatment

The mechanical-biological waste treatment plants (MBTP), which include the municipal waste biogas plants, have an important role in sustainable urban development. Some plants are equipped with a sewage pre-treatment plant, which is then directed to the sewerage system and the treatment plant. Others, on the other hand, have only a non-drainage tank. The parameters of technological sewage (TS) or processing technology could reduce sewage contamination rates. In addition to the quality of sewage from waste treatment plants, the emission of odours is also an important problem, as evidenced by the results obtained over the sewage pumping station tank. The conducted statistical analysis shows a significant positive correlation between odour concentration (cod) and volatile organic compounds (VOCs). Analysing the individual compounds, a high positive correlation was also found—the strongest being between H2S, NH3 and VOCs. In the case of sewage compounds, the insignificant correlation between P total and other parameters was found. For the rest of the compounds, the highest positive correlation was found between COD and BOD and N-NO2 and N-NH3 as well as COD and N-NO2. The dilution of sewage is only an ad hoc solution to the problem. Further work should be aimed at reducing sewage pollution rates. The obtained results indicate large pollution of technological sewage and a high level of odour and odorants concentration. The novelty and scientific contribution presented in the paper are related to analyses of various factors on technological sewage parameters and odour and odorant emission from TS tank at biogas plant processing municipal waste, which may be an important source of knowledge on the management of TS, its disposal and minimisation of emitted compound emissions.


Introduction
The MBTP, which include the municipal waste biogas plants, an important element of sustainable management for future generations and a circular economy [1][2][3][4], are essential from the point of view of renewable energy, but also from minimising the odour nuisance of waste management facilities (the encapsulation of the first, most odorogenic phase of the biological process) [5]. The anaerobic digestion and composting are recommended for waste treatment processes, mainly for the biodegradable waste collected at the source. Both processes aim to convert waste into the least harmful form for the environment. The anaerobic digestion process is particularly proposed as an environmentally friendly and more cost-effective alternative to treating both household waste and waste activated sewage [6][7][8][9][10][11][12]. Nkoa [13] wrote that the digestate can cause, inter alia, nitrate leaching and ammonia emissions into the atmosphere. However, the inherent impact of waste treatment is the emission of the odorant compounds, but also of the production of technological sewage with a potentially high pollutant load [14,15]. The olfactory compounds characteristic of the waste management includes mainly VOCs (sulphur-containing VOCs, volatile fatty acids, phenolics and indolics), ammonia (NH 3 ) and hydrogen sulphide (H 2 S) [16,17].
The TS is the leachate from individual waste treatment processes, e.g., treated at landfills, MBP plants and biogas plants, usually connected with rainwater. Technological Olfactometer Nasal Ranger is a lightweight, portable device with two replaceable filter cartridges with activated carbon for air purification. It includes a built-in channel system for mixing and sharing gas streams-deliberate targeting is known part of the inhaled air by bypassing filters. The control valve is used to adjust one of the eleven values of D/T (2,4,7,15,30,60,100,200,300,400,500) and to set the value of "blank", at which the researcher breathes by purified air stream [28,31].
Scentroid SM 100 is a field olfactometer, using compressed air from the cylinder under high pressure (31 MPa) to dilute the test sample. The apparatus consists of a dilution valve control. Its high accuracy is used to provide a constant flow of diluted air through the device which allows the user to select one of the 15 positions, which correspond to ratios of clean air to the dilution of the test air sample. The range of the device is between 2 and 30,000 ou/m 3 , and the detection limit of the olfactometer is 3 ou/m 3 [32,33].
The TS parameters were determined using the following methods  [40].
The independent samples t-tests-Student, Welch and Mann-Whitney-were used to assess whether the means of two populations are equal to each other. To check assumptionsnormality and equality of variances-the Shapiro-Wilk's test and the Levene's test were made. To check correlations between examined variables, Pearson's r, Spearman's rho and Kendall Tau B coefficients were calculated. Furthermore, linear regression and Bayesian regression models were made. It contained four predictors with odour concentration as a dependent variable. Those regression analyses resulted in a hypothetical model of the relationship between the outcome and predictor variables. Bayesian linear regression model uses probability distributions rather than point estimates-its response is assumed from a probability distribution.

