Next Article in Journal
Environmental Impact of Uncontrolled Sites in the Central Region of Mexico: A Life Cycle Assessment
Previous Article in Journal
Advancing Innovative Climate-Resilient and Net-Zero Technologies to Enhance Rice Productivity and Sustainability Amidst Climate Change
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Sustainability
Volume 17
Issue 20
10.3390/su17209323
sustainability-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Sustainable Management of Wastewater Sludge Through Co-Digestion, Mechanical Pretreatment and Recurrent Neural Network (RNN) Modeling

by
Raid Alrowais
1,*,
Mahmoud M. Abdel-Daiem
2,
Basheer M. Nasef
3,
Amany A. Metwally
4 and
Noha Said
2,*
1
Department of Civil Engineering, College of Engineering, Jouf University, Sakakah 72388, Saudi Arabia
2
Environmental Engineering Department, Faculty of Engineering, Zagazig University, Zagazig 44519, Egypt
3
Computer and Systems Engineering Department, Faculty of Engineering, Zagazig University, Zagazig 44519, Egypt
4
Agricultural Engineering Department, Faculty of Agriculture, Zagazig University, Zagazig 44511, Egypt
*
Authors to whom correspondence should be addressed.
Sustainability 2025, 17(20), 9323; https://doi.org/10.3390/su17209323
Submission received: 5 October 2025 / Revised: 15 October 2025 / Accepted: 16 October 2025 / Published: 21 October 2025
Browse Figures
Versions Notes

Abstract

This study investigates the combined effect of wheat straw particle size and mixing ratio on the anaerobic co-digestion (ACD) of waste activated sludge under mesophilic conditions. Ten batch digesters were tested with varying straw-to-sludge ratios (0–1.5%) and particle sizes (5 cm, 1 cm, and <2 mm). Fine straw particles (<2 mm) at 1.5% loading achieved the highest removal efficiencies for TS (43.55%), TVS (47.83%), and COD (51.52%), resulting in a 140% increase in biogas yield and methane content of 60.15%. The energy recovery reached 14.37 kWh/kg, almost double the control. The developed Recurrent Neural Network (RNN) model (3 layers, 13 neurons, 500 epochs) predicted biogas production with 99.8% accuracy, a root mean square error (RMSE) of 0.0038, a mean absolute error (MAE) of 0.0093, and an R2 close to 1. These results confirm the potential of integrating agricultural residues into wastewater treatment for renewable energy recovery and emission reduction. This study uniquely integrates mechanical pretreatment of wheat straw with RNN-based modeling to enhance biogas generation and predictive accuracy. It establishes a dual-experimental AI framework for optimizing sludge–straw co-digestion systems. This approach provides a scalable, data-driven solution for sustainable waste-to-energy applications.

1. Introduction

Anaerobic digestion (AD) is a widely applied biological process for stabilizing wastewater sludge while simultaneously recovering renewable energy in the form of biogas [1,2,3,4]. Municipal wastewater treatment plants generate large volumes of activated sludge that, although rich in organic matter, show limited biodegradability and consequently produce relatively low methane yields when digested alone [5,6,7]. At the same time, agricultural crop residues, such as wheat straw, are produced in significant quantities worldwide, particularly in wheat-producing countries, and often remain underutilized [8,9,10,11,12,13]. Instead of open burning or landfilling, wheat straw represents a potential feedstock for renewable energy generation [9,13]. Therefore, combining sewage sludge with wheat straw in anaerobic co-digestion (ACD) presents a promising approach to enhance waste valorization while addressing environmental and energy challenges [1,5,14].
ACD improves process performance by balancing the nutrient composition of the feedstocks [9,13,15]. Activated sludge is generally nitrogen-rich but carbon-deficient, whereas wheat straw, as a carbon-rich biomass with a high C/N ratio, helps balance nutrients and stabilize microbial activity [10,16,17]. Several studies have shown that adding lignocellulosic residues like wheat straw to sludge enhances both the volumetric methane production and the specific methane yield [1,5,6,16,17,18]. Beyond improving energy recovery, co-digestion also contributes to better sludge stabilization, reducing environmental risks associated with final disposal [5,10]. Thus, integrating wheat straw with activated sludge digestion holds technical, economic, and environmental advantages.
Despite the promising advantages of co-digesting sewage sludge with wheat straw, one of the main challenges lies in the recalcitrant structure of straw [8,13]. Its high lignocellulosic content, particularly lignin, forms a physical barrier that restricts microbial access to cellulose and hemicellulose, making hydrolysis the rate-limiting step in AD [1,11,16,19,20]. To overcome this limitation, mechanical pretreatments including grinding, roll milling, and extrusion are widely applied to reduce particle size and disrupt straw fibers [8,13,16]. Reducing particle size markedly enhances the surface area accessible to microbial activity, improves solubilization of organic matter, and enhances enzymatic accessibility, thereby boosting biogas yields [9,19,21].
Simple physical pretreatments, such as mechanical grinding, can reduce particle size and sufficiently condition biomass for effective bioconversion, offering a cost-efficient alternative to more complex methods [22,23]. Mechanical pretreatments are widely applied at full scale to decrease feedstock particle size, thereby enhancing methane production as well as enhancing mixing, heat transfer, and mass transfer within biogas plant digesters [24]. Conversely, thermal, chemical, and biological pretreatments involve substantially higher installation and operational costs, which can limit the economic feasibility of bioenergy and bioproduct generation [16].
The incorporation of agricultural residues such as wheat straw into wastewater treatment systems aligns with global sustainability objectives, particularly those related to renewable energy generation and the transition toward a circular economy [6,16,25]. This approach supports the United Nations Sustainable Development Goals (SDGs), notably SDG 7 (Affordable and Clean Energy), SDG 12 (Responsible Consumption and Production), and SDG 13 (Climate Action) [26]. ACD not only facilitates renewable energy recovery but also mitigates greenhouse gas (GHG) emissions by preventing open-field burning of straw, a widespread practice that contributes significantly to air pollution [5,14]. Research on the ACD of sewage sludge and wheat straw has expanded considerably, with increasing emphasis on optimizing substrate ratios, reducing particle size, and improving pretreatment efficiency [5,11,16,17,20,25]. These investigations provide critical insights into system optimization, scalability, and techno-economic feasibility. Ultimately, ACD represents a sustainable waste-to-energy strategy that simultaneously addresses waste management challenges, enhances renewable energy production, and contributes to climate change mitigation [6,13]. Therefore, beyond its technological potential for waste valorization, ACD serves as a viable pathway toward resource-efficient, low-carbon development while fostering environmental awareness and responsible behavior [27].
Artificial Neural Networks (ANNs) are regarded as among the most powerful modeling techniques compared to conventional analytical and statistical approaches [2]. Designed by researchers to support forecasting, measurement, modeling, and decision-making, they have been widely adopted across various scientific fields, including mathematics, engineering, medicine, economics, and agriculture [5,20,28,29,30]. Unlike traditional methods, ANN-based systems do not rely on predefined assumptions or rigid mathematical formulations about the system under study. Instead, their strength lies in uncovering hidden patterns and complex relationships within experimental datasets, making them highly effective for modeling nonlinear and multifaceted processes [20]. Through iterative training, neural network approaches can learn and capture nonlinear dependencies, which underscores their value as predictive tools in diverse applications [17].
ANNs have emerged as powerful tools for predicting biogas production from complex feedstocks such as sewage sludge and agricultural residues [31,32,33]. Unlike traditional kinetic models that rely on simplified assumptions, ANNs can capture nonlinear relationships between multiple input parameters, including substrate composition, particle size, pretreatment conditions, organic loading rate, pH, and temperature [17,18]. This enables them to provide accurate forecasts of methane yield under varying operational scenarios. Studies have demonstrated that ANN-based models often outperform classical models like first-order kinetics or modified Gompertz in terms of predictive accuracy, with higher R2 values and lower root mean square error (RMSE) [10,16,17,31]. By applying ANN models, operators can optimize process parameters, detect potential failures, and design more efficient co-digestion strategies [18]. Thus, ANN modeling provides a valuable decision-support tool for advancing biogas technology and improving the sustainability of waste-to-energy systems.
This study primarily aims to explore how reducing the particle size of wheat straw influences biogas and methane production during ACD with wastewater activated sludge. The novelty of this study lies in its integrated approach, which combines experimental evaluation of wheat straw particle size reduction with the application of an advanced Recurrent Neural Network (RNN), which is selected for its efficiency, computational simplicity, and high predictive accuracy, enabling reliable forecasting of biogas production based on time, straw proportion, and particle size. While previous research has explored the benefits of co-digestion and mechanical pretreatment separately, this work uniquely integrates both aspects with the application of an advanced RNN model. This integration uniquely demonstrates how mechanical pretreatment enhances biodegradability and energy recovery, while the RNN-based modeling improves predictive performance and process optimization. In contrast, the co-digestion process mitigates GHG emissions by reducing reliance on open-field straw burning and uncontrolled sludge disposal. Furthermore, this study contributes to sustainability by promoting circular bio economy practices through waste-to-energy conversion, resource recovery, and reduced environmental footprint, aligning with global goals for sustainable waste management and renewable energy transition.

