Comprehensive experiences on the operation of a full-scale continuous dry anaerobic digestion plant treating mechanically sorted OFMSW Open Access

Over the past two decades, municipal solid waste (MSW) management has undergone significant transformations. In the late 1980s, landfilling and mass burn incineration predominated as the primary disposal methods for MSW, with composting constituting a minor fraction due to quality challenges arising from contaminants like heavy metals and inert materials (De Baere & Mattheeuws 2013). Recycling efforts were limited mainly to paper, glass, and materials that were relatively easier to recover. However, since then, notable advancements have occurred across various waste management fronts. One standout development during this period has been the widespread adoption of anaerobic digestion (AD) technology for the treatment of organic fractions derived from MSW.

AD of organic fraction of municipal solid waste (OFMSW) minimizes landfill use and reduces greenhouse gas emissions significantly. It is regarded as a highly established technology for managing and stabilizing food waste (Voelklein et al. 2016) under controlled conditions in contrast to landfilling, thus minimizing fugitive methane emissions from organic waste and maximizing renewable energy output.

Co-digestion of the OFMSW with other substrates is an attractive alternative for sustainable management of various waste streams produced in large amounts (Dereli et al. 2010; Ersahin et al. 2011). Most wastes with a high solid content, e.g., municipal sludge, animal manure, food waste, and agricultural residues contain significant amounts of biodegradable organic carbon. These organic-rich wastes and residues are ideal feedstocks for renewable energy generation through AD (Dereli et al. 2012).

Dry AD, also referred to as solid-state AD or high-solid AD, is recognized as a process specifically tailored for biogas production, where total solids (TS) content in bioreactors exceeds 20% (Rocamora et al. 2020). Dry AD systems emerged as more efficient alternatives to slurry digesters (wet AD systems), requiring less volume, higher organic loading rate (OLR), reduced water addition, lower energy consumption, and less sophisticated pre-treatment methods (De Baere & Mattheeuws 2013; Rocamora et al. 2020). However, these systems also come with certain drawbacks including prolonged degradation periods and the risk of inhibitory compounds building up, such as volatile fatty acids (VFAs), ammonia, and heavy metals, primarily due to the high total solid (TS) content (Ajay et al. 2011).

Continuous dry AD systems are preferred due to lower investment costs and less sophisticated pre-treatment systems compared to wet AD systems and due to higher biogas yields compared to batch dry AD systems (Bolzonella et al. 2006; Fernandez-Rodríguez et al. 2013). Although there are many laboratory (Patinvoh et al. 2017; Veluchamy et al. 2019; Chen et al. 2020) and pilot-scale (Gong et al. 2020) studies on continuous dry AD of various organic wastes in the literature, studies on OFMSW are limited for laboratory-, pilot- and full-scale plants. Basinas et al. (2021) and Rossi et al. (2022) researched continuous dry AD system treating OFMSW in laboratory scale, whereas Zou et al. (2022) and Seruga et al. (2020) have studied full-scale systems. Further attention on the relation between feedstock composition, OLR and mixing regimes is required for optimization of continuous dry AD systems (Rocamora et al. 2020). It is difficult to simulate the physical and chemical environments of the full-scale digesters in laboratory- and pilot-scale systems. The high TS and large sized impurity content of mechanically separated (MS)-OFMSW is mostly not possible to handle in small size feeding equipment and digesters. To overcome this problem, many studies dilute the substrate or reduce its particle size through grinding; however, this approach fails to accurately reflect the substrate's true rheological properties, settlement risk, or the biochemical conditions that influence metabolic reactions. The limited number of studies on full-scale similar plants in the past was primarily due to the low number of such installations. The number of full-scale plants has increased in recent decades, driven by advancements in technology and accumulated operational experience. While technological improvements have focused on feeding, mixing, and discharging systems, practical know-how has been developed in inoculum preparation and improved process control during both start-up and routine operation. Mixing technologies designed to withstand high viscosity from solids and impurities – particularly those using transversally arranged paddle agitators that prevent sinking or floating – played a key role in these improvements due to their enhanced mixing performance.

In Türkiye, there were two main drivers of the increasing capacity of AD plants. According to Turkish Landfill Regulation, MSW is not accepted to landfills without pre-treatment and the amount of MSW to be sent to landfill is limited to 40% of its weight by 2035 (MoEUCC 2010). AD is used for pre-treatment of MSW before landfills and to reach the landfill diversion targets. Turkish Law on the Use of Renewable Energy Resources to Generate Electrical Energy provides incentives such as feed-in tariff mechanisms for various renewable energy systems, including wind, hydroelectric, geothermal, biomass, and solar energy. Accordingly, AD plants had a 10-year purchase guarantee at a rate of 13.3 cents per kWh for electricity from biogas which provided sustainable financing for AD plants (MoENR 2005). However, in 2021, the incentive rate for new plants was reduced to an interval of 8.0–9.0 $cents per kWh and set in Turkish Lira (the local currency), exact buying price to be announced by government each year.

