Anaerobic co-digestion of food waste, bio-flocculated sewage sludge, and cow dung in CSTR using E(C2)Tx synthetic consortia

In the current study, a E(C2)Tx synthetic consortia was tested for anaerobic co-digestion of food waste (FW), bio-flocculated sewage sludge (BFS)/ raw wastewater (RW) and cow dung (CD) at varying proportions in 0.25 L and 6.5 L mesophilic continuously stirred tank reactors. Anaerobic co-digestion of FW with CD and RW at the ratio of 1:1:8 in 0.25 L batch-reactor with E(C2)Tx inoculum resulted in the highest H 2 production with least CO 2 release. The microbial dynamics of FW:CD:RW samples were studied using 16S metagenomic sequencing which indicated a predominance of hydrolysing microbes at the end point of the digestion cycle. Subsequently, the experiments were scaled up in two continuous digesters, namely, R1 (fed with 50% FW and 50% BFS) and R2 (fed with 2% FW and 98% BFS) with 6.5 L working volume at 2.5 g VS L − 1 D − 1 organic loading rate (OLR) for 120 days. The highest VFA production of 19,183 mg L − 1 and 3,265 mg L − 1 with maximum bio-methane yield of 142.21-and 225.03-mL CH 4 g − 1 VS added were recorded in reactors R1 and R2, respectively. In addition, a numerical analysis was conducted to visualize the mixing and temperature distribution within the digesters, and the velocity and temperature profiles were obtained using Ansys Fluent.


Introduction
The increasing human growth rate and lifestyle put the thrust on dwindling natural resources and direct the whole population towards the usage of fossilized fuels for energy consumption.Globally, the dependency of human needs on fossil fuels paves the way for the exploration of alternative energy production pathways.During the past one decade, waste recycling/reuse and minimization strategies have been implemented to mitigate the amount of waste production and the environmental effects of waste management.One of the alternatives for energy recovery is the treatment of domestic food waste (FW) and sewage sludge (SS).The collection of approx.1.3 billion tonnes of FW across the world is seeking attention for its effective treatment towards bioenergy generation (Sinha and Tripathi, 2021).FW means unutilized cooked food and it mainly comes from homes, cafeterias, bars, factory lunchrooms, and restaurants (Agarwal et al., 2021).
In 2019, FW generation was estimated as 931 million tonnes worldwide, in which households contributed 61%, whereas 26% and 13% came from food service and retail (UNEP, 2021).This signifies around 17% wastage of total food production with 11%, 5%, and 2% share from household, food services, and retail (UNEP, 2021), respectively.The UNEP claims in the food waste index that the global consumer food waste can be the double of that of the earlier surveys (UNEP, 2021).In India, the household food wastage is recorded as 50 kg per capita per year which was lesser as compared to Bhutan, Bangladesh and Afghanistan (79, 65 and 82 kg per capita per year, respectively) (Byun et al., 2021).These estimates provide an opportunity to utilize FW as an effective energy source.
The SS or sometimes referred to as bio-solids, are solids which are produced from the domestic and municipal wastewater treatment (Kumar et al., 2016).SS contains carbohydrates (50%), fat (20%), organic matter (30%-40%), phosphorous (1.5%), total nitrogen (3%), and higher concentration of metal ions such as copper and zinc (Hassan et al., 2023;Xu and Brattebø, 2014).The potassium content and C/N ratio in SS remains 0.7 and 10%-20%, respectively, with a pH ranging between 6.5-7.0 (Xu and Brattebø, 2014).The domestic wastewater treatment plants produce sludge in two forms, namely primary and secondary.Primary sludge is produced by removing suspended solids from wastewater and secondary sludge is obtained after the treatment of soluble organic matter and its conversion into bacterial biomass (Samocha and Prangnell, 2019).Consequent to the presence of heavy metal ions, the use of sewage sludge is not prudent due to their harmful impacts on environment and especially on food chain once it gets accumulated (Kumar and Chopra, 2016).In India, only 37% of generated wastewater is connected to the central sewage treatment system, rest is being discharged untreated into rivers, ponds etc. which is a serious threat to environmental health (CPCB, 2021).The raw sewage water contains 1% solids and 99% water.The energy and nutrient enriched characteristics of sewage sludge makes it a useful raw fuel for energy production and in industrial applications (Kumar and Chopra, 2016).
Various existing treatment processes are available for the conversion of carbon enriched fuel into energy via thermochemical and biochemical conversion of waste substrates.Due to the energy intensive characteristics of the thermochemical processes, the biochemical conversion routes are gaining prominence (Khoo et al., 2010).Incineration and landfill strategies are available for treatment of biodegradable waste but the advantages of AD such as less energy requirement, killing of pathogen, providing essential micronutrients, and help in reduction in size of waste, makes it a popular and preferable method (Esposito et al., 2012;Wang et al., 2023).AD is a technical route in which the waste substrates get decomposed and converted into biogas in zero oxygen environment (Vu et al., 2020;Wang et al., 2023).It possesses the capability of reducing 80% of the greenhouse gas emission into the environment (van Lier et al., 2008).
AD has the potential of managing various organic wastes such as energy crops, municipal solid waste, lignocellulosic biomass, and FW etc. (Awasthi et al., 2016).It undergoes complex biochemical reactions and substrate decomposition to produce bio-methane (CH 4 ) and carbon dioxide (CO 2 ) in four step processes i.e., Hydrolysis, acidogenesis, acetogenesis and methanogenesis (Campuzano et al., 2022).The first three steps are not affected by operational factors and environmental conditions whereas methanogenesis is more vulnerable to pH, temperature, organic loading, accumulation of acids, and total ammonia nitrogen (TAN) (Xin et al., 2018).The accumulation of VFAs signifies the overloading of reactor and results in the suppression of methanogens since the former exhibits toxicity at different concentrations.For example, the concentration of acetate up to 3 g L −1 is favourable for the growth of Methanosarcina whereas toxic for Methanosaeta (Karakashev et al., 2005).However, the inhibitory limit for anaerobic co-digestion of FW and SS is not well reported in literature.Protein-rich substrates release ammonium ions on degradation and act as a nutrient source for microbial growth.Ammonium ions' concentration above 1500 mg L −1 has been reported inhibitory for methane production and promotes the presence of a specific type of microbial species if present in higher concentration.Methanosaeta is more sensitive to higher ammonium concentrations whereas Methanosacrina is more resilient (Chen et al., 2008).The C/N ratio is one of the critical operating parameters of anaerobic digestion which indicates the equilibrium between carbon content and microbial population inside the reactor (Koch et al., 2015).The optimum C/N ratio is reported as 25-30:1 for stable anaerobic digestion, whereas a C/N ratio ≥ 30 and ≤6 negatively affects the process (Koch et al., 2015).
SS and FW are the potential substrates for CH 4 generation which has been explored by various researchers.The codigestion of FW and SS is gaining attention worldwide due to high CH 4 yield, sufficient nutrients availability, and balanced C/N ratio than their mono-digestion (Koch et al., 2015).The mono-digestion of food waste acidifies the reactor due to rapidly biodegradable characteristics of food waste substrate and can be tackled using co-digestion (Montecchio et al., 2019).The mixing of the substrate within the digester also affects its performance due to non-uniform heat distribution, if fed with more than 10% total solids (TS) content (Bridgeman, 2012).In previous studies, it is reported that biogas production remains unaffected in the case of less than 5% fed reactors, whereas it produces 30% more biogas as compared to unmixed reactors if fed with more than 10% TS (Karim et al., 2005).The biochemical reactions in AD are mainly carried out by the microbial consortium or inoculum and its concentration and composition determines the start-up time, substrate digestion rate and reactor stability.The inoculum selection is reported as a useful operational parameter by Córdoba et al. (2016) and various studies have been conducted regarding the suitability of different inoculums for different substrates (Wang et al., 2022).Conversely, limited literature studies are available for the identification of suitable inoculum for the co-digestion of FW and BFS.
Hence. the current study was performed to investigate the activity of a newly prepared synthetic microbial consortia (E(C2)Tx) [Rashmi et al. submitted] for the anaerobic co-digestion of FW, CD and/or RW or BFS at different feeding ratios.
The study was conducted at mesophilic temperature range in 0.25 L and 6.5 L continuously stirred tank reactors (CSTR).The uniformity of temperature distribution and mixing inside the reactor was also analysed using Ansys Fluent.The scope of integrating anaerobic reactors in decentralized STPs has also been explored by deciding the feeding ratio based on the availability of FW and BFS from typical domestic households.The physiochemical parameters such as pH, VFA, Temperature, VS reduction , and bio-methane yield were recorded to analyse the performance of the reactors.