2. Methodology

2.1. Materials and Experimental Setup

Samples of waste-activated sludge were obtained from the oxidation ditch of Al-Duwadmi municipal wastewater treatment plant, in Riyadh Province, Saudi Arabia, and preserved at 4 °C until use. The sludge exhibited total solids (TSs) and total volatile solids (TVSs) contents of 1.41% and 0.99%, respectively, and a carbon to nitrogen (C/N) ratio of 6.71. No external seeding was added; all digesters were seeded with the collected waste-activated sludge, which served simultaneously as both substrate and inoculum due to its active microbial population adapted to wastewater treatment conditions. The total bacterial count in the sludge inoculum, as quantified by the plate count agar method, was approximately 15 × 106 CFU/mL.
Wheat straw was sourced from a local agricultural field and mechanically processed into three particle sizes: 5 cm, 1 cm, and <2 mm. The prepared straw was then sealed in plastic bags and kept at room temperature for later experiments. AD was carried out in ten laboratory-scale batch digesters, each with a 5 L working volume, filled with 2.5 kg of sewage sludge. Wheat straw was added to sludge at different ratios (0.5%, 1.0%, and 1.5% on a weight basis) to study the effects of feedstock loading and particle size reduction. The reactors were coded as follows: D1: sludge only (control), D2–D4: sludge + 0.5% straw (5 cm, 1 cm, <2 mm), D5–D7: sludge + 1.0% straw (5 cm, 1 cm, <2 mm), D8–D10: sludge + 1.5% straw (5 cm, 1 cm, <2 mm), see Table 1.

2.2. Reactor Setup and Analytical Methods

The experimental setup contained a glass anaerobic reactor that was connected to a gas collector and an open water jar to record gas displacement. The reactors were operated in batch mode under mesophilic conditions (35 ± 1 °C). They were placed in a thermostatically controlled water bath equipped with a heater to maintain a constant temperature and were tightly sealed to ensure anaerobic conditions throughout the experiment. Biogas production was evaluated by measuring the water volume displaced in the collector because of internal gas pressure, and the measured biogas volume was expressed as ml biogas per g of substrate. All gas measurements were performed at room temperature and atmospheric pressure. Following each biogas measurement, the reactors were manually agitated once per day to maintain homogeneous mixing of the substrates. Biogas composition for each reactor was analyzed three times, and the mean values were recorded. Samples were obtained using an airtight 10 μL syringe and subjected to analysis with a gas chromatograph (Agilent 7890A) (Agilent Technologies, Santa Clara, CA, USA) featuring a Thermal Conductivity Detector. TS were determined using the gravimetric drying method at 105 °C, and TVS were measured by loss on ignition at 550 °C, both following the standard procedures [34]. Total nitrogen (N) content was quantified using the standard Kjeldahl method [34]. Total carbon (C) was analyzed via elemental analysis using an Oxford xMET7500 Scrap Metal Analyzer (Oxford Instruments, Oxfordshire, UK). The chemical oxygen demand (COD) was determined by a spectrophotometer (DRB200 Reactor 1 Block 9X16 MM/2X20 MM) (HACH, Loveland, CO, USA), and pH values were recorded using an Ultrameter 6PII FCE pocket-sized pH meter (Myron, Carlsbad, CA, USA). All parameters were analyzed in triplicate, and the mean values were calculated.

2.3. Energy and Environmental Aspects

The potential electric energy of biogas produced from AD of different substrates can be estimated via the following equation and according to the procedure assessed in an earlier study [5]:
BE   =   0.278   ×   HV M   ×   η CHP   ×   Y B   ×   M %  
where BE is the potential electric energy of biogas produced per kg substrate in kWh, HVM is the methane heating value (57 MJ/m3), η CHP is the efficiency of the combined heat and power generation for combustion of methane (84%,) and YB is the total biogas production per kg substrate in m3, and M% is the methane percentage of biogas produced [5].
The main GHG emitted from this process include carbon dioxide generated from the AD process and combustion of methane to generate electricity. They can be estimated based on procedures reported by the previous literature [35]. To demonstrate the environmental benefits of AD technology, GHG emissions were calculated for the scenario of sludge disposal in an open environment (82.32 g CO2-eq/kg sludge), as well as from straw open field burning (2.64 kg CO2-eq/kg straw) [14,36].

