Effect of substituting organic fraction of municipal solid waste with fruit and vegetable wastes on anaerobic digestion

The components of the OFMSW were prepared by recording the total waste of 5 households with average income in the city of Karaj, Iran with a population of 3 million. Recordings were fulfilled in two different seasons, winter (January and February) and summer (July) in 2017. Collecting points were explained to families and only household waste was delivered. After separating inorganic components such as glass, plastic, cardboard and metals, the organic fraction is divided into five general categories: fruit waste (pulp, shell and corn), vegetables (leaf and non-leaf or summer crop), starches (breads, cereals and pasta, rice, sweets, chips, etc.) protein and fat (dairy, chicken skin, cooked fish, meat, oil, bone, egg shell, etc.), cellulosic wastes, (such as tissue napkins, baby diapers, paper patches, tea waste, sunflower, walnut and pistachio shells that absorb a lot of moisture except cardboard and packs). Then, each part was crushed below 1 cm size by a crusher-blender and maintained at 4 °C until use.

FVW was collected from the central wholesale market of fruits and vegetables in Karaj, Iran, from large waste containers including 3 main categories of fruits, leafy vegetables, and non-leafy vegetables. The total daily FVW on that wholesale market was 40 tons. The total amount of leafy and non-leafy vegetables was determined approximately 70% and fruit dedicated 30% of the waste in the summer by inquiring expert sellers. A mixture of each group was randomly selected from several 1100 L containers, mixed, and then crushed.

The inoculum was selected from the Tehran biogas plant (Abali). The digester was two stages mesophyllic wet system that processed mechanically separated OFMSW. After setting the concentration in the pre-enclosure, it was pumped to the first digester for the hydrolysis and acidification stage and then transmitted to the secondary digester for methanogenesis. Due to the need for a high-concentration inoculation for rising TS content of reactors, three types of primary, secondary, and stored thick digestate were selected and mixed at the ratio of 20, 40 and 40% based on wet weight. The TS of these digestates were 4.68, 4.64 and 12.79%, respectively. The inoculum was incubated for 2 weeks before the experiment.

Experiments were carried out on the batch mode in 1-L glass reactors at 37 ± 1 °C in a water bath. The total loaded feedstock and inoculum was 650 g as a useful volume. The ratio of FVW to OFMSW began at 0% on wet basis and increased by steps of 15%, runs R1 = 0%, R2 = 15%, R3 = 30%, and R4 = 45% (Table 1). Each mixture was tested at 2 reactor concentrations in 8% TS and 15% TS. Although some mixtures tested in 25 TS%, the preliminary results showed that after a few days, production of biogas was stopped due to intensive acidification and pH fell down blow to 5.3. For 15 TS% tests, the concentration of TS in the inoculum and feedstock was increased [15]. The concentration of inoculum was increased by adding the dried inoculum at 70 °C and using the weighted average. The TS% of feedstock was also adjusted to 50% at 70 °C with the continuous weighing of samples in an oven dryer. The ratio of inoculum to feedstock (F/I) was set at 1:1 on VS basis [22, 45]. Deionized water was used for TS correction, if necessary. Before starting, each reactor was flushed with nitrogen gas for 1–2 min. The final VS % (based TS) in each reactor was calculated proportional to VS amount of each component, Eq. (1):

where VS_t is final VS (based on TS) and VS_i stands volatile solid content of each component.

The reactors were shaken manually once a day. Each treatment including blanks was performed in triplicate. Water displacement system with 3 M NaOH solution was used for the determination of the methane content (Fig. 1). Volumetric gas production was expressed in normal conditions (273 K, 1 bar). Given that the biogas composition changes day to day, the weighted average was used to calculate the average methane content [16]. Experiments were stopped when the gas production reached less than 1% increase on cumulative production [17].

