The effect of brewery spent grain application on biogas yields and kinetics in co-digestion with sewage sludge

Material characteristics

The main substrate (SS) was obtained from the separately thickened sludge from primary and secondary clarifiers at the Puławy WWTP (Poland). The research was conducted on the basis of a contract between the Municipal Water and Sewage Company S.A. in Puławy (WODOCIĄGI PUŁAWSKIE, Skowieszyńska 51, 24-100 Puławy, Poland) and Faculty of Environmental Engineering (Lublin University of Technology). Under laboratory conditions, these were mixed at the recommended volume ratio of 60:40 (primary:wastesludge), homogenized, then screened through a three mm sieve and stored at 4 °C in a laboratory refrigerator for no longer than one week. Before supplying the SS to the reactors, this was kept at ambient temperature indoors until it reached 20 °C (Szaja & Montusiewicz, 2019).

The BSG was used as a co-substrate to SS. It was obtained from a local brewery, Grodzka 15 in Lublin (Poland). In order to ensure a stable substrate composition, the BSG was dried at 60 °C for 2 h in a laboratory dryer. Then, this sample was milled to a particulate size of 2.0 mm, partitioned in accordance with the assumed doses (Fig. 1) and stored in dry closed boxes. In co-digestion runs, the SS and BSG were homogenized using a low-speed mixer (Szaja & Montusiewicz, 2019). The characteristics of the substrates are presented in Table 2.

Experimental set-up—installation and operational set-up

The study was performed in semi-continuous reactors (supplied once a day) operating under mesophilic conditions (35 °C) at different HRTs. The R1 and R3 digesters were supplied with SS only (the control runs), while the R2 and R4 reactors were fed using a mixture of SS and BSG (the AcoD runs). In the AcoD runs, the BSG mass to the feed volume ratio was constant and was maintained at 1:10, but the HRT differed from 20 to 18 d. The detailed operational set-up is presented in Fig. 1 and Table 3.

The laboratory installation is shown in Fig. 2. Each reactor had an active volume of 40 L, while the volume of head space was 20 L. To maintain a stable temperature each digester was equipped with a heating jacket. The gas installation consisted of a digital mass flow meter, gaseous pipes, a gas sampler, pressure equalization tank and valves. The feedstock was provided to the reactor using a peristaltic pump. Moreover, each reactor had storage vessels both for the feedstock and digestate. Mixing was carried out using a mechanical stirrer with a rotational speed of 50 1/min (Szaja & Montusiewicz, 2019).

An inoculum for all reactors was collected from the mesophilic anaerobic digester with a volume of 2500 m³, operating at HRT of 25 d at the Puławy WWTP. The equalized sample of such a digestate was immediately transported to the laboratory, then divided into four parts supplying, each reactor with a volume of 40 L. Throughout the adaptation phase that lasted 30 d, all reactors were operated without feeding to ensure inoculum post-digestion, indicated by slight daily biogas production (0.01NL/d). As a result, the following inoculum characteristic was achieved: TS of 19.7 ± 0.26 g/kg, VS of 12.2 ± 0.44 g/kg and pH of 8.37 ± 0.01. After adaptation, the experiments of semi-continuous AD/Aco-D started. Every day the digesters were supplied by the feedstock according to the adopted schedule (Fig. 1), with an analogous volume of digestate discharged from them. The semi-continuous experiment lasted 90 for days, including 30 days for microorganism acclimatization to the specific feedstock composition and 60 days for the measurements.

Analytical methods

The analytical methods used in the present study were as previously described in Szaja & Montusiewicz (2019). The SS composition was controlled once a week, while the BSG characteristic was determined once for the whole experiment. The analyses were carried out immediately after the substrate delivery. The following parameters were monitored for both the SS and BSG: total chemical oxygen demand (COD), total solids (TS), volatile solids (VS), total nitrogen (TN), total phosphorus (TP), VFA, alkalinity, pH level, ammonia nitrogen (NH₄⁺ −N) and orthophosphate phosphorus (PO₄³⁻ −P). These were performed with Hach Lange UV–VIS DR 5000 (Hach, Loveland, CO, USA) according to Hach analytical methods (hach.com). Additionally, soluble chemical oxygen demand (SCOD) was determined for the SS using the aforementioned method. The pH values were controlled by the HQ 40D Hach-Lange multimeter (Hach, Loveland, CO, USA). Total and volatile solids were performed in accordance with the Standard Methods for the Examination of Water and Wastewater (APHA, 2005).

