From macrostructure to machine learning: Lava rock as a superior carrier in anaerobic co-digestion of manure and molasses residue
This study investigates the anaerobic co-digestion of cow manure (CM) and molasses residue (MR), focusing on the impact of various support carriers on reactor performance and machine learning model predictions. BMP tests identified a 50:50 CM:MR ratio as optimal for methane production, yielding the...
| Main Authors: | , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2025
|
| Online Access: | http://psasir.upm.edu.my/id/eprint/119694/ http://psasir.upm.edu.my/id/eprint/119694/1/119694.pdf |
| Summary: | This study investigates the anaerobic co-digestion of cow manure (CM) and molasses residue (MR), focusing on the impact of various support carriers on reactor performance and machine learning model predictions. BMP tests identified a 50:50 CM:MR ratio as optimal for methane production, yielding the highest biogas production (1540 mL), SMP (45.05 mLCH₄/gVSadded), and VS removal (51.4 %). Semi-continuous experiments were conducted with support carriers—lava rock (LR), nanoparticles (NPs), biochar (BC), and synthetic grass (SG), under mesophilic conditions with the 50:50 CM:MR ratio and organic loading rates of 1–6 gVS/L/day for 100 days. LR showed the best performance, producing the highest biogas (170 mL), SMP (22.5 mL CH₄/gVSadded), and VS removal (59.8 %). Compared to other support carriers, LR exhibited the largest pore size at 53.7 nm (92 % larger than BC and 88.6 % larger than NPs), which significantly enhanced nutrient diffusion and microbial accessibility. Machine learning models, including ANN and SVM, were developed from BMP data, with SVM showing superior predictive accuracy (R² = 0.84373) compared to ANN (R² = 0.71367). SEM and EPS analyses revealed a higher microbial population on LR than on BC. These results suggest LR's large pore size make it a promising support carrier for improving AD performance. |
|---|