Quali-quantitative characterization of biogas with the temporal behavior of organic load on wastewater treatment plant with upflow anaerobic sludge blanket reactors through measurement in full-scale systems

1Universidade Federal do Paraná – Curitiba, (PR), Brazil. 2Companhia de Saneamento do Paraná – Curitiba (PR), Brazil. Correspondence address: Orlando Antonio Duarte Hernandez – Graduate Program in Water Resources and Environmental Engineering, Department of Hydraulics and Sanitation, Universidade Federal do Paraná – PO Box 19011 – Jardim das Américas – CEP: 81531-990 – Curitiba (PR), Brazil. E-mail: oranduher1@gmail.com Conflicts of interest: the authors declare that there are no conflicts of interest. Funding: Brazil–Germany Project on Biogas (PROBIOGÁS), SANEPAR, UFPR, and INCT ETEs Sustentáveis (Brazilian National Science and Technology Institute for Sustainable WWTPs), Organization of American States (OAS), through cooperation with Brazilian Coordination for the Improvement of Higher Education Personnel – CAPES/UFPR. Received on: 02/21/2021. Accepted on: 08/27/2021. https://doi.org/10.5327/Z217694781059 Revista Brasileira de Ciências Ambientais Brazilian Journal of Environmental Sciences Revista Brasileira de Ciências Ambientais Brazilian Journal of Environmental Sciences ISSN 2176-9478 Volume 56, Number 1, March 2021 This is an open access article distributed under the terms of the Creative Commons license. Hernandez, O.A.D. et al.


A B S T R A C T
This study aims to present the time behavior of wastewater flow parameters, organic matter, biogas flow, biogas composition, and its relations, measured through online sensors, in a municipal wastewater treatment plant (WWTP) operating full-scale upflow anaerobic sludge blanket (UASB) reactors, installed in the south of Brazil. WWTP has online measurement devices to evaluate some physicochemical variables of the sewage and the biogas. The COD analyzer (UV-Vis probe), ultrasonic flow meter, biogas flow meter, and biogas composition analyzer were the equipment used. The monitoring occurred for two time periods each of 72 h and one time period for 48 h in the year 2018. Data were checked with descriptive statistics, data independence was checked through the autocorrelation Box-Ljung test, normality behavior was checked with several tests Lilliefors,D'Agostino K 2 ,, and Spearman's correlation coefficient was used to evaluate the correlations among the parameters. The mean sewage flow was 345 ± 120 L.s -1 ; removed organic load was, in average, 48%; biogas quality values were 82.32% ± 3.62% v/v (CH 4 ), 2.66% ± 1.19% v/v (CO 2 ), and 3453 ± 1268 ppm (H 2 S); and the production per capita