2.4. Recurrent Neural Network (RNN) Modeling

The ANN is a computational model that simulates the structure and function of the brain and can be used to learn from input-output data with no need for specific programming. They are widely used for modeling, pattern recognition, forecasting, and other applications in various engineering fields. An ANN typically comprises an input layer, one or more hidden layers, and an output layer—simulated here in MATLAB (R2020a) [28,37].
Every neuron has a weighted input, a bias value, and an activation function that is used to compute the output of the neuron (e.g., Sigmoid, Tanh, Radbas). The quantity of hidden layers and neurons is based on the complexity of the problem, while input and output neurons are obviously matched to the variables found in the dataset [38,39].
RNNs, a deep learning architecture style, are designed to work with sequential and time-series data via loops and hidden states. This architecture enables RNNs to learn patterns over time, which is helpful for things like signal classification and prediction. In the model investigated, there are inputs of time, size, and percentage of straw that feed into the network. This uses tapped delays to make predictions on cumulative amounts, such as biogas production [40]. Figure 1 depicts the layout of an RNN with two tapped delays from the input. As can be seen in Figure 1, the input layer is the place from which the model inputs come; the factors time, wheat straw size, and straw percentage are the inputs that supply these. The hidden layer has two tap delays, and the output layer is the predicted cumulative biogas.