The feedstock composition was controlled once a week, whereas the digestate was analyzed twice a week. For both, the analogous parameters were determined: COD, TS, VS, VFA, alkalinity, and pH.

Biogas production was estimated every day using an Aalborg (Orangeburg, NY, USA) digital mass flow meter. Its composition (CH₄, CO₂, N₂ and H₂S) was measured using a ThermoTrace GC-Ultra (Thermo Fisher Scientific, Milan, Italy) gas chromatograph coupled with a conductivity detector fitted with divinylbenzene (DVB) packed columns (RTQ-Bond). The assay procedure was consistent with themanual of this device (assets.thermofisher.com)

The parameters applied in the analysis were 50 °C for the injector and 100 °C for the detector. The carrier gas was helium with a flux rate of 1.5 cm³/min(restek.com). The peak areas were determined by means of the computer integration program (CHROM-CARD).

The kinetics of biogas production were evaluated by determining the constant of the biogas production rateand the untapped biogas potential. The latter parameter represents the difference between the maximum biogas production that can theoretically be obtained from a portion of feedstock introduced to the digester every day(V_max) and the related experimental value achieved from the system for continuous data acquisition (V_e). The biogas production curves were constructed on the basis of the averaged experimental data acquired from an XFM Control Terminal. The biogas production was described using a first-order kinetic equation (Gavala, Angelidaki & Ahring, 2003): (1)${\displaystyle V_f=V_{max}\left[1-exp\left(-k\cdot t\right)\right]}$

where V_f is the biogas volume in time (L), k is a constant of the biogas production rate (1/h) and t is the operational time (h). This method is typically applied for kinetics evaluation in batch systems. However, it was also successfully adopted for semi-continuous reactors (Szaja & Montusiewicz, 2019). In the present study, high values of the determination coefficients (R²) were achieved confirming the accuracy of such an approach.

Statistical analysis

The biogas production curves required for evaluating kinetics were prepared based on the averaged experimental data downloaded from an XFM Control Terminal. The kinetic parameters, such as the constant of the biogas production rate and the maximum biogas production were calculated involving a nonlinear regression method. The strength of the relationships between the results achieved experimentally and those obtained using the equation of the first-order reaction, were established using Pearson’s correlation coefficient (R) and determination coefficient (R²). The statistical analysis was conducted by ANOVA (Shapiro–Wilk’s, Levene’s and Tukey’s tests were included) with StatsoftStatistica software (v 13). The differences were assumed to be statistically significant at p < 0.05.

Removal efficiency of organic compounds

The application of BSG improved the feedstock composition as compared to SS (Fig. 3) and the differences were of statistical significance for VS, TS and COD. In comparison to the control runs, the VS content was enhanced by 29.6 and 21.1% in R2 and R4, respectively, with the related average values of 34.8 and 35.9 g/kg. For SS this was only 26.9 and 29.7 g/kg (in R1 and R3, respectively).

A similar tendency was observed for TS concentration. In this case, the TS improvements were 22.4 and 15.4%, while the TS concentration reached 43 and 46.9 g/kg for R2 and R4, respectively. In the control runs the TS content was much lower and constituted 35.1 and 29.7 g/kg for R1 and R3, respectively.

Considering the COD, the average values in co-digestion runs were 50.4 (R2) and 55.7 (R4) g/L. The SS was characterized by a significantly lower concentration of 39.2 (R1) and 48.8 (R3) g/L. In relation to the control runs, the increases of 28.6 and 14.2% were found in R2 and R4, respectively. Analogously, in the BSG presence the soluble fraction of chemical oxygen demand (SCOD) in the feedstock was enhanced (Fig. 3C), however, the observed differences were of no statistical significance. In this case, the SCOD reached 2,880 and 3,842 mg/L in R2 and R4, respectively, while for the SS the average values were 2147 and 3,800 mg/L (in R1 and R3, respectively).