Introduction
The treatment of sewage in warm regions, such as South America and Caribe, generally occurs via anaerobic technologies, such as upflow anaerobic sludge blanket (UASB) reactors. Von Oliveira (2009), Noyola et al. (2012), Chernicharo et al. (2015), and Mainardis et al. (2020) recognized the great advantages of UASB, since it allows the reduction of the costs of implementation, operation, and maintenance of wastewater treatment plants (WWTP); besides, it requires a low initial investment.
UASB reactors are well known for their efficiency on removal of organic matter and solids, low energy demand, and without adding chemicals. The structure of these reactors basically consists of a tank with a bottom layer of biological sludge and a settler and gas deflector on the top container. With the proper operation, a tendency of separation of solid, liquid, and gas phases occurs (Lettinga et al., 1983;Chernicharo et al., 1999). For these authors, among the main parameters related to the design of UASB reactors, hydraulic volumetric rate (HVR), hydraulic retention time (HRT), volumetric organic loading rate (Lv), and upflow velocity should be accounted.
Many studies have been conducted expressing or comparing the mean volumetric organic loading to the efficiencies of UASB reactors in the treatment of sewage. In this regard, volumetric organic loading is recommended to be between 2.5 and 3.5 kg COD.m 3 .d -1 (Chernicharo et al., 1999;von Sperling and Chernicharo, 2005;Chernicharo et al., 2015). Previous studies from Lettinga et al. (1983) reported lower loads, similar to Aisse et al. (2002), presenting values of 1.80 kg COD.m -3 .d -1 for the hydraulic retention time of 8 h. Aisse et al. (2002) mentioned the COD of (151 ± 64) mg.L -1 in the effluent of a UASB reactor treating urban wastewater. Considering the influent COD of (453 ± 147) mg.L -1 , the authors obtained the COD efficiency removal of 67%.
The gas phase, inherent to sewage treatment in UASB reactors, represents a great advantage, especially regarding biogas production with elevated methane content. Biogas in UASB reactors, treating municipal and domestic wastewater, presents its composition as follows: methane (70-80%), nitrogen (10-25%), carbon dioxide (5-10%), and H 2 S (1,000-2,000 ppm) (Noyola et al., 2006;Possetti et al., 2019). The proportion among these components depends on the type of biological treatment applied and on the substrate, which could be urban solid residues, domestic and municipal wastewater, sludge from municipal wastewater treatment, animal waste, among others (Venkatesch and Elmi, 2013;Mainardis et al., 2020).
Methane is associated with greenhouse gases, with CH 4 global warming potential (GWP) being 28 times superior to CO 2 ; thus, biogas combustion for energy production could avoid methane emissions and substitute fossil fuels, also reducing the CO 2 eq tons released to the atmosphere (IPCC, 2014). Methane has a lower calorific value of 9.9 kWh.Nm -3 , and its concentration defines the potential of recovering energy from the biogas; electric power production from biogas is an alternative with great expansion potential in Brazil. Biogas production rates, verified by Lobato et al. (2012), from 9.8 to 17.1 NL.hab -1 .d -1 , and Cabral et al. (2017b), from 3 to 138 NL(CH 4 )/kg(COD remov ), have been used by researchers and wastewater treatment plant managers. Possetti et al. (2013), Waiss and Possetti (2015), and Cabral et al. (2017b) observed a direct correlation between the influent sewage flow and rainfall, with the consequent lowering of HRT and the production of biogas. For Possetti et al. (2018), the rainwater results in sewage dilution (increase of flow and lowering of COD concentration), significantly reducing the biogas production. Mota et al. (2019) studied the variations in the concentration of methane (CH 4 ), carbon dioxide (CO 2 ), and oxygen (O 2 ), during 24-h periods, in a sanitary landfill, located in the Northeast Region of Brazil, with a predominantly hot tropical and mild semi-arid climate. The research area showed no significant seasonal variation, only periods with more or less rainfall. There were few changes in the climate of the semi-arid region of Northeastern Brazil during the year. Pagliuso and Regattieri (2008) observed that the increasing municipal demand for electric power requires alternative sources, thus making it necessary a deep knowledge of the time behavior of biogas obtained was 4.51 ± 1.65 NL.hab -1 .d -1 . It was estimated an electric power generation of 3118.6 kWh.d -1 , which is equivalent to an installed power of 130 KW. The behavior of removed organic load and biogas flow (Nm 3 .h -1 ), produced in the treatment plant, showed variable, periodic, and nonstationary time behavior.
New technologies rising in the market, especially those related to online and remote sensing, allow measurements in loco and in real time of biogas production in UASB reactors. Mota et al. (2019) recommended the development of further research, and estimating the potential biogas is particularly important to assess the feasibility of its exploitation for energy purposes.
In this context, this study aims to present the time behavior of wastewater flow parameters, organic matter, biogas flow, and biogas composition, measured with online sensors, in a municipal wastewater treatment plant operating with UASB reactors, in full scale.

Materials and Methods
This study took place in a medium-size WWTP, installed in the south of Brazil, with a design flow of 420 L.s -1 of domestic sewage and serving approximately 180,000 inhabitants. The wastewater pretreatment occurs with two mechanized screens and one grit chamber. The biological treatment is done in six UASB reactors (secondary treatment), and post-treatment of anaerobic effluent occurs in aerated followed by sedimentation ponds. The biogas generated by the UASB reactors at the plant is destroyed in an enclosed flare.
The treatment plant has online measurement devices to evaluate the behavior of some physicochemical variables of the sewage and the biogas (Figure 1). The COD meter (probe) in the sewage, the sewage flowmeter, the biogas flowmeter, and the biogas quality analyzer were the equipment used in this research.