The solid content was determined by placing wet specimens in an oven at 105 °C for 24 h [18]. The dried samples were then milled and burned to determine the VS% at 550 °C in a muffle furnace until it reached constant weight [18]. pH of feedstock and inoculum were measured using a portable pH meter (Milwaukee pH55, Australia). Total organic carbon (TOC) was determined in terms of TS% by wet dichromate oxidation method according to 5310 APHA standard [18]. Total Kjeldahl nitrogen (TKN) was analyzed using an auto Kjeldahl apparatus (Kjeltec 2100, Foss, Sweden) and consequently titration with 0.1 normal sulfuric acid according to ISO 1871 standard [19]. Alkalinity ratio was determined by a titration method at pH 4.4 and at 5.0, respectively [20]. A HACH Lange DR 3900 (Germany) spectrophotometer was used to measure total ammonium nitrogen (NH4-N) and total volatile fatty acids (TVFA). Digestate samples were centrifuged at 6000 rpm for 10 min and then filtered through 45 micron paper filter. Total volatile fatty acids equivalent to acetic acid (mg/L) and ammonium (mg/L) were measured by LCK 365 and LCK 303 cuvette number. The amount of VS removal as an effectiveness indicator of anaerobic digestion in converting the material and reducing its contamination was calculated based on total mass balances of VS in each reactor before and after the digestion test with subtracting the average VS content of blanks from that of the each test [16].

Analysis of variance (ANOVA) was carried out using SPSS18 software to find significant difference between the treatments in a completely randomized design (One factor ANOVA). Independent variables were TS% and feedstock mixture ratios (R1, R2, R3, and R4) and dependent variables were biomethane yield, methane content, and VS removal. The significant difference between results was considered when p value was lower 0.05 at a probability level of 5% (α = 0.05).

According to the application of the first-order model (FO), which is repeatedly used by researchers for easy biodigestible materials, the cumulative production of biomethane is exponentially increased to reach the final value according to the following equation [21]:where B is cumulative biomethane production (L/kg VS) at day t (day), B₀ stands biomethane yield (L/kg VS), and k is the first-order model constant (1/day). Also, wastes with an initially low decomposition rate require a lag phase (λ) to trigger the activity of microorganisms that it should be considered in the model. In the modified Gompertz model (MG), biogas production is specified according to the following equation [21, 22]:where λ is lag phase (day), t is the time (days), R_max is the maximum biomethane production rate per day (ml/g VS day), and e is the Euler’s number (2.7183).

Non-linear fitting was carried out by MATLAB software (2013a) (curve fitting toolbox) to fit both of first-order and modified Gompertz models. The model’s suitability was expressed by Root Mean Square Error (RMSE) and determination coefficient (R²).

Table 1 reports the OFMSW composition for winter and summer periods. The summation of fruit and vegetable parts in the two seasons accounts for an average 66% of the total OFMSW. From the perspective, more than half of OFMSW comprises fruit and vegetable waste is consistent with results of other researchers [3, 23, 24]. Boni et al. [25] estimated the total amount of fruit and vegetable waste in dining hall garbage of 81%. The amount of only fruit waste in urban waste composition in China is 28%, which is equal to this research [26]. Campuzano and González-Martínez [27] concluded that the difference in researches of OFMSW depends on country change rather than city-to-city differences. In the winter, the amount of fruit fraction is double than vegetable waste, but this proportion is reversed in the summer due to more cultivation and consequently more consumption, especially in Iran. The total amount of fruit and vegetable in OFMSW in the summer was more than the winter shows noticeable seasonal changes.

The average amount of protein and fat fraction share in OFMSW was 6.2%, which is lower compared with other studies conducted on anaerobic digestion OFMSW with 7% [3], and 11% [28].

The characteristics of the synthetic OFMSW and three treatments with FVW replacement are presented in Table 2. The TS% of MSW in Tehran, Iran resulted higher, up to more than 27% [4]. This amount of solid content was close to synthetic source sorted OFMSW waste [15, 29]; However, it was lower than the average value of different country with 27.2% [27]. In general, the OFMSW sampled from a waste treatment site or composting plant has lower moisture content than the synthetic feedstock. The cellulosic waste part moisture content in this study was measured at more than 50%. The amount of VS% of OFMSW was 91.34%, which is close to the same research that composition of organic fraction of MSW was determined [3, 13, 27]. Increasing addition of FVW caused a decline in VS% that is attributed to attached soil to the stem and root of the vegetables.