3. Results and Discussion

3.1. Experimental Results

The characteristics of sludge, straw, and their mixtures at different mixing ratios are summarized in Table 2. The TS content of the raw sludge was relatively low (1.41%), while wheat straw exhibited a very high TS value (92.8%), reflecting its lignocellulosic and fibrous nature [8]. When straw was added at 0.5%, 1%, and 1.5% levels, the TS content of the mixtures increased progressively to 1.98%, 2.3%, and 3.1%, respectively. A similar pattern was observed for TVS, where the sludge alone contained 0.99% TVS. In comparison, mixtures reached 1.01%, 1.7%, and 2.3%, indicating that the straw addition substantially enriched the organic fraction available for microbial degradation [5,16,41]. Similar results were reported in previous research, where adding 2% wheat straw to waste activated sludge increased TS from 1.43% to 3.22% and TVS from 1.00% to 2.33%, respectively [5]. Another study demonstrated that adding 3% wheat straw to waste activated sludge increased TS from 1.9% to 4.5% and TVS from 1.3% to 3.6%, respectively [41]
The COD of the sludge was 20 g/L, which increased consistently with straw supplementation to 24, 29, and 33 g/L for 0.5%, 1%, and 1.5% mixtures, respectively. This reflects the higher organic load introduced by straw, which provides additional biodegradable carbon to fuel AD [17,25]. Similar results were reported in previous research, where adding 2% wheat straw to waste activated sludge increased COD from 22 to 35 g/L [5]. Another study demonstrated that adding 3% wheat straw to waste activated sludge increased COD from 15 to 40 g/L [41]. However, the pH of the mixtures showed a slight decline with increasing straw addition (from 7.0 in sludge alone to 6.77 at 1.5% straw), possibly due to the release of volatile fatty acids (VFAs) [12,16]. Nevertheless, the pH remained within a range generally considered suitable for methanogenic activity [20,42]. Similar observations were noted in previous studies, which reported a slight decline in pH values following the addition of straw to sewage sludge [29,41]
Carbon and nitrogen fractions also shifted markedly. In sludge, carbon and nitrogen contents were 30.2% and 4.5% (TS basis), yielding a C/N ratio of 6.71. Straw, on the other hand, was carbon-rich (48.9%) and nitrogen-deficient (0.6%), with an extremely high C/N ratio (81.5). Mixing straw with sludge progressively improved the C/N balance, raising it to 9.86, 13.19, and 16.02 at 0.5%, 1%, and 1.5% straw addition, respectively. Blending straw with sludge preserved nutrient balance, promoting the growth of microorganisms vital for anaerobic digestion, as discussed and found in the previous literature [10,16,17,25]. Comparable findings were reported in earlier studies; for example, the addition of 2% wheat straw to waste activated sludge raised the C/N ratio from 6.64 to 17.85. Similarly, incorporating 3% wheat straw increased the C/N ratio from 6.56 to 20.60 [5,41]
Overall, the data clearly indicate that wheat straw supplementation enhanced the organic matter content (TS, TVS, and COD) and improved the C/N ratio of sludge, both of which are favorable for biogas production. The slightly lower pH observed in mixtures suggests enhanced acidogenesis but does not appear to have reached inhibitory levels. These findings align with previous reports [5,6,17,25], which emphasize that co-digestion of lignocellulosic biomass with nitrogen-rich sludge improves substrate balance, boosts biodegradability, and enhances methane yields.
The experimental results clearly demonstrate that both straw addition ratio and particle size influenced biogas yield, see Figure 2. Compared with the control (D1), which achieved 3.46 mL/g, the wheat straw addition led to improvements in the biogas production. At a 0.5% mixing ratio (D2–D4), biogas production increased modestly, ranging from 4.53 to 4.95 mL/g, corresponding to 31–43% improvement over sludge alone. When the straw ratio was raised to 1% (D5–D7), yields rose substantially to between 6.32 and 6.87 mL/g, reflecting 83–99% improvement. Further increasing straw to 1.5% (D8–D10) provided the highest outputs, with yields reaching 7.53–8.29 mL/g, equivalent to 118–140% enhancement. These trends are consistent with previous studies reporting that co-digestion of sludge with lignocellulosic biomass, such as wheat straw, enhances biogas production by supplying additional carbon sources and improving microbial activity. For instance, co-digestion of wheat straw (particle size <1 mm) with activated sludge at a 2% ratio resulted in a 168% increase in biogas production compared to mono-digestion of sludge [5]. Similarly, ACD of barley straw (particle size <1 mm) with waste-activated sludge at a 1% mixing ratio led to a 145% improvement relative to sludge digestion alone [20]. Other studies have reported that co-digesting sewage sludge with wheat and rice straw at various mixing ratios enhanced biogas and methane production by approximately 68–109% and 52–235%, respectively [1,10].
Particle size also played an important role: compared to 5 cm straw, reactors with 1 cm straw increased biogas production by approximately 3–4%, while those with finely ground straw (<2 mm, D4, D7, D10) achieved about 8–10% higher yields under the same mixing ratios. These findings confirm that co-digestion with straw not only boosts biogas potential by supplying additional carbon but also that mechanical pretreatment through size reduction enhances microbial accessibility and accelerates hydrolysis. These findings align with a previous study that reported an approximately 11% increase in biogas yield and 8% increase in methane yield during the ACD of microbial inoculum with milled wheat straw at a particle size of <2.4 mm compared to >2.4 mm [19]. ACD of wheat straw with sludge sourced from the digester of a municipal wastewater treatment plant resulted in a 21% higher methane yield when the straw particle size was reduced to less than 3 mm, compared to using chopped straw [21]. ACD of wheat straw and silage with different additives resulted in a 26% increase in methane potential when the straw particle size was reduced from 2 cm to 2 mm [43]. ACD of waste-activated sludge with wheat straw of particle size <2 mm in semi-continuous mode doubled biogas production compared to sludge mono-digestion [17]. ACD of rice straw and sludge in a semi-continuous AD achieved a 27% higher methane yield when the straw particle size was reduced to 1 mm compared to 2 cm [44]. ACD of sewage sludge with rice straw at a particle size of 0.075 mm produced 1.8 times more methane compared to straw of 2 cm size [44]. Anaerobic co-digestion of digested sludge with wheat, barley, and rye straw generated 1.7-, 1.3-, and 1.8-fold higher methane yields, respectively, when the straw was ground to 1 mm compared to 4 mm particle size [16]. These studies have reported and discussed comparable results; however, the observed differences among them can be attributed to variations in feedstock characteristics, particle size distribution, pretreatment methods, inoculum source, and operating conditions of the AD process.
Figure 3 shows the results of biogas analysis for different digesters. The methane concentration in biogas was highest in sludge mono-digestion (62.84% in D1) but decreased steadily with both straw addition and particle size reduction. At a 0.5% straw loading, CH4 content declined slightly from 62.25% (5 cm, D2) to 61.85% (1 cm, D3) and 61.78% (<2 mm, D4). At 1.0% straw, values dropped further from 61.44% (D5) to 60.98% (D6) and 60.92% (D7), while at 1.5% straw, the decrease was more pronounced, from 60.95% (D8) to 60.72% (D9) and finally 60.15% (D10). This trend coincided with a rise in CO2 and other gases from 3.30% in D1 to 7.93% in D10. The reduction in methane fraction can be attributed to VFA accumulation and rapid hydrolysis at finer straw sizes and higher loadings, which increase CO2 production and temporarily inhibit methanogens, thereby shifting gas composition away from methane [5,19,44]. However, despite this decrease in methane percentage, the total methane yield increased significantly with straw addition and size reduction, since smaller particles improved hydrolysis and substrate conversion, resulting in higher cumulative methane output. Similar findings were found and discussed by previous studies [5,13,16,19,21]. For example, a previous study found that while the methane concentration in biogas from mono-sludge digesters (59.23–63.12%) was slightly higher than that from straw–sludge co-digesters (55.33–61.14%), the total methane yield was higher in the co-digesters because of their greater overall biogas production [5]. Similarly, the results are consistent with a previous study, which found that the methane content during ACD of microbial inoculum with milled wheat straw was slightly lower for finer particles (<2.4 mm, 53.02%) compared to coarser ones (>2.4 mm, 54.7%) [19].