Characteristic of Analysed Plants
Both MWTP, which are the subject of the research, are mechanical-biological treatment installations of municipal waste which are equipped with biogas installation and methane fermentation in the biological part. The feedstock for the fermentation chambers at A plant is a fraction of biodegradable waste separated mechanically from the mixed waste stream and at plant B it is biodegradable waste selectively collected. The fermentation process is carried out under dry mesophilic conditions at both plants. The input material in the design assumption should be anaerobically treated for 21 days. At plant A, after this period, the digestate should be stabilised under aerobic conditions in an aeration chamber for 14 days and then subjected to a second-stage aerobic stabilisation at the ripening site for four weeks. Due to the limitations of the area of the technological yard on the premises of the plant, as well as a large amount of delivered waste, the plant periodically operates in "emergency mode". During these periods, the digestate is not always subjected to first and/or second-degree aerobic stabilisation after the end of the fermentation process but is sent to landfill. The measurements were carried out in the tank where flow sewage from stored, stabilized and composted waste and rainwater. This tank is a non-drainage tank, requiring periodic emptying and transport to the drainage station.
At plant B, after the first stage of the biological process (21 days), the digestate is directed to the processing site and there it undergoes oxygen stabilisation (approx. 28 days). The TS from the technological yard is directed by gravity to the collective sewage system (together with rainwater and domestic wastewater) and then is directed to a sewage treatment plant.
Sewage parameters were determined under laboratory conditions after earlier sampling directly from sewage tanks. The parameters of emitted gases were determined in field conditions using direct measurements. The measurements were taken from the sewage tanks.  Table 2 contains independent sample T-test results of air compounds determinations, while Table 3-its assumption checks.

Air Compounds
The difference between the groups is statistically significant at the 0.05 level for NH 3 and VOCs. Cohen's d was used as an effect size statistic for a paired t-test. It is calculated as the difference between the means of each group, all divided by the standard deviation of the data. The effect size was medium for c od and CH 3 SH and large for the rest of the effects.
The assumption checks were statistically significant in most cases, except the normality test of VOCs in plant B and the test of variances equality for CH 3 SH. For that last parameter, we consider the Welch version of the t-test, because the Welch version does not assume that the variances in the two groups are equal. Therefore, p values calculated by both t-test versions were more than 0.05. Furthermore, the equivalence independent samples t-test was made, which allows one to test the null hypothesis that the population means of two independent groups fall inside a by the user-defined interval. This procedure follows the two-one-sided tests (TOST). Only when both the upper bound and the lower bound statistic are rejected, the initial non-equivalence hypothesis is rejected-in the present study, that situation was only in one case-CH 3 SH. In the rest cases, p-value for the lower bound test was <0.05, so the effect was smaller than or equal to the lower bound. Figure 1 contains the correlation plot of air components, with Pearson's r, Spearman's rho and Kendall Tau B coefficients.  The Pearson correlation coefficient is used to assess the linear relationship between the two variables. Kendall Tau B measures the monotonic relationship. While Kendall's Tau B is to be interpreted in terms of probability, Spearman's rho is to be interpreted in terms of the percentage of the variance of the rank of one variable explained by the other. For this reason, all three coefficients have been considered in these analyses. Almost all correlations are significant at alpha = 0.001 level except the correlation between CH 3 SH and VOCs, which is significant at alpha = 0.05 level. Therefore, all of them are significant.
only in one case-CH3SH. In the rest cases, p-value for the lower bound test was <0.05, so the effect was smaller than or equal to the lower bound. Figure 1 contains the correlation plot of air components, with Pearson's r, Spearman's rho and Kendall Tau B coefficients. The Pearson correlation coefficient is used to assess the linear relationship between the two variables. Kendall Tau B measures the monotonic relationship. While Kendall's Tau B is to be interpreted in terms of probability, Spearman's rho is to be interpreted in terms of the percentage of the variance of the rank of one variable explained by the other. For this reason, all three coefficients have been considered in these analyses. Almost all correlations are significant at alpha = 0.001 level except the correlation between CH3SH and VOCs, which is significant at alpha = 0.05 level. Therefore, all of them are significant.
Considering values of Pearson's r, Spearman's rho and Kendall Tau B coefficients, correlation between odour concentrations and main odorants is high, between 0.7 and 0.9 (Pearson's r), 0.8 and 0.9 (Spearman's rho) as well as 0.7 and 0.8 (Kendall's Tau B). Correlation between odour concentration and VOCs is lower, but still high or moderate: 0.8 (Pearson's r), 0.7 (Spearman's rho) and 0.6 (Kendall's Tau B). Furthermore, considering Pearson's r, there is a very high correlation between H2S and NH3 and VOCs-coefficients are 0.9. Furthermore, correlation coefficient values are more than 0.9 in case of NH3 and Considering values of Pearson's r, Spearman's rho and Kendall Tau B coefficients, correlation between odour concentrations and main odorants is high, between 0.7 and 0.9 (Pearson's r), 0.8 and 0.9 (Spearman's rho) as well as 0.7 and 0.8 (Kendall's Tau B). Correlation between odour concentration and VOCs is lower, but still high or moderate: 0.8 (Pearson's r), 0.7 (Spearman's rho) and 0.6 (Kendall's Tau B). Furthermore, considering Pearson's r, there is a very high correlation between H 2 S and NH 3 and VOCs-coefficients are 0.9. Furthermore, correlation coefficient values are more than 0.9 in case of NH 3 and VOCs. Considering Spearman's rho coefficient, correlations are moderate or high in all cases. The highest-more than 0.9 values of the correlation coefficient are for CH 3 SH and H 2 S as well as NH 3 and VOCs. In the case of Kendall's Tau B coefficient, its value is more than 0.9 for H 2 S and CH 3 SH. In rest cases, correlation is from moderate to high-the lowest (0.5) is for VOCs and c od , H 2 S and CH 3 SH. All of p-values of the Shapiro-Wilk test for bivariate normality were <0.001.
The linear regression model-which contains four predictors with c od as the dependent variable-was made. The model coefficients are in Table 4. The R 2 factor, 0.92, is high and the adjusted R 2 , 0.90, drops only a little, showing robust model (probably not very high overfitting). The test of the fit of the model was also prepared. The explained variance of the model is statistically highly significant (p < 0.001). Therefore, the variance inflation factor is more than 10 for NH 3 (24.6) and H 2 S (22.5), so the multicollinearity in an ordinary least square is high. Furthermore, a Bayesian Linear Regression was made (Table 5). The posterior model probabilities express the probability of a model after seeing the data. The Bayes factor quantifies the data-induced change from prior model odds to posterior model odds. The prior probability of the respective models was between 0.033 and 0.200-the highest coefficient was in case of all four covariates. The maximum posterior model probabilities and the Bayes factor for the model were, respectively, 0.336 and 14.7-for NH 3 + H 2 S variant. Maximum R 2 coefficient was calculated for the sum of all four factors (R 2 = 0.922). Table 6 contains independent sample T-test results of air compounds determinations, while Table 7-its assumption checks.