As a result of these drives, 24 AD facilities treating OFMSW were installed between 2011 and 2023 in Türkiye, where 21 of these plants were dry AD plants with 19 being continuous processes. One of the two main reasons for this popularity increase is due to the robustness of continuous dry AD systems to cope with high impurities in mechanically treated MSW. This advantage minimizes the pre-treatment investment and operational costs considerably for investing institutions and companies. The second reason is the flexibility of the technology to handle high OLRs resulting in high biogas productivity in smaller reactor sizes, which also optimizes the investment costs for these systems compared to wet AD or batch dry AD systems. Considering the waste management strategy of Türkiye in near future, a new incentive mechanism would be necessary for further development of the continuous dry AD system, which could be in the form of subsidized electricity buying price or establishment of gate-fee for treatment.

As highlighted previously, there are not many studies in literature that document long-term performance and efficiency of continuous dry AD plants treating MS-OFMSW and complementary co-substrates. This study demonstrates the performance of a full-scale continuous dry AD facility treating MS-OFMSW operated for approximately four years, exploring mono and co-substrates with varying organic loads in triplicate digesters. It provides valuable insights about the systems' response to challenges caused by the inert materials, high solid content and high ammonia concentrations during long-term operation (including the start-up period). The most valuable contribution of this study is to demonstrate long-term experiences gained through the operation of a full-scale continuous dry AD plant including (a) achievable OLRs and biogas productivity, (b) limits for adding nitrogen-rich co-substrates to maximize biogas production, (c) importance of physical conditions (e.g., TS content, viscosity) in high solids digesters. These aspects may not be fully replicable using bench- or pilot-scale test setups. Therefore, the results reported here provide valuable insights for scaling up dry AD systems more effectively.

The studied full-scale AD plant was part of a mechanical-biological treatment (MBT) facility located in the northwest of Türkiye. In the facility, 1,000 tons/day mixed MSW were separated mechanically using manual and automatic separation for metal, plastic, paper, etc. After a bag opener and manual pre-sorting steps, waste was screened by a trommel screen. Around 0–70 mm fraction from the trommel screen and 0–40 mm fine of organic fraction from the ballistic separator were collected in feeding bunkers and fed to the digesters via belt conveyors. The equipment used in the mechanical separation and feeding process was manufactured by Disan Ltd. (Türkiye). The mechanical separation and anaerobic reactors of the MBT plant started to operate in January 2021. On the OFMSW feeding line, a hard particle separator (HPS) (Strabag, Austria) was installed later in September 2021 to reduce impurities before feeding to the AD system.

The quantity of waste fed to the reactors was measured by load cells of ZSFY-A30 T (KELI, China) installed in the automatic feeding bunkers. Biogas production was measured using Prosonic Flow B200 flowmeters (Endress + Hauser, Germany), and methane content was analyzed with an online biogas analyzer SWG 100 (MRU, Germany). All parameters were measured throughout the entire operation except for nitrogen analysis, which commenced after the start-up phase was completed. Biochemical methane potential (BMP) tests for OFMSW and co-substrates were conducted with Bioprocess Control AMPTS^® II (Sweden) according to VDI 4630 standard (VDI 4630 2006). In the BMP tests, liquid fraction of the digestate sample taken from the plant's reactors were used as inoculum. For each substrate type, three samples were used, and average values of the result were taken. The BMP test equipment consisted of an incubation unit, a CO₂ absorption unit and a gas measurement unit. The reaction bottles had a volume of 500 mL each. As control sample, starch was used. The tests were conducted at 39 °C for durations between 21 and 29 days depending on the substrate type. The tests continued until the daily biogas production equalled less than 0.5% of the total biogas production on three consecutive days. TS concentrations of the mixture were kept at <10%. The TVS concentration of the substrates and inoculum mixture was kept at 40 g VS/L. The VS-based inoculum to substrate ratio (ISR) was kept between 3 and 4 as per the standard and also recommended by Weinrich et al. (2018). BMP tests results were also used in the calculation of the biochemical biogas potential (BBP) by using the methane content of the biogas for each substrate (Dragoni et al. 2017).