Figure 4a–d indicate the changes in TS, TVS, COD, and pH, respectively, before and after digestion across the ten experimental digesters. The results clearly demonstrate the dual effect of wheat straw addition and particle size reduction on the stabilization of sewage sludge and enhancement of AD efficiency. The AD process showed a significant decrease in TS, TVS, and COD values. This reduction is primarily due to the microbial breakdown of organic matter under anaerobic conditions, producing biogas in the process [5,19]. The removal ratios of TS, TVS, and COD serve as a direct indicator of substrate biodegradability and are consistent with the measured biogas production in Figure 2. The control digester (D1), containing only sludge, showed the lowest efficiency degradation, with TS dropping from 1.41% to 1.05% and TVS from 0.99% to 0.70%, corresponding to modest removal ratios of 25.5% and 29.29%, respectively. At the same time, COD declined from 20 g/L to 12.9 g/L, equivalent to a 35.53% removal. In contrast, co-digestion treatments displayed significant improvements, especially those with higher straw loading and finer particle sizes. At 0.5% straw addition (D2–D4), TS removals ranged from 28.28 to 31.82%, TVS removals from 31.68 to 35.64%, and COD removals from 38 to 41%, confirming that even small amounts of straw could enhance degradation through improved carbon availability and C/N balance. Increasing the straw ratio to 1.0% (D5–D7) further elevated the removal efficiencies, with TS removals reaching 35.65–39.13%, TVS removals 37.06–42.35%, and COD removals 45.34–48.28%, indicating a synergistic effect between substrate ratio and microbial accessibility. The highest performance was observed in the 1.5% straw group (D8–D10), where TS removals reached 43.55%, TVS removals reached nearly 47.83%, and COD removals peaked at 51.52% in D10 with finely ground straw (<2 mm), clearly illustrating that particle comminution is a decisive factor in accelerating hydrolysis and mineralization. COD levels after digestion in these reactors were reduced to 16–17.5 g/L, compared to initial values of 33 g/L, confirming efficient organic matter breakdown. The TS, vs. and COD removal efficiency results verified biogas generation across the different reactors, with reduction percentages showing a direct proportionality to biogas yields, as found and confirmed by earlier studies [20,42,45]. A slight pH increase was detected after digestion; however, values remained within the optimal methanogenic range (6.79–7.1) across all treatments, ensuring effective performance and stable operation of the AD process [46]. This variation can be explained by pH fluctuations through the biological conversion phase, in which acidogenic bacteria generate large amounts of organic acids that may accumulate and alter system stability. Under typical conditions, however, this pH drop is buffered by bicarbonate generated by methanogenic bacteria and by ammonia release [20,42]. Consistent results were obtained in earlier research, where a modest rise in pH values was observed after ACD of sewage sludge with straw [5,29].
The observed solid removals and degradation are consistent with previous findings [17,20,41]. The removal efficiencies of TS, VS, and COD from sludge alone were 43.4%, 54.0%, and 61.6%, respectively, while co-digestion with barley straw at 1% mixing ratio achieved higher removals of 47.03%, 56.32%, and 64.4%, respectively [20]. The ACD of waste-activated sludge with 2% wheat straw achieved higher removal efficiencies of TSs (49%), TVSs (56%), and COD (59%) compared to mono-sludge digestion, which recorded 29%, 36%, and 38%, respectively [5]. In sludge mono-digestion, TS, TVS, and COD reductions were 45.83%, 53.90%, and 62.50%, respectively; however, co-digestion with wheat straw at a 3% ratio resulted in enhanced reductions of 54%, 59%, and 66.67%, respectively [18]. Sludge mixed with microalgae showed TS, TVS, and COD reductions of 22.8%, 33.1%, and 46.7%, respectively, while the co-digestion of this mixture with wheat straw resulted in significantly higher reductions of 48.1%, 58.2%, and 77.5%, respectively [41]. When compared to coarser straw (>5 cm), which often leads to incomplete hydrolysis and lower gas yields due to the recalcitrant lignocellulosic structure, the present results demonstrate that fine straw particles dramatically increase microbial accessibility and enzymatic attack, leading to higher degradation efficiencies, as observed in earlier findings [19,21,44]. Therefore, Figure 4 not only confirms that co-digestion with wheat straw improves solids degradation, organic matter mineralization, and process stability, but also reinforces recent literature consensus that the combined effect of dosing strategies and particle size reduction is central to enhancing biogas production, improving sludge stabilization, and reducing environmental impacts in waste-to-energy systems.
Figure 5 demonstrates how both straw addition ratio and particle size reduction influenced biogas electric energy potential and GHG emissions generated from methane production and combustion for electricity generation. The control reactor (D1) produced the lowest energy recovery (6.29 kWh/kg substrate) but also the lowest GHG emissions (0.029 g CO2-eq/kg), reflecting the limited biodegradability of sludge alone. With 0.5% straw (D2–D4), energy recovery rose to 8.13–8.81 kWh/kg, while GHG emissions increased to 0.038–0.041 g CO2-eq/kg. At 1.0% straw (D5–D7), energy recovery reached 11.17–12.05 kWh/kg with GHG values of 0.052–0.056 g CO2-eq/kg, and at 1.5% straw (D8–D10), energy peaked at 13.31–14.37 kWh/kg, more than double the sludge-only case, but GHG emissions also rose to 0.061–0.066 g CO2-eq/kg. Across all ratios, particle size reduction amplified both outcomes: finely milled straw (<2 mm) produced the highest energy yields but also the greatest emissions, while coarse straw (5 cm) produced lower energy but also lower GHG within each ratio group. Importantly, the emissions reported here arise from methane production and combustion and are thus biogenic, meaning they originate from renewable organic matter rather than fossil fuels, which substantially reduces their net climate impact [5,14,47]. Furthermore, AD offers broader environmental benefits that offset these process emissions: diverting wheat straw from open-field burning avoids extremely high emissions (2.64 kg CO2-eq/kg straw [14] and using sludge in AD prevents uncontrolled releases from open disposal (82.32 g CO2-eq/kg sludge [36]. Therefore, although higher straw loading and finer particle sizes enhance energy recovery while slightly increasing biogenic GHG emissions, the overall environmental performance of ACD remains highly favorable, as it provides renewable energy, stabilizes sludge, minimizes methane emissions, and prevents the substantially greater emissions associated with straw burning and uncontrolled sludge disposal [7,25]. These findings align with recent literature, which consistently shows that straw treatment enhances energy recovery while AD significantly reduces life-cycle climate impacts compared with conventional disposal practices [7,16,19,25,44].