Preparation of inoculum
In this study, we have used a novel synthetic microbial consortia (E(C2)Tx) as inoculum which was designed by our group previously [Rashmi et al. submitted].Briefly, this inoculum was composed of waste materials including aged sludge (AS), papermill sludge (PM), electro-flocculated wastewater (EF), and cow dung (CD) in a definite proportion.Firstly, the acclimatisation of heat pre-treated AS and PM at 80 • C for 2 h was performed for 21 days at 37 • C with EF.Thereafter, this acclimatised sludge was mixed in a definite proportion with a freshly prepared heat pre-treated mixture of AS:PM:CD resulting in the final synthetic microbial inoculum, E(C2)Tx.The initial parameters for inoculum, including TS, VS, COD, C/N and pH, were analysed.

Substrate collection
In this study, we collected domestic raw wastewater (RW) from the Sewage Treatment Plant (STP) Unit of IIT Mandi, South Campus (Kamand, Mandi, Himachal Pradesh, India).The collected RW was further processed into Bio-flocculated solids (BFS) using commercially available flocculent Eco-clean-20 for concentrating the carbon content.The fresh cow dung (CD) was collected from the local cowshed (Salgi, near North Campus, IIT-Mandi).The fresh food waste (FW) was collected from the student mess, South Campus (Kamand, IIT-Mandi), and was churned for its effective digestion.

Experimental plan for small lab-scale bottle batch reactors
Multiple co-digestion experiments of FW were performed in different ratios of RW, and/or CD with and without E(C2)Tx inoculum to check the total biogas, CH 4 , H 2 , CO 2 , and other gases production in batch mode with a minimum experimental duration of 14 days.While the final co-digestion experiment of FW with CD and RW in a ratio of 1:1:8 (FW:CD:RW) was performed with E(C2)Tx inoculum for 20 days along with the respective negative control.The mixture was also supplemented with 0.1% glucose and bottles were tightly sealed with a rubber cork connected to a 65 mL syringe and purged with nitrogen to maintain an anaerobic environment.The starting point (SP) and endpoint (EP) samples were collected at the beginning and end of experimental duration, respectively, for 16S metagenomic sequencing and were stored at 4 • C until further processing.The controlling parameters and experimental conditions are mentioned in Table 1.

DNA extraction and 16S amplicon metagenomic sequencing of samples from 1:1:8 FW:CD: RW experiment
DNA extraction was performed using Genomic DNA from soil (NucleoSpin ® 96 Soil) kit as per the manufacturer's protocol (Wang et al., 2023).The quantification and purity of extracted DNA samples was checked by using 0.8% agarose gel electrophoresis and Qubit 3.0 fluorometer (Invitrogen Corporation).Finally, the samples were subjected to the 16S rRNA gene-based metagenomic library preparation using V3-V4 regions universal primers and Nextera XT Index Kit following Illumina protocol.The sequencing of the final pooled library was performed on the MiSeq platform (Illumina, San Diego, CA, United States) by using MiSeq v3 reagents (paired end, 2 × 300 bp) at our in-house NGS facility.