The C/N ratio of OFMSW was found to be 22.10. The average carbon and nitrogen content, based on the percentage of TS for OFMSW in literatures are 46% and 2.9% respectively (C/N of 16) [27]. Kjeldahl Nitrogen has more variance in the articles, which is related to the difference in composition of OFMSW in a country or how it is synthesized. The amount of nitrogen in the source separated organic waste is higher than mechanically sorted one. Nitrogen content (based on TS%) for source separated and mechanically sorted MSW was determined at 6.4 and 3.75, respectively [30], and for water sorted organic fraction of municipal solid waste was measured at 2.3 [31]. For all treatments of R1, R2, R3, and R4 C/N ratios were within proper range of 20–30 for anaerobic digestion [12, 32].

Experimental runs indicated that at high TS concentration of the reactor, BMP decreased significantly (Table 3). The cumulative biomethane yield (L_N/kgVS.day), as well as the daily methane production have shown for 8 and 15 TS% in Figs. 2, 3, respectively. The results of ANOVA analysis revealed that the interaction effects of TS% of BMP and co-substrate mixer have a significant effect on biomethane yield. However, for TS concentration of 8%, no significant difference was observed between treatments (p > 0.05). Nevertheless, R2 with 15% addition of FVW has a slightly higher yield of methane with a 385.7 L_N/kg VS. The BMP value for R4 in this study (373.2) is close to the results of Pavi et al. (2017) for OFMSW: FVW of 1:1 (50%, w/w) with 350.6 L_N/kgVS. The biomethane yield for OFMSW (R1) with 385.2 L/kgVS was close to the literatures, which has been taken in diverse researches for different countries, which have 415 L/kgVS [27]. In general, if the quota of food waste (especially protein waste) in OFMSW composition be higher than yard trimming and paper waste, more BMP will be obtained [22].

According to research on FVW, whether single-feedstock or codigestion, such wastes are susceptible to acidification of digester because of their easy digestibility [7]. Therefore, a start of decline was observed in more than 30% of FVW addition in BMP. The BMP values of FVW were reported in the range of 351 L/kgVS [33], 335 L/kgVS [34], 276 L/kgVS [14], 403.46 for VW, and 358.27 for fruit waste [35], which is less than the average worldwide value for OFMSW and its co-digestion. In this study, OFMSW was considered as the main feedstock that is more available on mechanical biological treatment plants than FVW. In codigestion of food waste and FVW, it has been proved that mono digestion of both feedstocks has less BMP than 20, 38, and 50% codigestion with food waste. The best results of biomethane yield obtained for about 38% FVW addition with 725 L/kgVS and more than 50% is prone to acidification [6] that is in line this research.

Considering that the interactions of TS% and the amount of FVW supplementation have been significant, these two factors affect simultaneously the deterioration of BMP. Although dry AD has its own advantages such as a smaller digester volume, methane yield and methane production rate (L/kgVS day) decrease with increasing TS% from the threshold defined (15–20) in the dry AD [36]. In fact, lower water content in the reactor results in a rapid accumulation of volatile fatty acids, especially for easily digestible feedstock that hinders the activity of methanogens bacteria [37]. On the other hand, the mobility and mass transfer of microorganisms in dry AD more than 15% TS are lower. For water sorted OFMSW, by increasing the reactor TS content from 11 to 16, the methane yield decreased from 314 to 273 L/kg VS, and VS removal from 41.8 to 26.1% [31], which is inconsistent with present results.

The effects of the reactor TS concentration and FVW: OFMSW substitution were significant on the VS removal of added feedstock, but the interaction effect was not significant (p > 0.05). Fruit and vegetable waste due to high water content and easily biodegradable compounds has a high VS conversion [33]. Therefore, the results were close to VS removal of FVW with 95% [7], 81.3% [33], 82% [38] and 82% for food waste [16]. Up to 30% (w w) addition of melon waste to OFMSW and a carbon to nitrogen (C/N) ratio of 15.9 caused better VS removal up to 57.2% [12]. On the other hand, in 8 TS%, a higher VS removal was observed due to the better mass transfer of microorganisms. It has been proved that the increasing of reactor organic load decreases VS removal due to the production of intermediate products such as VFAs [39].