The digesters were commissioned in sequence at the beginning of 2021. For the first commissioned digester (Digester 2) cattle manure was used as inoculum. The substrate to inoculum ratio was adjusted to 1:1. Digesters 1 and 3 were commissioned using the digestate from Digester 2 as inoculum. After commissioning at the beginning of 2021, start-up period took approximately three months (January to March 2021), followed by eight months of mono-digestion of OFMSW (April to November 2021). Then, the co-digestion period commenced in December 2021 to increase biogas production until the end of November 2024, lasting 35 months. Although data collection for this study finished in November 2024, co-digestion is still in progress in the facility.

Daily feed flow rates per digester ranged from 90–175 and 70–200 ton/day during mono- and co-digestion periods, with average values of 141 and 139 tons/day, respectively (Table 2). The corresponding hydraulic retention times (HRTs) were between 17–30 and 15–44 days, respectively, with an average of approximately 21 days for both periods. The average TS of the feed was 37 and 35% during the mono-digestion and the co-digestion, while the average TVS ratios were 52 and 64%, respectively, reflecting a significant increase in the co-digestion phase due to the higher organic content of co-substrates and the integration of HPS equipment.

During the mono-digestion period, due to inefficient pre-treatment of OFMSW, operational failures occurred once in Digester 1 (February 2022) and twice in Digester 2 (March 2021 and September 2021), causing interruption of operation for several months in each failure. During these failure periods, feeding was ceased and then restarted. The failures were mainly due to the blockage of mixers by large inert materials such as stones that could not be removed in the mechanical separation system.

OLR ranged between 3.4 and 13.8 kg TVS/m³.day during the mono-digestion period with an average of 9.1 kg TVS/m³.day. During the co-digestion period, OLR ranged from 4.2 to 14.9 kg TVS/m³.day, with a higher average of 9.9 kg TVS/m³.day. The digester temperature was controlled between 39.0 and 39.5 °C with automation (Table 1).

The physico-chemical characteristics, BMP test results and calculated BBP values for the OFMSW and each co-substrate are presented in Table 2. After the addition of the HPS system in the mechanical separation line, inert materials such as stones and glass were partially removed, resulting in a decrease in TS content and an increase in TVS/TS content.

Table 3 outlines the average values of the main process parameters and yields observed in the reactors during the operation time, which lasted for almost four years. The results are grouped separately for mono-digestion and co-digestion periods.

During the first year of co-digestion (2022), OLR was maintained more stably and initially ranged between 10 and 13 kg TVS/m³.day, later reaching 14 kg TVS/m³.day. In 2023, OLR was first reduced to 10–12 kg TVS/m³.day, then further dropped to 6 kg TVS/m³.day (due to increased free ammonia nitrogen concentration from chicken manure overfeeding) followed by a rapid increase to 12 kg TVS/m³.day and a subsequent decrease to 7 kg TVS/m³.day levels.

By the end of 2023 and the beginning of 2024, OLR reached 14 kg TVS/m³.day by increasing the co-substrate dosing. At this high level of OLR, process instability was observed when the co-substrate type was changed from WWTP sludge to chicken manure in February and March 2024. As a precaution, chicken manure addition and OLR were decreased until May 2024. From this period onward, OLR was again increased by adding only WWTP sludge as co-substrate and was maintained between 9.5 and 13.0 TVS/m³.day until Fall 2024 (Figure 3).

When OLR was sustained between 10.5 and 12.5 kg TVS/m³.day using OFMSW alone or OFMSW and WWTP sludge mixture, the digester operation was more efficient and stable. In most of the monitored operation period, OLR was kept within this range for all digesters, although there were significant fluctuations over time. In a comparable study, Rocamora et al. (2020) reported OLR values up to 15 kg TVS/m³.day for full-scale continuous dry AD systems treating OFMSW.

Zou et al. (2022), in their study on a full-scale plant with continuous dry AD reactor treating household kitchen waste, observed that increasing OLR until 15 kg TVS/m³.day linearly in 120 days did not cause a serious problem in the process. When OLR reached 15 kg TVS/m³.day, deteriorations in the process were observed such as accumulation of VFA, decrease in the biogas production and sharp drop in the CH₄ content in the biogas.

During the mono-digestion period, methane content of the biogas ranged from 54 to 60%, with an average value of 56%. Meanwhile, during the co-digestion period, the variation was between 50 and 65%, with an average value of 61%. During the period of ammonia inhibition in week 122, lower methane contents was observed. Other than this period, average methane content was higher due to protein rich co-substrates.

The average biogas production during mono-digestion stage was calculated as 114 Nm³/ton OFMSW fed. This value is similar to the findings of Seruga et al. (2020) as 105.3 and 111.1 m³/ton for MS-OFMSW and SS-OFMSW, respectively, in a thermophilic full-scale continuous AD plant.