Moreover, the application of BSG also significantly increased the OLR, and the major difference was observed at shortened HRT of 18 d. Therein, the enhancement reached 33% as compared to the control, while at HRT of 20 d the related improvement was 28% (Table 3). The observed improvements in the feedstock characteristic were due to the BSG composition (Table 2). As compared to SS, such a co-substrate had a significantly higher content of organic matter. This fact should contribute to enhancing process efficiency in the co-digestion systems.

Despite the improved feedstock composition in the presence of BSG, VS removal decreased to 44.1 and 36.1% at HRT of 20 and 18 d, respectively, whereas in the controls the average values were 46.3 (R1) and 36.6% (R3) (Fig. 3A). Interestingly, a major decline was noted when HRT were shortened to 18 d. The observed tendency might indicate the process inhibition caused by significant VFA concentration in the BSG (Table 2). Its high concentration in AD may lead to a decrease in pH value resulting in acidification and the creation of conditions which are especially toxic for methanogens (Murto, Bjórnsson & Mattiasson, 2004), thus contributing to a decrease in methane production and finally to the process failure (Angelidaki, Ellegaard & Ahring, 1999; Chen, Cheng & Creamer, 2008; Franke-Whittle et al., 2014). Moreover, the possible occurrence of phenolic compounds, formed through BSG drying and milling may affect the process performance (Sežun et al., 2011; Retfalvi, Tukacs-Hajos & Szabo, 2013; Sawatdeenarunat et al., 2015). It is known that phenolic compounds may damage microbial cells by affecting the membrane permeability, resulting in leakage of intracellular components and the deactivation of the enzymatic systems (Monlau et al., 2014; Milledge, Nielsen & Harvey, 2019).

Regarding the TS removal, the slight increase from 37.1 (R1) to 38.2% (R2) was found at HRT of 20 d. Conversely, shortening HRT to 18 d led to a minor decline from 29.9 (R3) to 28.4% (R4). Though, it should be pointed out that the differences noted were of no statistical significance. In co-digestion run at HRT of 20 d, the improvement in both SCOD and COD removal occurred as compared to SS mono-digestion (Figs. 3C, 3D). At HRT of 18 d a diminishing tendency was found, and a statistically significant decrease was observed for SCOD, which dropped from 28.8 (R3) to 2.7% (R4), respectively. The minor decline from 40.6 (R3) to 37.4% (R4) was found with the COD removal.

Considering the removal efficiencies, shortening the HRT from 20 to 18 d cannot be recommended for the co-digestion of SS with BSG. Generally, the lignocellulosic biomass requires prolonged retention times in comparison to other substrates (Yadvika et al., 2004). This observation might be attributed to the presence of highly resistant and recalcitrant compounds, mainly lignin (Montusiewicz et al., 2017). The BSG complex structure indicated that such a substrate is not easily accessible to AD microbes which are especially associated with hydrolytic bacteria (Khanal, 2008; Sawatdeenarunat et al., 2015). An analogous trait was typical for other lignocellulosic wastes, such as wheat straw (Shi et al., 2017) and maize (Banks, 2004).

Process stability

The process stability was evaluated by estimating the pH value, alkalinity, VFA concentration and the VFA/alkalinity ratio (Table 4). Considering the feedstock composition, the major differences in the BSG presence were noted for alkalinity and VFA concentration. As compared to the control run (R3), a statistically significant decrease in alkalinity of almost 8% was found only in R4. Conversely, the VFA content increased in the presence of the BSG. The enhancements of 42 and 4.5%occurred in R2 and R4, respectively (Table 4). This tendency was attributed to the implementation of BSG characterized by a large VFA content (Table 2), but the observed differences were of no statistical significance. After anaerobic digestion, a growth in digestate pH was observed. For all runs, the average values were at the levels favorable for methanogens (Table 4). Additionally, the alkalinity increased more than four-fold and the digestate revealed a relatively low VFA concentration, which indicated a stable process performance. However, at shortened HRT of 18 d, a statistically significant increase of digestate VFA content was found, as compared to the control run. This effect might result from a possible digester overload (Chen, Cheng & Creamer, 2008). The co-digestion stability was confirmed by the values of VFA/alkalinity ratio which increased slightly in the presence of the co-substrate. For both controls, the related average values were 0.12, while in co-digestion runs these reached 0.13 and 0.16 in R2 and R4, respectively. The results might suggest a minor inhibitory effect of BSG accompanying the HRT shortage (R4). It should be mentioned that for the lignocellulosic biomass, the shortened HRT might also contribute to the instability of the process (Shi et al., 2017). Importantly, the VFA/alkalinity ratio still remained lower than 0.3, confirming stable process conditions (Bernard et al., 2001).