Post-sequencing metagenomic data analysis
The bioinformatics analysis of the generated raw sequences (FASTAQ format) was performed by using QIIME2, version 2021.8.(Quantitative Insights into Microbial Ecology) software suite (Estaki et al., 2020).Denoising was performed by considering the threshold value of Q20 phred score using DADA2 (Division Amplification Denoising Algorithm 2) (Callahan et al., 2016).Open-reference clustering was performed using the VSEARCH plugin at 99% identity to generate operational taxonomic units (OTUs) (Rognes et al., 2016).Feature-classifier plugin was used to assign taxonomy to ASVs/OTUs from the SILVA 138 reference database (Quast et al., 2012) which shared 99% similarity to generate taxonomy tables.2.4.Experimental plan for 6.5 L reactors

Substrate preparation
The fresh CD slurry was prepared with a concentration of 20 g COD/L and used for inoculum stabilization for 90 days as 80% by volume in two reactors, namely, R1 and R2.The initial pH of CD was 7.86 and had the composition as C-42.22%,H-5.66%, N-1.3%, S-0.12%, by mass measured using CHONS elemental analyser (Elementar India Private Limited; UNICUBE).The initial pH of FW was 5.3 which was increased by adding an appropriate amount of calcium hydroxide and the elemental composition was recorded as C-44.27%,H-6.54%, N-1.93%, S-0.28%, and O-1% by mass (Table 1).The composition of BFS was measured as C-36.85%,H-7.014%, N-3.12%, and S-0.92%, by mass.BFS was used in the current investigation for co-digestion with FW for the up-scaled experiment.The initial conditions and parameters of an inoculum and substrate are listed below (Table 2).

Reactors and operation
Two CSTR's (R1 & R2) of 6.5 L volumetric capacity were used to perform the current study in mesophilic temperature conditions of 37 ± 2 • C (Table 3).Initially, both reactors were operated in batch mode for a duration of 90 days and fed with CD substrate and E(C2)Tx as inoculum for stabilization.Thereafter a continuous mode operation was started for co-digestion of BSF and FW by feeding the reactor on a daily basis.The formation of biogas also depends on the hydraulic retention time (HRT) and the organic load (OLR) (Gautam et al., 2022;Liu et al., 2018).So, in this study an HTR of 4 days was considered for feeding the reactor R1 at a ratio of 2:98 for FW:BFS for the entire duration of the experiment.On the other hand, reactor R2 was fed with equal quantities of FW and BFS considering an HRT of 35 days.At the time of feeding, the effluent was collected from the outlet port of the reactor and was tested regularly for its physiochemical characteristics.The mesophilic temperature was maintained using a 250-W electric heater operated by a PID controller.
In order to maintain the homogeneity of temperature and composition, a DC motor driven impeller was employed at 240 rpm.The detailed layout of the reactor showing mountings of different auxiliaries is demonstrated in Fig. 1.The numerical simulations were also performed to determine the mixing and temperature distribution within the reactor.

Chemical analysis
During both the continuous and batch mode operations, the physicochemical parameters such as COD, VFA, TS and VS were monitored and measured using the titrimetric techniques.Alongside, pH and temperature of the reactors were also maintained within desirable limits and measured using pH meter (APERA; accuracy: 2% of FSD) and K-type thermocouple, respectively.The composition of the produced biogas was analysed using the Gas Chromatography system (Agilent; model: 7890B).Nitrogen and helium were used as carrier gases for capturing hydrogen (H 2 ) and CH 4 in thermal conductivity detectors and flame ionization detectors.The water displacement method was used for measuring the quantity of the produced biogas on a daily basis.

Numerical analysis
Computational fluid dynamics (CFD) has effectively demonstrated its use as a tool for analysing mixing and temperature distribution within a vessel successfully (El Ibrahimi et al., 2021).For the CFD study, the ANSYS Fluent 17.2 (parallel processing employing four cores at once) was employed, which uses a finite volume approach for computation.The package comprised geometry creation, meshing, fluent processing, and post-processing.The BFS was considered to be a non-Newtonian, single-phase, constant density substance.Table 4 illustrates the rheological parameters of the sludge for TS concentrations ranging from 2.5 percent to 12.1 percent at 35 • C. For the computations of R1 and R2, the data from TS 2.5% and TS 12.1% were utilized, respectively.

Table 4
The rheological parameters of the sludge for different TS ranging from 2.5% to 12.1% (Wu and Chen, 2008).

Non-Newtonian fluid identities
Sludge from the water treatment plant was classified as a non-Newtonian fluid.Eq. ( 2) describes the relation between the shear stress (τ ) and the shear rate ( γ ).Understanding the dynamics of these fluids could be performed by models like the Bingham model, Herschel-Buckley model, Power-law model, and the Bird-Carreau model.Herschel-Buckley model has also been used in literature for simulating digested sludge but due to scarcity of parameters and for the ease of simulation, the power law model was used (Fernandes del Pozo et al., 2020).Since it is a numerically efficient and easy depiction that specifically connects the apparent viscosity (η) and shear rate, the power-law approach was adopted for simulating the sewage sludge as depicted in the Eq. ( 1) (Bridgeman, 2012;Wu, 2010).
The power-law index and the consistency coefficient are denoted by the letters n and k, respectively.Whenever n would be less than one, shear-thinning happens, however, when n would be greater than one, shear-thickening occurs.The TS is approximately proportionate to the density as demonstrated below in Eq. (3) (Landry et al., 2005).

Equations for flow model
The absolute velocity (⃗ v) along with a rotating reference would be attributed to the relative velocity ( − → v r ) as given below in Eq. ( 4), where ⃗ r is the position vector and angular velocity is ⃗ ω.Because the sewage sludge was speculated to be both the incompressible fluid and the constant density, the continuity equation could be primarily determined by computing the divergence of the velocity.Algebraic form of this statement is depicted in Eq. ( 5) (Meister et al., 2018).
In sewage sludge, surface forces and body forces interfere with the microparticles.The particle's net forces are equivalent to the rate of change of momentum which is illustrated below in Eq. ( 6) (Meister et al., 2018).
The pressure, viscous stress, and gravitational acceleration of the sludge are represented by p, τ , and g.
The kind of flow inside a CSTR can be determined by Reynold's number.For non-Newtonian fluids, it is computed by the following Eq.( 7) (Metzner and Otto, 1957).
k, n, and ρ are the parameters presented in Table 2; N is the impeller's speed in rotations per second, and D m is the impeller's diameter.The laminar flow model is used when Re becomes lower than 10, otherwise, the turbulent flow model is used (Metzner and Otto, 1957).
Numerous models have already been investigated for turbulent flows of non-Newtonian fluids (Wu, 2011).When Reynold's number is significantly high, the standard k-ε model is employed (Mao et al., 2019).Either the standard k-ω model or the realizable k-ε model is suggested for modelling the turbulent flow.This study employed the realizable kε method for the CFD simulation because it had the minimum margin of uncertainty for fluids of TS 12.1 percent (Wu, 2011).