When adding co-substrates to increase energy production, it is critical to balance elements such as carbon and nitrogen (Bouallagui et al. 2009; Karthikeyan & Visvanathan 2013). Excessive free ammonia nitrogen concentrations can inhibit microbial communities in digesters, leading to process instability and reduced biogas production (Jokela & Rintala 2003) and eventually to process failure (Chen et al. 2008). Angelidaki & Ahring (1994) observed poor process performance above 700 mg/L of free ammonia concentration during the digestion of cattle manure. In this study, during mono-digestion, NH₄-N concentrations were observed between 1,300 and 2,000 mg/L with free ammonia nitrogen below 400 mg/L (Figure 8(c)).

During the first year of the co-digestion period in 2022, NH₄-N concentrations reached 5,500 mg/L without affecting biogas production significantly as free ammonia nitrogen concentration remained below 700 mg/L. However, in 2023, NH₄-N and free ammonia nitrogen concentrations increased to 6,000 and over 700 mg/L, respectively, due to overfeeding of chicken manure, which decreased biogas production. In spring 2023, at week 121, free ammonia nitrogen peaked at 2,000 mg/L, resulting in significant process instability and an increase in VFA/TA ratio, coupled with a drop in biogas production. Starting from 2023 summer until September 2024, free ammonia nitrogen concentrations were kept between 250 and 400 mg/L by controlling the substrate composition. As a result, the process performance was quite stable. In October and November 2024, the levels remained below 250 mg/L, lower than inhibition levels.

The nitrogen analysis of co-substrates showed that the majority of nitrogen at the digester inlet was originating from the broiler chicken manure (Table 2). When the ratio of chicken manure in the feed was evaluated throughout the entire operation period, free ammonia nitrogen concentration increase could be correlated with the increase of the CM ratio in the feed When the CM ratio exceeded 10% (w/w) in week 116, free ammonia nitrogen concentration also increased accompanied with an increased pH. Although the OLR value at this period was comparingly lower (6–8 TVS/m³.day), process instability was signaled by an increase in VFA/TS ratio and a decrease in the SGP. When the CM ratio peaked at 22% (w/w) in week 120, the highest free ammonia nitrogen concentration was observed at 2,027 mg/L with 2 weeks lag time. The VFA/TA average of the three digesters reached to 0.8, whereas in Digester 3 it reached to 1.0 in week 124, having even higher values in daily readings (Figure 6).

The ratio of SGP to BBP of the feed mixture, also defined as process efficiency P_eff (%), fluctuated between 85 and 95% throughout the whole operation period, with influences of changes in OLR, HRT, and moderate free ammonia nitrogen levels (Figure 8). During this period, SGP production fluctuated between 450 and 550 Nm³/ton TVS_fed, where the level of BBP calculated was between 550 and 625 Nm³/ton TVS_fed. However, between weeks 123 and 149, productivity in all digesters were below 80%, calculated by the average SPG of 560 to average BBP of 710. In Digester 3, which had the highest CM ratio of 25%, productivity values fluctuated between 65 and 75%. This corresponded to an interval of 460–530 Nm³/ton TVS_fed for SPG. From week 150 until week 197, CM ratio remained below 10% and process instabilities were not observed. P_eff was over 80% for all digesters in this period. When the free ammonia nitrogen level dropped below 250 mg/L and the OLR was maintained around 12 kg TVS/m³.day in October and November 2024, P_eff was observed close to 90%.

When the main operation parameters such as the OLR and ratio of nitrogen-rich substrates can be controlled in the full-scale plant, TVS removal rates and consequent biogas yields could be optimized and values between 90 and 95% of the BBP test results could be achieved. This result could be useful for designing high solids AD systems treating MS-OFMSW in countries where SS collection is not applied.

Overall, this study provided valuable insights into the operation of continuous dry AD plants for the treatment of MS-OFMSW by analyzing a large dataset spanning approximately four years. An OLR of 10.5–12 kg TVS/m³.day with an HRT of 16–18 days was recommended, which could provide a GPR of 5.0–6.6 m³_biogas/m³_reactor.day. The results of the study highlighted the importance of optimizing substrate composition and OLR to attain a sustainable process. High nitrogen-containing wastes such as broiler chicken manure should be carefully introduced to AD to prevent ammonia inhibition. If readily degradable but high nitrogen-containing co-substrates are needed to increase biogas production due to availability of spare digester capacity or economic reasons, it is recommended to run experimental studies (batch or continuous reactors) to determine the optimal co-substrate mixing recipe. In the investigated full-scale plant, broiler chicken manure was used as co-substrate and the critical ratio at the feeding was observed to be 10% (w/w) to maintain a stable and efficient AD process.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.