Instrumentation
The COD measurement system is composed of a spectrometer and a control unit; spectrometer probes work according to the principle of UV-Vis spectrometry. The system can determine concentrations between 100 and 3,250 mg(COD).L -1 . A detailed description of the probe can be found, e.g., in Langergraber et al. (2003) and Hernandez et al. (2018). The probe possesses an uncertainty of 1.8%, for a probability coverage of 95.45% (Hernandez, 2019).
The treatment plant possesses an ultrasound flowmeter, with a resolution of ±0.2%, located over a Parshall flume in the inlet of the treatment plant. The equipment has an output with analog standard 4-20 mA, with an uncertainty of ±0.001%, for a probability coverage of 95.45% (Hernandez, 2019).
The biogas flow was measured with a thermal dispersion transmitter, which is basically formed by two temperature probes (insert in the gas flow) and a heater. The energy required to maintain the sensor warm to a constant temperature is directly proportional to the gas velocity. Hence, correlations between energy and velocity are used to calculate the gas production. In this regard, the uncertainty of the equipment is 10.57% for a probability coverage of 95.45% (Hernandez, 2019).
The gas analyzer is a measurement system composed of a static unity and a portable measurement device, which receives biogas samples collected in the burning line. The biogas analyzer uses selective infrared probes to measure CH 4 (0-100%) v/v and CO 2 (0-100%) v/v, and electrochemical probes to measure O 2 (0-25%) v/v and H 2 S (0-5,000) ppm. Regarding the uncertainties, for the infrared probes, it is ± 1.5%, whereas for the electrochemical probes, it was assumed to be ± 0.03% for a probability coverage of 95.45%.

Energy recovery from biogas
The potential of energy generation via the use of the biogas produced in the WWTP was estimated through the following Equation 1 (Cabral et al., 2017a): Where: EP = energy potential (kWh.d -1 ); Q CH4 = methane flow rate (Nm 3 .d -1 ); EC = energetic content of methane (9.9 kWh.Nm -3 ); η electric = electrical efficiency of a combined heat and power engine (36%).
The power of the electric engine is calculated by dividing by 24 h, in case of continuous use.

Statistical evaluation criteria
Temperature and operational data collected in the treatment plant were used, and precipitation data were registered with a pluviometer also located in the plant. In addition, the obtained values were transmitted to a database and subsequently treated in electronic datasheets for the elaboration of the descriptive statistics. The monitoring period occurred hourly for three consecutive days (72 h), in August and in September (samplings 1 and 2); in October, the data were collected for 48 consecutive hours (sampling 3), all in the year 2018. The Spearman's rank correlation coefficient (r s ) was used to evaluate the monotonic correlations among the parameters for the significance level of 0.05.
Rough data were checked with descriptive statistics and analyzed for outliers identification with the interquartile amplitude method. Data independence was checked through the autocorrelation Box-Ljung test (Ljung and Box, 1978), and the normality behavior was verified with the following normality tests: the Shapiro-Wilk test of normality (Shapiro and Wilk, 1965) and the Kolmogorov-Smirnov, Lilliefors, Anderson-Darling, D' Agostino K2, and Chen-Shapiro tests (Adefisoye et al., 2016;Razali and Wah, 2011). If normal distribution and lack of autocorrelation are not to be rejected, for a 0.05 significance level, the p-values of the Shapiro-Wilk and Box-Ljung tests are higher than 0.05.

Results and Discussion
The climate of the South Region in Brazil, which is located below the Tropic of Capricorn in a temperate zone, is influenced by the system of disturbed circulation of the south, which produces the rains, mainly in the summer. In the evaluation period, the wastewater collection system was subjected to atmospheric precipitations of up to 38 mm/day. Regarding temperatures, the winter is cool and the summer is hot. The annual medium temperatures range from 14 to 22°C, and in places with altitudes above 1,100 m, it drops to approximately 10°C. Some parts of the southern region also have an oceanic climate. Table 1 shows the meteorological data obtained at the treatment plant.