3.2. RNN Modeling Results

The cumulative biogas ratio was estimated using an RNN model. As the model initializes with randomly assigned connection weights and partitions data into patterns, its performance was evaluated across multiple runs to ensure reliability [5,20]. All simulations were executed on a PC equipped with a 2.8 GHz Core i5 processor and 16 GB of RAM, utilizing the ANN Toolbox in MATLAB R2020a.
The findings demonstrate the effectiveness of the RNN, configured with 13 neurons in its hidden layer, in accurately simulating the process. A tansigmoid (tansig) activation function was applied in the hidden layer, while training was conducted using backpropagation with Bayesian regularization. On average, approximately 200 training epochs were sufficient to achieve optimal performance while minimizing the risk of overfitting.
In this study, the RNN model was developed using MATLAB R2020a and trained on experimental data from the ACD of waste-activated sludge, incorporating time, wheat straw particle size, and straw proportion as input variables. Multiple activation functions were tested to evaluate model performance. As shown in Table 3, the tansig function for the hidden layers and the purelin function for the output layer yielded the lowest average RMSE, making them the most effective configuration for accurate predictions. The trainbr training algorithm outperforms both trainlm and the Scaled Conjugate Gradient backpropagation method (trainscg). Additionally, the tansig activation function appears to be the most effective, whereas other functions, such as radbas or the triangular basis (tribas), yield poorer performance, reflected by higher RMSE values.
Table 3 highlights that the tansig activation function outperformed alternatives such as the triangular basis (tribas) and radial basis (radbas) functions, which yielded higher RMSE values. Overall, the optimal RNN configuration was determined after extensive testing, demonstrating strong alignment with experimental data and reliable predictive accuracy under comparable conditions. Furthermore, to evaluate the performance of the optimal RNN neural network model in predicting bio-gas production, several validation indices were calculated, as presented in Table 4. These indices include the RMSE, mean absolute error (MAE), and the coefficient of determination (R2). The optimal RNN model achieved notably low prediction error values during both the training and testing phases, indicating its high predictive accuracy. Therefore, the model can be directly applied for forecasting process characteristics, providing a rapid and efficient means of analyzing the biogas production process and potentially reducing analysis time. However, adjustments or retraining may be required if input parameters or experimental conditions vary.
Essentially, the optimal RNN training configuration was identified through comparative testing of different options. The resulting model performs very well with the current experimental data, indicating its ability to produce accurate predictions under similar conditions. However, if the experimental parameters or input values change, retraining or structural adjustments would be necessary. Nonetheless, the approach used here to select the most suitable regression model remains highly valuable for guiding future studies.
Therefore, Figure 6 illustrates the regression relationship between the observed values (targets) and the outputs of the RNN model during training, validation, and testing. The vertical axis represents the regression line equation, while the corresponding correlation coefficients (R) are also presented for each dataset. The model achieved nearly perfect performance, with R-values of 0.99993, 0.99995, and 0.99993 for testing, validation, and training, respectively. These results, generated using the MATLAB plot regression function, demonstrate that the RNN predictions closely align with experimental data. Overall, the findings confirm that the developed RNN model is highly accurate and efficient, reliably capturing the system’s behavior across all phases of evaluation.
The RNN demonstrates good performance, owing to its strong ability to capture nonlinear relationships and accurately map input–output patterns. Furthermore, this model serves as an efficient alternative to empirical methods, which are often time-consuming and require costly prototypes. The RNN can be applied under similar boundary conditions without the need for physical experiments. For different parameters or boundary conditions, the model can still be employed after appropriate training and validation, enabling reliable prediction of biogas production.
Figure 7 presents the RNN model predictions for the entire dataset, alongside the results of hyperparameter optimization used to determine the most suitable network structure and training parameters.

4. Conclusions

These results clearly demonstrate that ACD of waste-activated sludge with wheat straw, particularly when mechanically pretreated to fine particle sizes, enhances digestion performance, biogas yield, and energy recovery. At the highest straw loading (1.5%) with finely ground straw (<2 mm), removal efficiencies of TS, TVS, and COD reached 43.55%, 47.83%, and 51.52%, respectively, compared to only 25.53%, 29.29%, and 35.50% in the control. Biogas production rose from 3.46 mL/g in sludge mono-digestion to 7.53–8.29 mL/g with 1.5% straw, representing 118–140% improvement, while finer straw particles yielded approximately 10% more biogas than coarse particles. The energy recovery increased markedly, reaching 14.37 kWh/kg, more than double that of sludge alone. The RNN modeling demonstrated notable accuracy in forecasting cumulative biogas production. With a remarkably low RMSE of 0.0038, MAE of 0.0093, and R2 closely to 1, the model’s predictions exhibited a high degree of precision, closely approximating the observed values.
From a sustainability standpoint, the proposed co-digestion approach offers a viable pathway toward integrated waste valorization and renewable energy recovery. This alignment with sustainable development priorities highlights the potential of ACD as a scalable solution for achieving low-carbon, resilient bioenergy systems. Overall, the study advances knowledge by linking mechanical pretreatment, co-digestion performance, and RNN-based forecasting into a comprehensive strategy for sustainable biogas generation. However, the findings are limited to laboratory-scale experiments and future studies should investigate pilot- and full-scale implementation, assess long-term performance and evaluate economic feasibility.