Technological Sewage Compounds
The difference between the groups is statistically significant at the 0.05 level for all of compounds, except P total. The effect size statistic for a paired t-test was medium for P total and large for the rest of effects. The assumption checks were statistically significant in most cases, except test of normality of solids and N-NH 3 at plant A and test of variances equality for P total at plant B. Therefore, for that last parameter, we consider the Welch version of the t-test. According to the results of independent samples t-test for sewage results, reject initial non-equivalence hypothesis was rejected in two cases: P total and c od . In the rest cases, the effect was smaller than or equal to the lower bound.   Figure 2 contains the correlation plot of technological sewage components, with Pearson's r, Spearman's rho and Kendall Tau B coefficients. P total was excluded from analysis since correlation coefficients were between −0.013 and 0.075. All correlations of analysed components are significant at alpha <0.05 level (most of them <0.001 level). The highest correlation is between COD and BOD-Pearson's r correlation coefficient is 0.98. There is also a very high correlation between N-NO 2 and N-NH 3 (Pearson's rho 0.89) as well as COD and N-NO 2 (0.84). Therefore, Pearson's r correlation between solids and COD as well as N-NH 3  Pearson's r, Spearman's rho and Kendall Tau B coefficients. P total was excluded from analysis since correlation coefficients were between −0.013 and 0.075. All correlations of analysed components are significant at alpha <0.05 level (most of them <0.001 level). The highest correlation is between COD and BOD-Pearson's r correlation coefficient is 0.98. There is also a very high correlation between N-NO2 and N-NH3 (Pearson's rho 0.89) as well as COD and N-NO2 (0.84). Therefore, Pearson's r correlation between solids and COD as well as N-NH3 and COD is, respectively, 0.51 and 0.54, while Spearman's rho is, in the above cases, 0.82 and 0.89. Generally, Spearman's rho correlation values are bigger than Pearson's r and Kendall's tau-they are between 0.72 and 0.95 while Pearson's r and Kendall's tau values are, 0.45-0.97 and 0.50-0.84, respectively. Almost all of p values of Shapiro-Wilk test for bivariate normality were <0.001, except BOD-N-NO2 (p = 0.002). The linear regression model-which contains seven predictors with cod as the dependent variable-was made. Model coefficients there are in Table 8. The linear regression model-which contains seven predictors with c od as the dependent variable-was made. Model coefficients there are in Table 8. The R 2 factor, 0.95, is high and a little drop of the adjusted R 2 , 0.93, shows robust model (probably not very high overfitting). The explained variance of the model is statistically highly significant (p < 0.001). Therefore, the variance inflation factor is more than 10 for COD (40.0), BOD (37.9), N-NO 2 (16.0) and N-NH 3 (14.6), so the multicollinearity in an ordinary least square is high. Furthermore, a Bayesian Linear Regression was made (Table 9). The prior probability of the respective models was between 0.004 and 0.006. The maximum posterior the model probabilities and the Bayes factor for model were 0.337 and 141.8 respectively-for COD + BOD + N-NH 3 variant. Maximum R 2 coefficient was calculated for the sum of all four factors (R 2 = 0.954) and the sum of them, without N-NO 2 . For the pH of process wastewater, the correlation coefficients were negative. Statistically significant were six out of eleven-the correlation coefficient of pH and other parameters ranged from −0.4 to −0.5.