Organic load
In Figure 2, it is possible to observe the hourly behavior of the organic load, calculated from the relation of the hourly measurements of the ultrasound meter (flow) and spectrometer probe (COD concentration). The probe was used to measure COD in the influent and effluent of the reactor. The reported mean values for the three evaluated periods [sampling 1 (72 measurements), sampling 2 (72 measurements), and sampling 3 (48 measurements)] were 688 ± 243 mg.L -1 for the influent and 358 ± 116 mg.L -1 for the effluent. The mean sewage flow was 345 ± 120 L.s -1 , inferior to the design flow. Therefore, the organic influent load in the reactors was 19,782 ± 9,949 kg.d -1 and the organic effluent load was 10,133 ± 4,566 kg.d -1 .
The UASB reactors presented the mean COD removal efficiency of (47.25% ± 12.03%), and the mean removed organic matter was 9,989 ± 5,980 kg(COD).d -1 . Thus, the removal efficiencies were below the values reported by Aisse et al. (2002) and Oliveira and Von Sperling (2011). The removed organic matter was similar to the mean obtained by Bilotta and Ross (2016) for an equivalent treatment plant. The applied volumetric organic loading rate (Lv) was 1.70 ± 0.81 kg(COD).m 3 .d -1 , which is in accordance with the values reported by Lettinga et al. (1983) and Aisse et al. (2002).
The obtained HRT value of 9.58 ± 2.29 h is coherent with the values reported by Von Sperling (2011), Chernicharo et al. (2015), and Metcalf and Eddy (2016), between 6 and 10 h, in terms of the mean flow, respecting the recommendations of the Brazilian Regulation NBR 12209 (ABNT, 2011). Figure 3 presents the behavior of removed organic load (kg.d -1 ) and biogas flow (Nm 3 .h -1 ) produced in the treatment plant. The curves present variable, periodic, and nonstationary time behavior, corroborating the biogas production values found by Possetti et al. (2013), Cabral et al. (2017b), and Possetti et al. (2019). Figure 4A presents the behavior of the hourly biogas concentration (quality) and the histograms of these measurements. The collected values were 82.32% ± 3.62% v/v of methane (CH 4 ), 2.66% ± 1.19% v/v of carbon dioxide (CO 2 ), and 3,453 ± 1,268 ppm of hydrogen sulfide (H 2 S). In order to complete the 100% v/v in the biogas composition, the difference was attributed to nitrogen (N 2 ) (~15%) v/v, dissolved in the raw sewage, and removed in the gas phase of the UASB reactor (Noyola et al., 2006).

Characterization and biogas production
The presented results indicate that the control and the monitoring of the generated biogas characteristics should be performed continuously, since variation might occur. These variations could occur due to climate, characteristics of the basin, and population that contributes to the treatment plant or occurrence of disturbances in the process of anaerobic digestion (WEF, 1994(WEF, , 1998Brasil, 2017). Figure 4B shows the histograms of biogas hourly concentration. Regarding H 2 S, it was possible to observe greater clusters between 1,700 and 3,700 ppm, highlighting the bimodal feature of the data. For samplings 1 and 2, values ranged mainly between 3,400 and 5,500 ppm, while for sampling 3, the values were located primarily in the interval between 500 and 3,000 ppm (see Figure 4A). The multimodality generally occurs when the data are collected from more than one process or condition. It is believed that rainfall could be the explanation for such behavior. In the period of sampling 3, the mean rainfall was 19.5 mm.d -1 , in comparison with samplings 1 and 2, with a mean rainfall of 2 mm.d -1 . The gas emission did not show a significant difference between the end of the rainy period and the end of the dry period (Pinheiro et al. 2019).
It is noteworthy that the minimum concentration of H 2 S was 130 ppm, and the maximum was 5,457 ppm ( Figure 4B). The obtained data could be interesting to adequate, for example, the chemical dosage in the systems for controlling odor, in anaerobic treatment reactors, or to increase the dosage in periods where a greater concentration of H 2 S is expected. However, for the use of biogas to generate energy, gas treatment is required. For example, motor-generator groups typically demand concentrations of H 2 S below 130 ppm for proper functioning (Soreanu et al., 2011).
Carbon dioxide presented, as seen by Noyola et al. (2006), an asymmetry of the collected data distribution to the left, with the minimum concentrations of 0.7% and maximum concentrations of 6.2% v/v. The histogram also indicates bimodal behavior.
Methane was within the maximum of 94.5% and the minimum of 76.6%. It could be mentioned that in the greater data series, grouping is in the interval between 75 and 87.5%. Moreover, it is evident that the lowest values occurred during sampling 3, rainy period, which is coherent with meteorological conditions (see Table 1) and the data by Possetti et al. (2013) and Cabral et al. (2017b).
The biogas flow showed a relative symmetric distribution, presenting higher frequency in the measurements when the equipment measured between 25 and 45 Nm 3 .d -1 . Figure 4B shows a normal distribution line for the biogas flow; visual inspection indicates possible normality for this parameter but not for the gas concentrations. Biogas flow and removed organic load were tested with the OriginPro © software-based normality tests: Shapiro-Wilk, Chen-Shapiro, Anderson-Darling, Kolmogorov-Smirnov, Lilliefors, and D' Agostino K 2 (omnibus). Each collected sample and the ensemble of all samples were tested, and the results are shown in Table 2. Razali and Wah (2011) compared the power of the first four tests (Shapiro-Wilk, Anderson-Darling, Lilliefors, and Kolmogorov-Smirnov), verifying that they are in descending order (S-W being the most powerful and K-S the less one). Razali and Wah (2011) also showed that the maximum normality test power occurs for N > 200 for symmetric distributions and N > 50 for asymmetric distributions.    Only biogas flow sample #2 shows non-normal behavior.
Most normality tests show coherent results, the exceptions being Kolmogorov-Smirnov and Lilliefors, which is a modification of the K-S test. Normality test results indicate a non-normality trend for removed organic load and normality trend for biogas flow.
The biogas flow, along with the biogas quality, could be of great help, for example, in the operation of a sludge thermal drying system or the possible implementation of a gasometer, for the storage of biogas generated in the treatment plant.
When comparing the removed organic matter with the flow parameters, CH 4 percentage, CO 2 percentage, and concentration of H 2 S, it was observed that the organic load is positively correlated with the four parameters. The larger correlation coefficients were for removed organic matter versus biogas flow (r s = 0.44) and CH 4 percentage (r s = 0.34), respectively. Additionally, there is no direct influence between the percentage of CH 4 and H 2 S concentration, and a positive correlation of 0.52 was evidenced between biogas flow and H 2 S. The correlation coefficients obtained through the Spearman method are presented in Table 3. The correlation varies from negligible (|r s |~ 0) to moderate (|r s |~ 0.6).