Mixing characterization framework
The dead zone is a frequently adopted metric for measuring the mixing in a vessel.The dead volume is comprised of such volumes which have a velocity under 5 percent of the highest velocity (Vesvikar and Al-Dahhan, 2005).There is not any specific amount of dead volume to determine whether good mixing has been accomplished; nonetheless, a smaller dead zone indicates better mixing.
The velocity gradient is an essential criterion used by the water treatment sector to remark about mixing (Bridgeman, 2012).It is computed using the formula, presented in Eq. ( 8), in a three-dimensional space, where u, v, and w denote the cell's velocity for the x, y, and z axes, respectively (Sindall et al., 2013).

Shape and meshing
In Design Modeler, the digester's shape was designed using just a simple cubical primitive and a pyramidal primitive with that of a square base.The depth of the pyramidal portion was 90 mm, and the edge of the cubical and pyramidal parts was 180 mm.A cylinder and parallelepiped primitives were used to make the shaft and impeller, respectively.The impeller's breadth and diameter were 20 mm and 50 mm, respectively.A cylindrical zone (35 mm height and 80 mm diameter) with a rotational velocity of 240 rpm was generated, which is more than 1.5 times the impeller's dimensions (Wu, 2010).
The grid convergence index (GCI) measures how dependent the outcomes are on a grid in comparison to a benchmark grid.Tetrahedral units were used to form five grids numbered as 1 to 5, where 1 is for the lowest while 5 is for the finest quality grid.Table 3 shows the number of elements, the maximum skewness for determining grid quality along the time needed to perform 1500 iterations.With the use of the formula provided in Eq. ( 10), the root mean square error (e rms ) was determined at 1000 locations.The GCI was determined based on the relationship given in Eq. ( 9) after determining the refinement ratio (r) from Eq. ( 10) (Bridgeman, 2012).
The grid refinement ratio is r, and the number of elements in the coarse and fine grid is h1 and h2, respectively (Table 5).The RMS value of error in velocity, recorded at 1000 identical positions, between the two grids is depicted by e rms .While considering only two grids, the safety factor, represented as F s , is taken as 3, and it is taken as 1.25 if analysing more than two grids (Roache, 1998).In contrast to grids 2 and 3, grid 4 has a much lower GCI.Grid 5 has a slightly better GCI than grid 4, however, the duration of computation becomes very large.Also, when we went from grid 4 to grid 5, the maximum skewness present in the grid dropped from 0.827 to 0.896, which pointed towards a dent in the accuracy of outcome hence grid 4 was selected for further computation purposes in this study.
A two flat-blade impeller running at 240 revolutions per minute mixed the sludge.The multiple reference frame (MRF) approach was used to describe mixing in the Fluent.In this approach, the working space is split into zones with various rotating speeds.According to the discussion in Section 3.1.2,the turbulence framework was chosen, the rheological parameters were added as given in Table 2, and the zone conditions were defined using a pressure-based solver.The no-slip settings with zero thermal conductivity were considered for the walls of the digester, the heat flux and heat production rate at heater exterior wall was configured as per the 250-W heater.The impeller blades' walls were defined as the moving surfaces at zero rpm with reference to its rotating domain.
The computation was done using the SIMPLE technique, with converge criterion of 1e −5 for x, y, z, k, ε and 1e −6 on energy.The simulation was started with the parameters k = 0.001 and ε = 0.01 while the temperature was taken at 295 K.

Gas production during anaerobic co-digestion of food waste, cow dung and/or raw wastewater in 0.25 L reactors
We evaluated the production of CH 4 , H 2 and other gases during the co-digestion of FW with two other wastes namely, CD and RW, as substrates in varying concentrations along with the inhouse developed synthetic microbial consortia (E(C2)Tx) as inoculum under mesophilic conditions (37 • C) using small-scale bottle batch reactors (Table 6).FW is known to have a high potential for CH 4 production which makes it as one of the promising organic substrates for AD processes (Zhang et al., 2014).Clearly the use of FW in large volumes in Expt_1 resulted in the production of a very large volume of total biogas, where both CH 4 (12.59ml) and H 2 (11.23 ml) gases were recovered in low volumes and in almost equal proportions.However, the percentage and volume of CO 2 obtained in this sample is very high.
The digestion of FW alone results in an inhibition of the process in the long-term operation.This inhibition is caused by the presence of higher lipid concentration, pH imbalance, and nutrients imbalance such as insufficient trace elements (Zn, Mo, Fe, etc.) and excessive macronutrients (Na, K, etc.) (Zhang et al., 2014).Besides, during anaerobic digestion of FW, the rate of hydrolysis is prompt which can lead to an instability of the system due to acidification (Ma et al., 2019).To overcome these limitations, co-digestion of FW with several additional organic substrates, such as CD, sewage wastewater, sewage sludge, etc. has been widely employed (Zhang et al., 2014).Co-digestion of organic waste can boost the stability of the AD process, balance nutrient supply, stabilize pH, optimize the C/N ratio, improve buffer capacity, and boost biogas generation as compared to AD on a sole substrate (Ma et al., 2019).Under anaerobic conditions, co-digestion can accelerate the degradation of each substrate.In other words, addition of co-substrate can have synergistic effects, as opposed to additive effects, where an increase in CH 4 production simply results from a greater mass of accessible biodegradable organic matter per unit volume (Xie et al., 2017).Taking into account these advantages, various co-digestion strategies were explored with varying composition of different wastes in our experiments.
Complex lignocellulosic substances are known to be bound on CD (Zhang et al., 2014).The presence of recalcitrant lignocellulosic matter in CD can result in a slower hydrolytic rate upon co-digesting with FW thereby mitigating the effect of hydrolytic acidification due to fast hydrolysis of FW (Ma et al., 2019).Besides, CD is known to have a good buffering capacity and ability to maintain the COD value along with nutrient balance (Zhang et al., 2014).Thus, co-digestion of FW and CD is expected to yield better results by improving the stability of AD systems.Towards this, the co-digestion of FW with CD, by keeping the concentration of the earlier higher (50% (1:1) or 70% (7:3)), was performed (Expt_2, and Expt_3) (Table 6).Although now, a comparatively lower amount of total biogas was recovered as compared to Expt_1, slightly higher volumes of H 2 gas could be recovered.It is notable that negligible changes were observed in the percentage of CO 2 and CH 4 in these experiments as compared to Expt_1.
The composition and characteristics of the co-substrate that contribute to a synergistic impact are linked to various factors, including the C:N ratio, an adequate supply of micro-and macro-nutrients, a higher proportion of easily biodegradable organic components, and a stronger buffering capacity.For instance, to reach an ideal C/N ratio of 20-25 the highly carbon-content rich wastepaper could possibly be co-digested with low C/N sludge (Xie et al., 2017).FW is shown to have a C/N ratio of ∼22 while that of CD is ∼33 (Table 2).The mixing of these two substrates is expected to result in a C/N ratio higher than desired.The C/N ratio of RW can be in the range of 2-3 (Li, 2016) and RW is known to have a

