Work

Enhancing anaerobic co-digestion of cow manure and molasses residue with various support carriers

Year: 2023
Type: dissertation
Abstract: Cow manure (CM) contains pollutants such as growth hormones, antibiotics, heavy metals, and pathogens, whereas molasses residue (MR) has high organic and sugar content, acidity, chemical oxygen demand (COD), and lethal metal ions, posing risks to human health and the environment. Anaerobic Digestion (AD) is selected as the optimal waste control method for both CM and MR; however, its mono-digestion (AMoD) is not recommended as it causes inhibition. Hence, this study focused on the co-digestion (ACoD) of CM and MR to overcome these shortcomings. Despite MR's potential as a mono-substrate, CM is chosen due to overproduction in Malaysia and its higher environmental hazard. Biomethane Potential (BMP) tests were performed in various operational condition, including mono- and co-digestion, varying co-digestion ratios, and various support carrier applications (granulated activated carbon (GAC), synthetic grass (SG), nanoparticles (NPs), biochar (BC), and lava rock (LR)). BMP were applied using 30 bottles, which is kept at mesophilic temperature in a water bath. The BMP results guide the selection of substrate mixing ratios and support carriers for a semi-continuous system. Then, the semi-continuous study was carried out with comparison of suspended system (SR) and attached system (AR) employed with several support carriers, including SG, NPs, BC, and LR at mesophilic temperature. The sample used was co-digestion of CM and MR at 50:50 ratio, from Organic Loading Rates (OLR) 1 until OLR 6 g/L/day, in total of 100 days with daily feeding. After the attached system with SG, NPs, and BC was inhibited, combinations of support carriers (NPs+LR, BC+LR, BC+NPs, and BC+NPs+LR) were tested. Additional research on the machine learning model fitting between Support Vector Machines (SVM) and Artificial Neural Networks (ANN) was done, utilizing data from the BMP test. According to the batch study (BMP), co-digestion outperformed mono-digestion (CM) in terms of performance, and it was discovered that 50:50 was the ideal ratio. With the exception of GAC, every support carrier had shown potential, hence, among the five support carriers, only four are chosen, since they produced high amount of biogas. Based on the semi-continuous results, only LR significantly enhances performance in the attached system, where it has resulted in increased biogas output (155 mL), SMP (22.5 mLCH4/gVSadded), and VS elimination (59%). Meanwhile, combination of support carriers did not show any significant improvements, with LR still showing the best performance. For machine learning, two sets of ANN and SVM models were created, and it was discovered that the SVM model has greater forecast accuracy compared to ANN, and a more resilient system for controlling performance, and may operate as a forerunner in commercial applications. This study emphasized the potential of MR as a co-substrate for CM treatment and LR's role in improving reactor performance.
Language: English
Field: Energy
Open Access status: green
OpenAlex ID: W7204680520
Cites: 279
Cited by:
Related to: 10