Author Contributions

All authors whose names appear on the submission made substantial contributions. Conceptualization, R.A., M.M.A.-D., B.M.N., A.A.M., and N.S. methodology, M.M.A.-D., B.M.N., and N.S. The first draft of the manuscript was written by R.A., M.M.A.-D., B.M.N., A.A.M., and N.S.; revised and presented with suggestive comments about the previous versions of the manuscript, R.A., M.M.A.-D., B.M.N., A.A.M., and N.S. All authors have read and agreed to the published version of the manuscript.

Funding

Deanship of Graduate Studies and Scientific Research at Jouf University under grant No. (DGSSR-2025-FC-01035).

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

The data for this work can be found within the article, and for further data, feel free to contact the corresponding authors.

Acknowledgments

This work was funded by the Deanship of Graduate Studies and Scientific Research at Jouf University under grant No. (DGSSR-2025-FC-01035).

Conflicts of Interest

The authors confirm that there is no conflict concerning the publication of this manuscript.

References

  1. Azarmanesh, R.; Qaretapeh, M.Z.; Zonoozi, M.H.; Ghiasinejad, H.; Zhang, Y. Anaerobic co-digestion of sewage sludge with other organic wastes: A comprehensive review focusing on selection criteria, operational conditions, and microbiology. Chem. Eng. J. Adv. 2023, 14, 100453. [Google Scholar] [CrossRef]
  2. Korni, R.D.; Bora, T.; Majji, A.; Panda, J. Liquisolid technique for solubility enhancement of a poorly soluble thrombin inhibitor: Optimization using design of experiments and artificial neural networks. Prospect. Pharm. Sci. 2025, 23, 97–109. [Google Scholar]
  3. Jafari, G.; Ardabili, S.; Pourdarbani, R.; Abbaszadeh, B.; Hernandez-Hernandez, M. Sustainable biomethane production from sewage sludge and wheat straw co-digestion in the presence of polypyrrole Fe3O4 nanoparticles and alkaline pretreatment: Life cycle assessment point of view. Acta Technol. Agric. 2023, 3, 133–141. [Google Scholar] [CrossRef]
  4. Gao, J.; Li, J.; Wachemo, A.C.; Yuan, H.; Zuo, X.; Li, X. Mass conversion pathway during anaerobic digestion of wheat straw. RSC Adv. 2020, 10, 27720–27727. [Google Scholar] [CrossRef]
  5. Alrowais, R.; Said, N.; Alotaibi, A.; Hatata, A.; Essa, M.A.; Abdel daiem, M.M. Comparing the effect of mesophilic and thermophilic anaerobic co-digestion for sustainable biogas production: An experimental and recurrent neural network model study. J. Clean. Prod. 2023, 392, 136248. [Google Scholar] [CrossRef]
  6. Rani, P.; Bansal, M.; Pathak, V.V. Experimental and kinetic studies for improvement of biogas production from KOH pretreated wheat straw. Curr. Res. Green Sustain. Chem. 2022, 5, 100283. [Google Scholar] [CrossRef]
  7. Said, N.; Alrowais, R.; Abdel-Daiem, M.M.; Mostafa, N.A. Bio-Waste to Bioenergy: Critical Assessment of Sustainable Energy Supply Chain in Egypt. Resources 2025, 14, 131. [Google Scholar] [CrossRef]
  8. Said, N.; Alrowais, R.; Abdel-Daiem, M. Effect of physical and chemical treatment on the characteristics of wheat straw as fuel for energy applications. BioResources 2024, 19, 953. [Google Scholar] [CrossRef]
  9. Elsayed, M.; Andres, Y.; Blel, W. Anaerobic co-digestion of linen, sugar beet pulp, and wheat straw with cow manure: Effects of mixing ratio and transient change of co-substrate. Biomass Convers. Biorefin. 2023, 13, 11831–11840. [Google Scholar] [CrossRef]
  10. Rahmani, A.M.; Tyagi, V.K.; Ahmed, B.; Kazmi, A.A.; Ojha, C.S.P.; Singh, R. Critical insights into anaerobic co-digestion of wheat straw with food waste and cattle manure: Synergistic effects on biogas yield and kinetic modeling. Environ. Res. 2022, 212, 113382. [Google Scholar] [CrossRef] [PubMed]
  11. Zdeb, M. Anaerobic Digestion of Wheat Straw Pretreated with Soaking in Water and Alkali Medium. J. Ecol. Eng. 2021, 22, 246–254. [Google Scholar] [CrossRef]
  12. Rajput, A.A.; Sheikh, Z. Effect of inoculum type and organic loading on biogas production of sunflower meal and wheat straw. Sustain. Environ. Res. 2019, 29, 4. [Google Scholar] [CrossRef]
  13. Kamusoko, R.; Mukumba, P. Potential of wheat straw for biogas production by anaerobic digestion in South Africa: A review. Energies 2024, 17, 4662. [Google Scholar] [CrossRef]
  14. Shang, X.; Song, S.; Yang, J. Comparative environmental evaluation of straw resources by LCA in China. Adv. Mater. Sci. Eng. 2020, 2020, 4781805. [Google Scholar] [CrossRef]
  15. Sarkar, O.; Rova, U.; Christakopoulos, P.; Matsakas, L. Biogas potential of organosolv pretreated wheat straw as mono and co-substrate: Substrate synergy and microbial dynamics. Sci. Rep. 2024, 14, 18442. [Google Scholar] [CrossRef]
  16. Álvaro, A.G.; Palomar, C.R.; Redondo, D.H.; Torre, R.M.; de Godos Crespo, I. Simultaneous production of biogas and volatile fatty acids through anaerobic digestion using cereal straw as substrate. Environ. Technol. Innov. 2023, 31, 103215. [Google Scholar] [CrossRef]
  17. Abdel daiem, M.M.; Hatata, A.; Said, N. Modeling and optimization of semi-continuous anaerobic co-digestion of activated sludge and wheat straw using Nonlinear Autoregressive Exogenous neural network and seagull algorithm. Energy 2022, 241, 122939. [Google Scholar] [CrossRef]
  18. Abdel daiem, M.M.; Hatata, A.; Galal, O.H.; Said, N.; Ahmed, D. Prediction of biogas production from anaerobic Co-digestion of Waste Activated sludge and wheat straw using two-dimensional mathematical models and an artificial neural network. Renew. Energy 2021, 178, 226–240. [Google Scholar] [CrossRef]
  19. Witaszek, K.; Kupryaniuk, K.; Kupryaniuk, J.; Panasiewicz, J.; Czekała, W. Optimization of Straw Particle Size for Enhanced Biogas Production: A Comparative Study of Wheat and Rapeseed Straw. Energies 2025, 18, 1794. [Google Scholar] [CrossRef]
  20. Lee, W.; Kim, Y.; Kim, H.; Kim, M. Comparison of anaerobic co-digestion of food waste and livestock manure at various mixing ratios under mesophilic and thermophilic temperatures. Sustainability 2024, 16, 7653. [Google Scholar] [CrossRef]
  21. Victorin, M.; Davidsson, Å.; Wallberg, O. Characterization of mechanically pretreated wheat straw for biogas production. Bioenergy Res. 2020, 13, 833–844. [Google Scholar] [CrossRef]
  22. Stanley, J.T.; Thanarasu, A.; Kumar, P.S.; Periyasamy, K.; Raghunandhakumar, S.; Periyaraman, P.; Devaraj, K.; Dhanasekaran, A.; Subramanian, S. Potential pre-treatment of lignocellulosic biomass for the enhancement of biomethane production through anaerobic digestion-A review. Fuel 2022, 318, 123593. [Google Scholar] [CrossRef]
  23. Uthirakrishnan, U.; Sharmila, V.G.; Merrylin, J.; Kumar, S.A.; Dharmadhas, J.S.; Varjani, S.; Banu, J.R. Current advances and future outlook on pretreatment techniques to enhance biosolids disintegration and anaerobic digestion: A critical review. Chemosphere 2022, 288, 132553. [Google Scholar] [CrossRef]
  24. Garuti, M.; Sinisgalli, E.; Soldano, M.; Fermoso, F.G.; Rodriguez, A.J.; Carnevale, M.; Gallucci, F. Mechanical pretreatments of different agri-based feedstock in full-scale biogas plants under real operational conditions. Biomass Bioenergy 2022, 158, 106352. [Google Scholar] [CrossRef]
  25. Al-Da’asen, A.; Al-Harahsheh, A.; Al-Hwaiti, M.; Irshaid Irshaid, F. Biogas production via anaerobic codigestion of chemically treated wheat straw with sewage sludge or cow manure. Biomass Convers. Biorefin. 2024, 14, 5505–5516. [Google Scholar] [CrossRef]
  26. SDGR. The Sustainable Development Goals Report; The United Nations Statistics Division: New York, NY, USA, 2025. [Google Scholar]
  27. Al Husban, W. The Impact of Integrating Sustainable Development Goals on Students’ Awareness and Pro-Environmental Behavior: A Case Study of Jordan. Sustainability 2025, 17, 2588. [Google Scholar] [CrossRef]
  28. Isenkul, M.E.; Güneş-Durak, S.; Kocak, Y.P.; Pir, İ.; Tüfekci, M.; Demirkol, G.T.; Sevgen, S.; Çığgın, A.S.; Tüfekci, N. Predicting biogas production in real scale anaerobic digester under dynamic conditions with machine learning approach. Environ. Res. Commun. 2025, 7, 65016. [Google Scholar] [CrossRef]
  29. Ashraf, M.T.; Sieborg, M.U.; Yde, L.; Rhee, C.; Shin, S.G.; Triolo, J.M. Biomethanation in a thermophilic biotrickling filter—pH control and lessons from long-term operation. Bioresour. Technol. Rep. 2020, 11, 100525. [Google Scholar] [CrossRef]
  30. Shroff, K.C.; Shah, N.G. Prediction Modelling to Enhance Anaerobic Co-digestion Process of OFMSW and Bio-flocculated Sludge Using ANN. Pollution 2024, 10, 481–494. [Google Scholar]