Table 6
Table showing the details of the five experiments performed for anaerobic co-digestion (AD) using varying substrate compositions.The total volume (in ml) of the biogas, carbon dioxide (CO 2 ), methane (CH 4 ), and hydrogen (H 2 ) produced during the AD process along with the cycle length (in days) is provided.The proportion of food waste and other co-digestion substrates used during various experiments is also provided.

Experiment
diluted carbon content and less COD value.Thus, we adopted the strategy of using RW as a co-digestion substrate to dilute FW (Expt_4).Upon mixing FW:RW in 3:7 ratio, we observed a significant reduction in the total biogas volume (Table 6).However, the percentage of CO 2 released is also significantly reduced with four times increase in H 2 gas production as compared to our earlier experiments.Notably, we could not recover any CH 4 gas in this experiment as well.Finally, in Expt_5 we co-digested FW with CD and diluted this mixture with RW in a ratio of 1:1:8 of FW:CD:RW and fermented with our synthetic inoculum.It is interesting to note that the production of total biogas was doubled as compared to Expt_4, although it still remained lower than that in Expt_1, Expt_2, and Expt_3.We did not recover any CH 4 gas in this experiment as well, however, a significantly high percentage of H 2 gas (100 ml, 59.2%) was recovered with significant reduction in CO 2 gas production.The ratio of dosage and varied co-digestion substrates significantly influence the synergistic impact of anaerobic co-digestion (Ma et al., 2019).For example, in this study FW is co-digested with RW and/or CD in different ratios.With the addition of RW and/or CD in FW and treatment with E(C2)Tx synthetic inoculum increased the biogas yield per gVS added as compared to the FW alone and their respective negative controls.Additionally, the added advantage of microbial synergism enhances the digestion process (Anjum et al., 2017).Co-digestion can sometimes have a negative or antagonistic impact as well instead of an overall synergistic effect (Aichinger et al., 2015;Xie et al., 2017).For example, the hydrolysis of complex polymeric compounds, e.g.extracellular polymers in sewage sludge is one of the rate-limiting stages in the co-digestion.Notably, CH 4 was not recovered in the co-digestion experiments.For the production of CH 4 , the presence of methanogens is very crucial.Methanogens are very sensitive to pH change and prefer neutral or slightly alkaline pH for their growth.A slightly acidic pH, which avoids methanogenesis and solventogenesis (5.0 to 6.5), is one of the favouring parameters to support the growth of H 2 -producers and enhances the H 2 -production rate (O-Thong et al., 2018).Whereas an optimum pH for methanogenic bacteria ranges between 6.0-7.5 and anaerobic bacteria work well below pH 6 (Boopathy and Daniels, 1991;Van Ginkel et al., 2005).However, FW is rich in various inhibitory components which can lead to instability of the anaerobic system, including decrease in pH.It has been previously demonstrated that low levels of CH 4 gas could be recovered due to low pH in the reactors (Nzeteu et al., 2021).It is important to note that the pH of the batch reactors was not maintained during the entire course of the experiments.This could be a potential reason for the extremely low yield of CH 4 and higher yield of H 2 in these experiments.
The inoculum (E(C2)Tx) used in this study is composed of different waste components and is prepared by using two methods, namely heat pre-treatment to enrich H 2 producers because of their endospore formation ability and acclimatisation to enrich active methanogenic microbial community (Rashmi et al. under review).The use of this synthetic inoculum demonstrated a substantial increase in the cumulative gas production with respect to the negative control over a period of 20 days (Fig. 2), and also resulted in early gas production.A notable gas production in the control sample started only after the 6th day, while in the case of the sample with inoculum it started within 1 day of setting up the experiment.