Specific biogas production and potential of energy generation
Currently, the treatment plant attends a population of approximately 180,000 inhabitants. Since the average biogas production, in the evaluation period, was 36.46 ± 12.35 Nm 3 .h -1 , the biogas production rate per capita was calculated as 4.51 ± 1.65 NL.hab -1 .d -1 . The biogas production rate with the removal rate was 80.4 ± 29.68 NL.kg -1 (COD).
The unitary relations obtained in the studied treatment plant were close to the inferior limit reported by Lobato et al. (2012). When sampling 3 is studied separately, its biogas production rate per capita presents a considerable reduction, with the mean of 2.72 ± 1.03 NL. hab -1 .d -1 . The periods of intense rain resulted in the lowering of biogas production.
Power generation potential estimative based on the average biogas flow and methane content values found during the monitoring peri-od of WWTP was 3,118.6 kWh.d -1 , which is equivalent to an installed power of 130 KW. According to Rosenfeldt et al. (2015), Cabral et al. (2017a), Gomes et al. (2017), and Possetti et al. (2019), the decision on the best way to use biogas energy depends on the size and operational conditions of each WWTP and on on-site specific requirements, including social and environmental aspects.

Conclusions
The presented results revealed the behavior of different sewage parameters, such as organic load in the influent/effluent and removed organic matter in a wastewater treatment plant implemented, with UASB reactors operating in full scale, including biogas production, and adopting the time behavior in a full-scale approach. Mean hourly values were reported in the evaluation period for COD in the influent sewage, COD in the effluent sewage, sewage flow, biogas flow, and biogas composition (82.32% of methane).
Visual inspection indicates normality for biogas flow, but not for the gas concentrations. Most of the applied normality tests showed coherent results, the exceptions being Kolmogorov-Smirnov and Lilliefors, which is a modification of the K-S test. Normality test results indicate a non-normality trend for removed organic load and a normality trend for biogas flow.
The organic load [kg(COD).d -1 ] was inferior to design parameters, and the removed organic matter efficiency was, in average, 48%. Both removed organic load and biogas flow (Nm 3 .h -1 ), produced in the treatment plant, showed variable, periodic, and nonstationary time behavior. The hourly removed organic matter has shown a positive moderate Spearman's rank correlation coefficient with biogas flow, CH 4 percentage, CO 2 percentage, and concentration of H 2 S. Also, it was verified that there are no direct correlations between biogas flow and the concentration of H 2 S.
The mean biogas production per capita obtained was 4.51 ± 1.65 NL.hab -1 .d -1 , a value inferior to that reported in the literature. The values of biogas composition (82.32% ± 3.62%) v/v (CH 4 ) were in accordance with the values mentioned by Noyola et al. (2006), with H 2 S resulting in the superior limit reported in the literature (between 1,700 and 3,700 ppm). In the period of sampling 3, the mean rainfall was 19.5 mm.d -1 , resulting in the reduction of organic load and biogas production. It was estimated an electric power generation of 3,118.6 kWh.d -1 , which is equivalent to an installed power of 130 KW.