Biogas production and its kinetics

The supplementation of SS with BSG did not influence the biogas production (Table 5). Due to the BSG characteristics, including a significant amount of carbohydrates, a decreased methane content was observed and the major divergence was found at HRT of 18 d. Importantly, despite the reduction observed, biogas with such a characteristic may still be efficiently used at WWTPs in combined heat and power units (CHP). Consequently, in this case a diminished methane yield was observed and the average values were 0.29 and 0.27 m³CH₄/kgVS_added in control and co-digestion runs, respectively, but the difference was of no statistical significance. At HRT of 20 d, the methane yield was 0.21 m³/kgVS_added for both reactors. These results exceeded the values reported in different studies. Zou et al. (2018) investigated the anaerobic co-digestion of residual sludge and various lignocellulosic wastes in batch mode; therein, the specific methane yields varied between 0.13–0.16 m³CH₄/kg VS_added. However, these yields were significantly enhanced as compared to the sewage sludge mono-digestion. Comparing other wastes co-digested with SS, different values of biogas/methane production were found. In the co-digestion of SS and slaughterhouse waste, the highest biomethane production reached 0.55m³CH₄/kgVS_added (Salehiyoun et al., 2020). By using SS and an organic fraction of municipal solid wastes, biogas production varied between 0.4–0.6 m³/kgVS_added (Sosnowski, Wieczorek & Ledakowicz, 2003).

Considering kinetics, in the present study, the semi-continuous system was applied. The reactor was supplied once a day with the portion of substrate or substrates and simultaneously the same volume of digested medium was removed from it. Accordingly, the biogas production between each feeding related to a temporal interval 0–24 h (Fig. 4) (Szaja & Montusiewicz, 2019).

The supplementation of the feedstock with BSG led to a decrease in the constant of biogas production rates and a simultaneous increase of untapped biogas potential (Table 6). In comparison to SS mono-digestion, the k values dropped by 21 and 35% at HRT of 20 and 18 d, respectively. Simultaneously, the related untapped biogas potential was 2.5 and 3.5 times greater compared to the control, which indicated that using lignocellulosic matter as a co-substrate needed ensuring the prolonged, rather than shortened HRT. Li et al. (2013) noted that feedstocks with significant lignin content (more than 15% on TS basis) were characterized by low first-order rate constant as compared to other lignocellulosic and manure wastes. This tendency might be attributed to the presence of recalcitrant compounds and the inhibitory effects of phenolic compounds potentially appearing during the BSG pretreatment (milling and drying) (Retfalvi, Tukacs-Hajos & Szabo, 2013; Panjičko et al., 2017). It should also be noticed that the hydrolysis of lignocellulose constitutes the rate-limiting step through the conventional AD process (Khanal, 2008).

Energy balance

The energy balance was estimated on the basis of experimental data for a digester operating at the WWTP in Puławy (Poland). A detailed procedure of its calculation was adopted from the authors’ previous study (Szaja & Montusiewicz, 2019). Interestingly, in the presence of the BSG , significantly enhanced energy profits were found (Table 7). This trend most likely came from the improvement of the feedstock composition through the application of BSG, rich in organic compounds.

However, more beneficial results were obtained using longer HRT of 20 d. As compared to the SS mono-digestion, the thermal energy profit was enhanced by approx. 38 and 33% at HRT of 20 and 18 d, respectively, whereas the profit of theoretical thermal and electric power productions was improved by approx. 30 and 27.5% at HRT of 20 and 18 d, respectively. To sum up, the energy generated in the co-digestion system could completely cover the WWTP energy demand, therefore its surplus may be sold to other recipients, increasing company profits in this way.