Metagenomic profiling of the anaerobic co-digestion of FW:CD:RW
Being a co-digestion substrate, FW addition can provide large amounts of available substrates, which are necessary for microbial metabolism.The differences in FW compositions will affect the anaerobic microbial community structure in the AD systems (de Jonge et al., 2020;Xu et al., 2021).In order to explore the community dynamics during this co-digestion process, we performed the metagenomic sequencing of the SP and EP samples of the FW:CD:RW (1:1:8) substrate treated with the synthetic inoculum and the negative control (Fig. 3).
During the AD of FW phyla Firmicutes, Chloroflexi, Bacteroidetes, and Proteobacteria are observed as the dominant microbial taxa (Cui et al., 2022;Zamanzadeh et al., 2016).Firmicutes play vital roles in the fermentative and acetogenesis stages of the AD process and are also known as syntrophic bacteria which can anaerobically degrade VFAs (Yi et al., 2014).Proteobacteria also play a key role in the breakdown of organic substrates and are significant consumers of VFAs (namely butyrate, propionate, and acetate) (Yi et al., 2014).The members of phylum Bacteroidota are known for their proteolytic ability in addition to being capable of degrading complex and recalcitrant organic material such as cellulose, proteins, and lipids and simpler compounds including, amino acids and sugars by using hydrolytic enzymes (Gannoun et al., 2016).Phylum Chloroflexi is mainly detected in the SP sample of inoculum treated substrate (Expt_5_SP).The members of this phylum are able to consume glucose (Ariesyady et al., 2007) and play roles in hydrolysis and acetogenesis stages of the AD process (Nguyen et al., 2020;Zhang et al., 2020).
Important differences in the microbial community at the genus level are observed between the inoculum treated and control samples and also within the samples based on SP and EP of AD process (Fig. 3(B)).All the four samples showed an over-representation of Pseudomonas genera; however, their abundance is found to be more in the negative control samples.Pseudomonas spp.are of great importance for the robust and dynamic AD food webs by acting as a key candidate for all important methanogenic pathways in the AD process (Buettner et al., 2019).The other taxa that are found to be enriched in the negative control samples are genus Lactobacillus, which is majorly responsible for the production of lactic acid which can cause a drop in the pH of the solution.Lactobacilli are also known to produce some other secretions such as bacteriocins and antimicrobial substrates.The presence of these compounds and a lower pH can reduce the production of CH 4 (Guan et al., 2021;Mustapha et al., 2017).Not surprisingly, the control samples containing a higher abundance of Lactobacillus demonstrated much less gas production as compared to the samples containing the inoculum.
The microbial community in the SP of the inoculum treated samples is also found to be very different from that of EP.For example, the abundance of Prevotella is found to be significantly reduced in the EP samples as compared to SP. Prevotella is known to play a very important role in the digestion of cellulose, starch, hemicellulose and pectin (Dao Trong et al., 2021).The members of this genus are also known to produce propionate, which is one of the main substrates utilized by acetogenic bacteria and transform it into acetate, CO 2 , and H 2 or formate, which are the immediate precursors to CH 4 (Pytlak et al., 2020).In the fermenters, the presence of this mechanism and the extensive influence of Prevotella on the transformation of organic waste into CH 4 , not only by promoting hydrolysis but also by the acetogenesis process, have been confirmed by Pytlak et al. (2020).Interestingly, in the EP samples the abundance of genus Bacteroides is found to be very high as compared to the SP samples.Bacteroides and Prevotella have competitive potential for the degradation of highly biodegradable polysaccharides by using the generated polysaccharides utilization locus (Ausland et al., 2020).Bacteroides cellulosolvens is known to ferment cellobiose and cellulose for the production of acetic acid, CO 2 , H 2 , ethanol, and very little lactic acid while other members are capable of degrading starch (Jiang et al., 2019).This clearly indicates that during the AD process, the abundance of Prevotella gets reduced gradually and the activity of hydrolysis is taken over by Bacteroides.
The SP samples inoculated with E(C2)Tx are found to be enriched in Peptostreptococcaceae, which can play important roles in the fermentative metabolisms (hydrolysis and acidogenesis) during the AD process by converting organic matter into SCOD (soluble chemical oxygen demand) which is further transformed into various VFAs such as acetate and butyrate (Jiang et al., 2019).In addition, the SP sample show a higher abundance of Leptolinea and Longilinea, and other unculturable members of family Anaerolineaceae which are reported as acetate and H 2 producers by fermenting carbohydrates (Yi et al., 2014).On the other hand, the EP samples exhibit a high abundance of the members of Oscillospiraceae and Rikenellaceae.
The members of family Oscillospirales such as Oscillospiraceae-UCG-005 are saccharolytic and carbohydrate fermentative in nature (Wang et al., 2017).While Rikenellaceae_RC9_gut_group are known for fermenting carbohydrates or proteins (Su et al., 2014).While Rikenellaceae_RC9_gut_group are known for fermenting carbohydrates or proteins (Tiong et al., 2023) also reported the presence of Halobacterota and Nanoarchaeota from the archaeal group in the inoculum used in their digesters.The members of Halobacterota include acetoclastic, hydrogenotrophic, and methylotrophic methanogens that can act over a diverse range of substrates (Tiong et al., 2023).The methanogenic community was found to be absent in the EP samples, correlating well with observation that CH 4 gas could not be retrieved at the end of the experiment.A study by Nzeteu et al. (2021) showed that the relative abundance of archaea was inhibited during the AD process due to the initially prevailing low pH in the reactors which led to the low levels of CH 4 production.It is important to note that the pH of the batch bottle reactor was set to 6-6.5 only initially at the start of the experiment and was not maintained further.Due to the process of hydrolytic acidification during digestion of FW, the pH of the digestate is expected to be acidic which is unfavourable for methanogenic microbes.
The small-scale batch bottle reactor experiments have provided important leads towards the upscaling of the anaerobic co-digestion of FW.The regulation of VFA concentration is extremely crucial for the stabilization of pH because of the correlation between pH and the levels of VFAs, CO 2 , and bicarbonates (Liu et al., 2008).Towards this, we have adopted two strategies to maintain the pH optimal for methanogens (6.5-7.2) (i) by allowing the stabilization of our synthetic inoculum with CD for 90 days to increase the buffering capacity and (ii) by adopting alkali treatment of our substrate using bicarbonate.FW can easily undergo acidification, as it has a very low buffering capacity because of its characteristics of high oil, salt and C/N ratio, which lead to the inhibition of microbes and causes less or no CH 4 production (Wang et al., 2022).To address this issue, in the further experiments, the strategy of co-digestion of FW with bio-flocculated wastewater (BF) instead of RW was used to increase the carbon load, and also for the elimination of inhibitory effects and for the regulation of the nutritional structure (Wang et al., 2022).The batch bottle reactor experiments also led to the conclusion of adding lower volumes of FW for further experiments as the results clearly demonstrated that a high volume of FW led to the generation of higher volumes of CO 2 .Finally, during the further scaling-up experiments, changing the batch mode AD process into a continuous mode of digestion with proper pH regulation was decided as the prime strategy, which is well studied for more enhanced biogas production (Zappi et al., 2021).

Reactor start-up performance
The inoculum stabilization was initiated in duplicates with CD which is a commonly available substrate and used in the study in concentration of 20 g COD/L in both of the reactors for 90 days.The Fig. 4(i) shows cumulative biogas production trends which clearly indicates that the biogas production started within 24 h in both the reactors and consistently increased up to 90 days.The cumulative biogas production trends in both reactors were similar except for the small ramps observed in R1 on the 38th and 67th day.The highest biogas production of 470 mL was observed in R1 on the 75th day, whereas in case of R2, it was 360 mL on the 80th day.The bio-methane production trends for 90 days' batch study are illustrated in Fig. 4(ii) which shows that the production of combustible biogas (having 50% methane) in both reactors started from the 15th day and remained consistent throughout the investigation.The highest methane shares of 60.29% and 59.94% were observed on the 25th day in R1 and R2, respectively.
The pH is considered as one of the crucial operational parameters which affect the performance of AD processes.The optimal pH range is considered between (6.8-7.3) for methane production.The pH variation in inoculum stabilization operation varied between the 6.56 and 7.1 as depicted in Fig. 4(iii) which is optimum for the growth of methanogens.The current optimal pH range was achieved mainly due to the high buffering capacity of cow dung and minimal addition of calcium hydroxide as a buffering agent.The pH value between 6.5 and 7.1 results in the production of more than 50% methane throughout the experiment and shows the equilibrium between the organic acid production and their consumptions in both reactors.However, Lindner et al. (2015) observed a slow hydrogenase activity of microbes at pH 5.5 which resulted in low methane yield of 40.9%.

Reactor stability during co-digestion operation
The stability of co-digestion operation was determined by capturing the daily pH variation and VFAs production inside the reactors.The pH variations in R1 and R2 are illustrated in Fig. 5(i).A pH below 6.5 inhibits the methanogenic activity and results in the instability of the process.The more frequent variation of pH in R1 was due to the more amount of food waste percentage than that in R2.The food waste has rapid degradation rate and low buffering capacity, whereas sewage sludge is enriched with several salts and possesses high buffering capacity when mixed with food waste in proper ratio.The fluctuations in pH may also vary with concentration of VFAs accumulation in the reactor as it can cause instability of the process by lowering down the pH from the optimal range.The vulnerable range of VFAs for mono-digestion of food waste are reported up to 3500 mg L −1 (Wang et al., 2018) whereas it is not well reported for different co-digestion ratios of food waste and sewage sludge.The reactor R1 shows increasing trends of VFAs production up to 19 183 mg L −1 on the 53rd day as shown in Fig. 5(ii) which decreases throughout the study.The enormous amount of VFAs generated in R1 drops the pH frequently as discussed earlier and causes an instability of the system.The maximum VFAs accumulation in R1 was 3265 mg L −1 on the 95th day whereas, on the 25th day it was 436 mg L −1 .The deviation in pH from the optimum range was less observed in the case of reactor R2 at the aforementioned VFAs concentration range.Therefore, the current investigation points towards a high VFAs accumulation in the feeding ratio used in the case of R1 than R2 at an OLR of 2.5 g VS L −1 D −1 .
The stability of an anaerobic reactor is also indicated by the C/N ratio inside the reactor.The ratio between 25:1 and 30:1 is considered as optimum for effective anaerobic digestion (Kim et al., 2003).Raw food waste contains a C/N ratio between 11.1-36.4based on the composition whereas sewage sludge has a ratio between 6-9.The effective mixing for co-digestion of food waste and sewage sludge can improve between 6-15 (Kim et al., 2003).The C/N ratio more than the optimum range makes the reactor nutrient deficient and results in a low VS removal rate, whereas if it falls below 6, the inhibition due to ammonia will slow down the methanogenic activities.The C/N ratio of 14.5:1 was observed in R1 and 10.1 in the case of R2 without additional nutrient supply.This was due to the more food waste concentration in the case of R1 than in R2.Kim et al. (2003) also performed the co-digestion of food waste and sewage sludge at a mixing ratio of 20:80 and reported a C/N ratio of 8.20:1.

Impact of co-digestion on biogas production, methane yield & VS reduction
Fig. 6(i & iii) shows the biogas production and corresponding yield in the case of R1 and R2.Initially, similar trends were observed for 20 days, and subsequently, more biogas production was observed in the case of R2 than in R1 throughout the investigation.The highest biogas production was observed as 5.9 L in R1 on the 130th day and 6.7 L in the case of R2 on the 124th day, whereas the corresponding yield was 368 mL g −1 VS added and 407 mL g −1 VS added for R1 and R2, respectively.The results of biogas yield obtained at the current OLR were more than the previous co-digestion study of sewage sludge and municipal solid waste conducted at an OLR of 2.9 g VS L −1 D −1 by (Elango et al., 2007) which advocates the maximum biogas yield of 360 mL g −1 VS added.The bio-methane composition in Fig. 6(ii) indicates that in R2 the combustible biogas production (more than 45% methane) was observed throughout the study.The highest share of 69% was recorded on the 70th day in R2 whereas R1 has not produced more than 45.83% bio-methane in the entire investigation.In R2, a more consistent CH 4 production was due to the lesser accumulation of VFAs and lesser pH variations from the optimum range as compared to R1. Consequent to the higher concentration of FW in R1, the VFAs accumulation was also high which caused an instability in the reactor.The highest CH 4 yield of 142 and 225 mL CH 4 g −1 VS added was observed in R1 and R2 as depicted in Fig. 6(iv).Whereas the cumulative CH 4 yield per 2.5 OLR is depicted in Fig. 6(v).In the current study the results obtained in R2 are more than the 157 mL CH 4 g −1 VS added reported by Kim et al. (2003) at a co-digestion ratio of 20:80.
The methane yield recorded at a feeding ratio of 1:1 in R1 was less than the 215 mL CH 4 g −1 VS added reported by Kim et al. (2003) at thermophilic operating conditions.Sosnowski et al. (2008) explored the co-digestion of sewage sludge and the organic fraction of municipal solid waste at a mixing ratio of 75:25 and reported high 439 mL CH 4 g −1 VS added whereas, with the same substrate, Cabbai et al. (2013) reported the 293 and 365 mL CH 4 g −1 VS added at the mixing ratios of 90:10 and 50:50, respectively.The microwave pre-treatment of a single substrate at a time at 3:2 feeding ratio of food waste and sewage sludge was investigated by Zhang et al. (2016) and they reported 316.24 and 338.44 mL CH 4 g −1 VS added using anaerobic wastewater sludge as inoculum.Moreover, Lee et al. ( 2019) have produced only around 21% more bio-methane than R1 despite using thermal alkali pre-treatment, which has high energy demand.Very few studies are available on the co-digestion of primary sewage sludge and domestic food waste without any pre-treatment.The comparative results of digestate of R1 and R2 after the stabilization of the reactor are shown in Table 7.The average VS removal efficiency in the case of R2 was recorded as 53.2% whereas, in R1, it reaches 77.5% as shown in Fig. 6(vi).The less HRT of 4 days can be insufficient for less VS reduction in R2 than R1.Despite less VS reduction, high biogas production and methane yield was observed in R2 due to a stable operation throughout the investigation.

Velocity and temperature profiles in digesters
Fig. 7(i) & (ii) demonstrate the velocity field vectors derived from the CFD simulation and post-processing analysis for both the lab-scale reactors R1 and R2.On the vertical plane going from the centre of the digester, the velocity field vectors are depicted for t = 0 and infinity.The disparity between the reactors' dead volume is large, with R1's dead volume being 31.48% and R2's being 66.64%, respectively.The dead volume inside this R2 reactor is even more than double that of the reactor R1.The reactors R1 and R2 have mean velocity gradients of 7.05 s −1 and 2.11 s −1 , respectively.The optimum velocity gradient for lab-scale reactors is recommended to be between 7.2 and 14.3 per second (Sindall et al., 2013).The   velocity gradient for R1 is close to this range in our situation, while the velocity gradient for R2 is quite low due to the presence of larger quantity of TS.
The sewage was heated with one 250-W heater, and also the impeller has been deployed to evenly distribute the heat.As illustrated in Fig. 1, the thermocouple was mounted on the opposite side of the heater to prevent it from the early and false reading.Fig. 8(i) & (ii) show the temperature profile at 1 min 30 s, 5 min 30 s, and 9 min 30 s for R1 and R2, respectively.The profiles depict the temperature variation on a vertical plane that passes through the reactor.Observing the regions near the thermocouple, it took about 9 min for the temperature to rise from 22 • C to 37 • C.However, as visible from Fig. 8(ii), the heat is not spread evenly all across the reactor R2.The impeller was successful in spreading the heat for the instance of R1, as depicted in Fig. 8(i).In the case of R1, the temperature across the reactor became higher than 37 • C from the beginning temperature of 22 • C in just 9 min.
The dead volume is proportional to the total quantity of solids present in the sludge, as evidenced from the velocity distributions published by Bridgeman (2012).Under the same working parameters, an increase in the amount of total solids results in a greater dead volume.In accordance with the literature (Leonzio, 2018), the mixing in the reactor R2, assessed according to the velocity gradient value, was heterogeneous.This type of mixing resulted in such temperature variation that was not spread across the reactor.R2 has to be optimized more to get a minimal dead volume and well-dispersed heat across the reactor.

Conclusions
Initial batch-bottle-reactor experiments suggested a co-digestion of less proportions of FW and CD with RW (or BFS).Metagenomic analysis indicated maintenance of optimum pH for methanogenic microbial growth for enhanced CH 4 yields.Following this, R1 (2%-FW) produced more biogas (and CH 4 ) than R2 (50%-FW) due to more VFA production in R1.According to CFD simulations, the digestion in R2 (TS 2.5%) was due to a uniform temperature distribution, whereas R1 had dead zones due to high TS content (12%).The mixing ratio of R2 can be integrated with STPs due to its higher stability.

NGS data availability:
NGS sequence reads for Samples used in this study can be accessed from NCBI under BioProject ID viz., PR-JNA828359SRA with respective SRA IDs (viz, SRR23283309, SRR23283294, SRR23283290, and SRR23283293).

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Fig. 3 .
Fig. 3. Stack bar plots representing the relative abundance of top ten dominant bacterial or archaeal taxa: (A) Top taxa at the phylum level; (B) Top taxa at the genus level; (C) Observed Archaeal genera.These plots include each sample at the starting (SP) and ending (EP) points of the anaerobic digestion of FW:CD:RW in the proportion of 1:1:8 respectively, for a period of 20 days.The remaining microbial community is marked under the category of ''Other''.Food waste (F), cow dung (C), and raw wastewater (R); FCR::1:1:8 mixture, NC: Negative control, E(C2)Tx: Synthetic microbial consortia.

Fig. 4 .
Fig. 4. Biogas production trends during the 90 days batch experiments in the upscaled reactors R1 (black) and R2 (red).(i) Biogas production, (ii) Bio-methane composition, (iii) Cumulative methane yield (iv) pH trends.(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 5 .
Fig. 5. Reactor stability parameter variation in the upscaled reactors R1 (black) and R2 (red) during the experiment cycle.(i) pH and (ii) VFA as acetic acid.(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 6 .
Fig. 6.Gas production in the upscaled reactors R1 (black) and R2 (red) during the entire experiment.(i) Biogas, (ii) Bio-methane composition, (iii) Biogas yield, (iv) Methane yield, (v) Cumulative methane yield, and (vi) VS reduction.(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Table 1
The experimental parameters used for the small-scale bottle batch bioreactors are provided.

Table 2
Initial conditions of inoculum and substrate.

Table 5
The number of elements, refinement ratio, GCI, maximum skewness, and the time required to complete 1500 iterations for the grids 5 to 1.

Table 7
Comparison of the physiochemical parameters in R1 and R2 digestates.