Air and Technological Wastewater Compounds
In the remaining cases, correlation analyses of technological wastewater compounds, odour and air pollution concentrations, Spearman's rho correlation coefficients ranged For the pH of process wastewater, the correlation coefficients were negative. Statistically significant were six out of eleven-the correlation coefficient of pH and other parameters ranged from −0.4 to −0.5.
In the remaining cases, correlation analyses of technological wastewater compounds, odour and air pollution concentrations, Spearman's rho correlation coefficients ranged from 0.6 (VOCs-P tot. ) to 0.9 (c od -COD). Very high (>0.8) correlation coefficients were also determined for BOD and all the air compounds, as well as COD and almost all air compounds, apart from VOCs, also for pairs: N-NO 2 -NH 3 air, NO 2 -VOCs, N-NH 3 -c od , and NH 3 air.

Conclusions
The carried out research allowed to draw the following conclusions: 1.
None of the air pollution concentration values-ammonia, hydrogen sulphide and methyl mercaptan-meet the permissible reference values for assessing the degree of air pollution-respectively, 0.533 ppm, 0.013 ppm and 0.009 ppm (Regulation of the Minister of the Environment of 26 January 2010 on reference values for some substances in the air) [41].

2.
At AMWTP, where sewage is stored in a tank and only periodically pumped out, much higher values of both sewage and air parameters were observed than in the case of biogas plant A equipped with a sewage system, thanks to which sewage is directed to a sewage treatment plant. 3.
The analysis of the results of air compounds shown a significant positive correlation between the odour concentration and both the main odorogenic and volatile organic compounds. Analysing the individual compounds, a high positive correlation was also found-the strongest between H 2 S, NH 3 and VOCs.

4.
After analysis of the results of sewage compounds, the insignificant correlation between P total and other parameters was found. For the rest of the compounds, the highest positive correlation was found between COD and BOD and N-NO 2 and N-NH 3 as well as COD and N-NO 2 .

5.
According to the results, the impact of physico-chemical parameters of technological sewage on odour emission was significant-the strong correlation was observed between odour concentration and chosen air and wastewater parameters. To make these relationships more accurate, linear regression models were performed, which were characterized by high determination coefficients. 6.
Municipal waste treatment plants, especially those equipped with a biogas installation, are an indispensable element of urban infrastructure as well as an important part of a circular economy. Therefore, it is important to support the technological processes carried out at plants by analysing them in scientific studies. TS from biological waste treatment processes is very persistent, due to its diverse and variable composition, as well as uncontrolled emission of odours from tanks intended for their storage. The presented research results show the essence and complexity of the raised issues. 7.
It seems advisable to extend the research conducted in this study with an analysis related to the biomethane potential of technological wastewater after the fermentation process. Such a study for household food waste was conducted by Lytras et al. [12]. The mentioned researchers analysed the co-digestion of waste activated sludge and condensate, produced through drying and shredding of source-separated collected food waste, which proved to be an effective method for its valorisation.