The methane content of biogas has a significant difference at 8% and 15 TS% (P < 0.05), and in 8%, it was more than the same treatment at 15%. According to Table 3, co-digestion of OFMSW with FVW can improve biogas quality by increasing CH₄% (v/v). Pavi et al. [14] reported an increase in methane content in the codigestion of OFMSW with 50% FVW addition from 76.5 to 80.8%. Also, By increasing the percentage of vegetable waste in food waste from 56.5 to 78.3%, the methane content increased from 52 to 57% [40].This can be interpreted by the C/N ratio [41], and a slight increase in the amount of TKN from treatment R1 to R4. The average methane content for OFMSW in the literature is about 60% [42]. This parameter did not change in co-digestion of corn stover and vegetable waste and was almost 63% [43].

The final values of volatile fatty acids, pH, and ammonium are shown in Table 4. No signs of inhibition were observed based on Table 4. At 8% TS, the amount of TVFA equivalent to acetic acid was between 370 and 520 mg/L, which shows final TVFA has reached stable levels. These values were close to the final VFAs in codigestion of OFMSW and FVW [14] and for water sorted OFMSW with 400 mg/L [31]. However, for fruit and vegetable waste AD obtained was 1150 mg/L [44]. Although, in 15% TS, TVFA increased somewhat, ranging from 925 to 1100 mg/L and the alkalinity ratio raised beyond 0.5 threshold mg/L, but acute inhibition and cessation did not occur due to the appropriate buffering capacity (0.37 with TAC of 10150 mg/L). Mono-digestion of FVW above 8 TS% encountered inhibition due to the accumulation of volatile fatty acids and the reduction of pH [7].

Results of the kinetic study and time of 90% production of biomethane (T90) are shown in Table 5. The effective digestion time or technical digestion time is an important parameter that provides primary information to estimate hydraulic retention time (HRT) for continuous anaerobic digestion [45]. At 8% TS, there is not much difference between mixture treatments, although R4 has lesser T90 (about 10.2 days). While in the 15% TS it is obviously visible in Fig. 3 that increasing of FVW replacement in mixture decreased the T90 from 13.7 to 10.4 days. This is due to the easily biodegradable compounds such as free sugars in FVW and their rapid conversion by microorganisms and consequently the promotion of their reproduction rate, which accelerates digestion of more complicated substances. In similar studies, Wang et al. [6] found that adding 38% FVW to food waste can reduce T80% from 25 to 16 days. El-Mashad and Zhang [16] found that approximately 90% and 95% of the final biogas potential could be obtained after 20 days for codigestion of unscreened cattle manure with 32% and 48% food waste. They proposed an approximately 20 days HRT for continuous anaerobic digestion. Dong et al. [31] proposed lower solid content has a preference to shorten the HRT of the AD and increases the biogas production from water sorted OFMSW. The technical digestion time (80%) of fruit waste and vegetable waste was observed in the range of 11–15 and 10–17 days, respectively with increasing S/I ratio from 0.43 to 2.33 [35]. Regarding Figs. 2 and 3, it is possible to anaerobically digest OFMSW in continuous mode with lesser HRT through addition of FVW.

According to Table 5, the results of curve fitting with MATLAB, modified Gompertz model has privilege than first-order model because of greater R² (more than 0.990 versus maximum of 0.966 R²) and lower RMSE. The comparison of these two models for both 8 and 15 TS% is illustrated in Fig. 4. The highest k value was obtained for R4 in 8% TS with 0.185 as well as 15% TS with 0.159. By increasing k the rate of degradability of the feedstock increases, which is desirable for the AD [21]. Zhao et al. [21] by studying different parts of the waste from East Asian fruits including peels, seeds, shells, obtained the highest amount of k for pitaya peels (0.16) due to the higher carbohydrate content in this part of the fruit and the low lignin content than shell of seeds. Also, the biomethane production kinetics for FVW has revealed that 99% of the biogas is produced in less than 10 days, which is due to the easy digestibility of these wastes [33]. The preference of modified Gompertz model has been reported in other co digestion studies [12, 45].

The lower value of λ in the modified Gompertz model represents faster start-up and activation of microorganisms [21]. In this regard, there was some lag phase in 15 TS % than 8 TS %, which is likely due to lesser mass transfer between microorganisms and slight acidification. In general, it can be concluded that the addition of FVW reduces the required time of AD due to the presence of sugary and digestible compounds, provided that its share does not traverse beyond the optimal C/N ratio of the process.