  31. Ling, J.Y.X.; Chan, Y.J.; Chen, J.W.; Chong, D.J.S.; Tan, A.L.L.; Arumugasamy, S.K.; Lau, P.L. Machine learning methods for the modelling and optimisation of biogas production from anaerobic digestion: A review. Environ. Sci. Pollut. Res. 2024, 31, 19085–19104. [Google Scholar] [CrossRef]
  32. Adeleke, O.; Olatunji, K.O.; Madyira, D.M.; Jen, T.-C. Application of multimodal machine learning-based analysis for the biomethane yields of NaOH-pretreated biomass. Sci. Rep. 2025, 15, 24372. [Google Scholar] [CrossRef]
  33. Maruejols, L.; Höschle, L.; Yu, X. Energy independence, rural sustainability and potential of bioenergy villages in Germany: Machine learning perspectives. Int. Food Agribus. Manag. Rev. 2025, 1, 1–33. [Google Scholar] [CrossRef]
  34. Rice, E.W.; Bridgewater, L.; Association, A.P.H. Standard Methods for the Examination of Water and Wastewater; American Public Health Association: Washington, DC, USA, 2012; Volume 10. [Google Scholar]
  35. Jafari-Sejahrood, A.; Najafi, B.; Faizollahzadeh Ardabili, S.; Shamshirband, S.; Mosavi, A.; Chau, K. Limiting factors for biogas production from cow manure: Energo-environmental approach. Engineering Appl. Comput. Fluid Mech. 2019, 13, 954–966. [Google Scholar] [CrossRef]
  36. IPCC. Climate Change 2023: Synthesis Report; Intergovernmental Panel on Climate Change (IPCC): Geneva, Switzerland, 2023. [Google Scholar]
  37. Alharbi, H.S. Efficient Swell Risk Prediction for Building Design Using a Domain-Guided Machine Learning Model. Buildings 2025, 15, 2530. [Google Scholar] [CrossRef]
  38. Ibrahim, A.F.; Hussein, M.A. Leveraging machine learning for prediction and optimization of texture properties of sustainable activated carbon derived from waste materials. Sci. Rep. 2025, 15, 11313. [Google Scholar] [CrossRef] [PubMed]
  39. Habchi, S.; El Bari, H. Optimization study using response surface methodology and artificial neural networks on the co-digestion of food waste and poultry slaughterhouse waste, with hydrochar as an effective enhancer. Bioresour. Technol. Rep. 2025, 29, 102061. [Google Scholar] [CrossRef]
  40. Sherstinsky, A. Fundamentals of recurrent neural network (RNN) and long short-term memory (LSTM) network. Phys. D Nonlinear Phenom. 2020, 404, 132306. [Google Scholar] [CrossRef]
  41. Ahmed, D.; Wagdy, R.; Said, N. Evaluation of biogas production from anaerobic co-digestion of sewage sludge with microalgae and agriculture wastes. BioResources 2019, 14, 8405–8412. [Google Scholar] [CrossRef]
  42. Hamzah, N.; Tokimatsu, K.; Yoshikawa, K. Solid fuel from oil palm biomass residues and municipal solid waste by hydrothermal treatment for electrical power generation in Malaysia: A review. Sustainability 2019, 11, 1060. [Google Scholar] [CrossRef]
  43. Gallegos, D.; Wedwitschka, H.; Moeller, L.; Zehnsdorf, A.; Stinner, W. Effect of particle size reduction and ensiling fermentation on biogas formation and silage quality of wheat straw. Bioresour. Technol. 2017, 245, 216–224. [Google Scholar] [CrossRef]
  44. Dai, X.; Hua, Y.; Dai, L.; Cai, C. Particle size reduction of rice straw enhances methane production under anaerobic digestion. Bioresour. Technol. 2019, 293, 122043. [Google Scholar] [CrossRef] [PubMed]
  45. Turek, A.; Wieczorek, K.; Wolf, W.M. Digestion procedure and determination of heavy metals in sewage sludge—An analytical problem. Sustainability 2019, 11, 1753. [Google Scholar] [CrossRef]
  46. Kitessa, W.M.; Fufa, F.; Abera, D. Biogas Production and Biofertilizer Estimation from Anaerobic Co-Digestion of Blends of Wastewater and Microalgae. Int. J. Chem. Eng. 2022, 2022, 3560068. [Google Scholar] [CrossRef]
  47. Hu, Y.; Ma, P.; Yang, Z.; Liu, S.; Li, Y.; Li, L.; Wang, T.; Siddique, K.H.M. The responses of crop yield and greenhouse gas emissions to straw returning from staple crops: A meta-analysis. Agriculture 2025, 15, 408. [Google Scholar] [CrossRef]
Sustainability 17 09323 g001
Figure 1. An RNN architecture incorporating a hidden layer.
Figure 1. An RNN architecture incorporating a hidden layer.
Sustainability 17 09323 g001
Sustainability 17 09323 g002
Figure 2. Cumulative biogas production from different digesters.
Figure 2. Cumulative biogas production from different digesters.
Sustainability 17 09323 g002
Sustainability 17 09323 g003
Figure 3. Biogas analysis for different digesters.
Figure 3. Biogas analysis for different digesters.
Sustainability 17 09323 g003
Sustainability 17 09323 g004
Figure 4. Characteristics before and after digestion for different digesters (a) TS, (b) TVS, (c) COD, and (d) pH.
Figure 4. Characteristics before and after digestion for different digesters (a) TS, (b) TVS, (c) COD, and (d) pH.
Sustainability 17 09323 g004
Sustainability 17 09323 g005
Figure 5. Biogas electric energy potential and GHG emissions from methane production and combustion.
Figure 5. Biogas electric energy potential and GHG emissions from methane production and combustion.
Sustainability 17 09323 g005
Sustainability 17 09323 g006aSustainability 17 09323 g006b
Figure 6. Cumulative biogas linear regression relationships: (a) Training, (b) Validation, (c) Testing and (d) All.
Figure 6. Cumulative biogas linear regression relationships: (a) Training, (b) Validation, (c) Testing and (d) All.
Sustainability 17 09323 g006aSustainability 17 09323 g006b
Sustainability 17 09323 g007
Figure 7. A graphical representation of the optimal RNN model prediction versus experimental data on the cumulative biogas production.
Figure 7. A graphical representation of the optimal RNN model prediction versus experimental data on the cumulative biogas production.
Sustainability 17 09323 g007
Table 1. Experimental digesters for ACD of sludge and wheat straw.
Table 1. Experimental digesters for ACD of sludge and wheat straw.
DigesterD1D2D3D4D5D6D7D8D9D10
Sludge mass (kg)2.52.52.52.52.52.52.52.52.52.5
Straw ratio (%) 0.00.50.50.51.01.01.01.51.51.5
Straw particle size –5 cm1 cm<2 mm5 cm1 cm<2 mm5 cm1 cm<2 mm
Table 2. Characteristics of raw materials and mixtures.
Table 2. Characteristics of raw materials and mixtures.
ParametersSludgeStrawSludge + 0.5% StrawSludge + 1% StrawSludge + 1.5% Straw
TS (%)1.4192.801.982.303.10
TVS (%)0.9975.601.011.702.30
COD (g/L)20.00ND24.0029.0033.00
pH7.006.206.906.856.77
C (% TS)30.2048.9016.3322.9029.02
N (% TS)4.500.601.661.741.81
C/N6.7181.509.8613.1916.02
Table 3. Effect of the training algorithm and the hidden layer’s activation function in the RNN model.
Table 3. Effect of the training algorithm and the hidden layer’s activation function in the RNN model.
The Structure of the RNN ModelActivation Function of the Hidden Layer’sErrorTraining Algorithm
TrainlmTrainscgTrainbr
(3–13–1)tansigRMSE0.00640.02670.0038
radbasRMSE1.12911.52491.3144
tribasRMSE1.04811.87759.3391
A bold number means the optimal case.
Table 4. Effect of the number of neurons used in the hidden layer of each model.
Table 4. Effect of the number of neurons used in the hidden layer of each model.
ModelError# of Neuron
35101315
RNN modelRMSE0.03550.00820.00510.00380.0142
R20.99730.99970.99991.00000.9999
MAE0.09570.02880.01340.00930.0121
A bold number means the optimal case.
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Alrowais, R.; Abdel-Daiem, M.M.; Nasef, B.M.; Metwally, A.A.; Said, N. Sustainable Management of Wastewater Sludge Through Co-Digestion, Mechanical Pretreatment and Recurrent Neural Network (RNN) Modeling. Sustainability 2025, 17, 9323. https://doi.org/10.3390/su17209323

AMA Style

Alrowais R, Abdel-Daiem MM, Nasef BM, Metwally AA, Said N. Sustainable Management of Wastewater Sludge Through Co-Digestion, Mechanical Pretreatment and Recurrent Neural Network (RNN) Modeling. Sustainability. 2025; 17(20):9323. https://doi.org/10.3390/su17209323

Chicago/Turabian Style

Alrowais, Raid, Mahmoud M. Abdel-Daiem, Basheer M. Nasef, Amany A. Metwally, and Noha Said. 2025. "Sustainable Management of Wastewater Sludge Through Co-Digestion, Mechanical Pretreatment and Recurrent Neural Network (RNN) Modeling" Sustainability 17, no. 20: 9323. https://doi.org/10.3390/su17209323

APA Style

Alrowais, R., Abdel-Daiem, M. M., Nasef, B. M., Metwally, A. A., & Said, N. (2025). Sustainable Management of Wastewater Sludge Through Co-Digestion, Mechanical Pretreatment and Recurrent Neural Network (RNN) Modeling. Sustainability, 17(20), 9323. https://doi.org/10.3390/su17209323

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop