Abstract
Abstract A novel acid-phase anaerobic dynamic membrane bioreactor (AnDMBR), inspired by the stomach of ruminant animals, was operated to achieve efficient hydrolysis at a hydraulic retention time (HRT) of 12 h with minimal membrane fouling. This rumen-mimicking AnDMBR achieved volatile fatty acid (VFA) yields of 0.4 ± 0.2 g VFA/g volatile solids (VS) fed over three years of continuous operation with both mono-digestion of food waste and co-digestion of food waste and wastewater sludge. The bioreactor retained core rumen-associated microbes, such as Prevotella, Prevotella_7, and Bifidobacterium, supporting its adaptability across different feedstocks. These results highlight the rumen-AnDMBR as a robust option for producing VFAs from diverse organic waste streams, suitable for downstream methane or medium chain fatty acid production. Introduction Anaerobic digestion (AD) is widely used to convert organic wastes into renewable bioenergy and nutrient-rich digestate but faces challenges like long retention times. Anaerobic membrane bioreactors (AnMBRs) have been employed for high-rate anaerobic treatment, retaining biomass and enabling operation at high solids retention time (SRT). However, AnMBRs are challenged by membrane fouling, particularly with high-solids waste streams, which reduces flux and increases maintenance costs (Osman et al., 2023). AnDMBRs offer a cost-effective alternative by using low-cost materials to replace conventional membranes and employing energy-efficient fouling mitigation strategies, such as backwashing and relaxation (Hu et al., 2018). We developed a novel AnDMBR that simulates the rumen environment and rumination processes, such as regurgitation and mastication (Fonoll et al., 2024). In this system, the hydraulic retention time (HRT) is decoupled from the SRT by forming a dynamic membrane on a mesh that supports the growth of microbes enhancing hydrolysis and fermentation (Mason and Stuckey, 2016). By periodically breaking the biofilm, similar to regurgitation and mastication in ruminants, the system reduces the inhibitory effects of high volatile fatty acids (VFA) concentrations, creating a favorable environment for microbial activity and enhancing hydrolysis. The rumen-AnDMBR successfully produced an average VFA yield of 0.55 ± 0.12 g of VFA/g of VS during mono-digestion of food waste (FW) at an OLR of 18 ± 2 g VS/Lreactor/d (Fonoll et al., 2024). In comparison, conventional AD systems that operate acid-phase treatment of FW typically produce around 0.3--0.4 g VFA/ g VS at an organic loading rate (OLR) below 11 g VS/Lreactor/d (Jiang et al., 2013; Lim et al., 2008). FW typically has high compositional variability compared to feedstocks like animal manure or wastewater sludge leading to challenges such as poor hydrolysis and VFA accumulation in mono-digestion systems (Dalke et al., 2021). Co-digestion of FW with organic wastes such as animal manure, lignocellulosic waste, and sludge can mitigate these issues by providing alkalinity, supplemental nutrients, and diverse microbial communities (Karki et al., 2021). However, FW generally offers a higher methane yield than other organic wastes, which can decrease with co-digestion due to dilution effects (Li et al., 2020; Luo et al., 2024). Co-digestion of FW with sludge has not been previously explored in rumen-mimicking AD systems. The lack of co-digestion studies with rumen-mimicking systems motivated us to study co-digestion of FW and sludge in a rumen-inspired AnDMBR. We demonstrated the long-term (over three years) stability of the rumen-AnDMBR process under FW mono-digestion and FW and sludge co-digestion. The hydrolysis performance data were complemented with 16S rRNA gene sequencing data to understand changes in a few key rumen microbial populations in response to operational changes. Materials and Methods Liquid and solid rumen contents collected from a fistulated cow at the Michigan State University dairy farm (East Lansing, MI, USA) and first-phase digestate obtained from the Flint Water Resource Recovery Facility (WRRF) (Flint, MI, USA) were used to inoculate the rumen-AnDMBR. FW was collected from two dining halls at the University of Michigan (Ann Arbor, MI). A blend of 60% primary and 40% thickened waste activated sludge, referred to as wastewater sludge in this study, was collected from the holding tank at the Ann Arbor WRRF (Ann Arbor, MI, USA). A 6.2 L AnDMBR was operated to mimic rumen conditions (Fig. 1) where a dynamic membrane containing a cake layer of biofilm and solids was formed, broken, and reformed on a 147 µm pore size mesh support during a 21 h period each day. The pH in the rumen-AnDMBR was maintained between 6.2--6.3 by automatically dosing 2.5 N NaOH. The operating conditions for the rumen-AnDMBR at different periods are summarized in Table 1. The FW, wastewater sludge, bulk digestate, and permeate samples were collected twice a week for chemical analyses. Total and suspended solids concentrations (APHA, 2012) were measured to adjust SRT; while COD concentrations (Hach, Loveland, CO, USA), and VFAs (Fonoll et al., 2024) were analyzed to determine the hydrolysis rate and VFA yield observed under different operating conditions. 16S rRNA gene sequencing was performed on the digestate samples to evaluate the microbial community structure and identify key microbes responsible for hydrolysis function. DNA extraction, amplification, and sequencing procedures have been detailed in Fonoll et al. (2024). The DNA sequences were analyzed using DADA2 (Callahan et al., 2016). The hydrolysis rate constant (d-1) was calculated using a steady-state particulate COD mass-balance approach in a continuously stirred tank reactor: khyd =[(Qin x Cin) - (Qdig x Cdig + Qperm x Cperm)]/ Cdig x V , where Qin, Qdig, and Qperm represent the flow rates of feed, wasted digestate, and permeate in the rumen bioreactor, respectively. Cin, Cdig, and Cperm represent the particulate COD concentrations of feed, digestate, and permeate, respectively. V is the volume of the bioreactor. The net VFA yields, reported in this study in terms of COD, are values calculated after subtracting the concentration of the respective VFAs in the feed. Results and Discussion Rumen-AnDMBR achieved high VFA yields and microbial adaptability under varied feedstock conditions The hydrolysis rate constants were consistently maintained across both mono- and co-digestion periods, highlighting the rumen-AnDMBR's adaptability to varied feedstocks, even with the dilution effect introduced by sludge (Fig. 2). The average net VFA yields in the rumen-AnDMBR was 0.42 ± 0.18 g VFA/g VS for FW mono-digestion and 0.35 ± 0.18 g VFA/g VS for co-digestion, reaching the upper limits of the typical range. This sustained VFA yield demonstrates the rumen-AnDMBR's robustness in maintaining hydrolysis and fermentation efficiency at a higher loading rate, even when feedstock characteristics vary, positioning it as a resilient alternative to conventional systems. The relative abundance of select genera was investigated to understand the retention of known hydrolytic rumen microbial populations, originating from the rumen inoculum, in the rumen-AnDMBR (Fig. 3). The genera Prevotella, Ruminococcus, Bifidobacterium, and Fibrobacter, typically present in the rumen microbiome, are highly efficient in cellulose and hemicellulose degradation, which are critical for VFA production (Mizrahi et al., 2021). Prevotella and Bifidobacterium maintained their abundance across mono- and co-digestion periods. Prevotella, which peaked during the mono-digestion period, was replaced by Prevotella_7 as a dominant genus in the co-digestion period, reflecting the system's ability to adapt to changing feedstock conditions. The fiber-degrading genera Ruminococcus and Fibrobacter exhibited relatively low abundances, likely attributed to the low fiber content of the feedstock. However, the increase in the relative abundance of Ruminococcus during the co-digestion period indicates some potential for fiber degradation. With steady hydrolysis rates, high VFA yields at high OLRs, and adaptable microbial dynamics, the rumen-AnDMBR demonstrates strong potential as a scalable, resilient solution for sustainable waste processing across diverse waste streams. Acknowledgements We thank Steve Donajkowski, Ethan Kennedy, and Tom Yavaraski for technical assistance. Thanks also to the staff at Michigan State University dairy farm, Flint and Ann Arbor WRRFs, and the University of Michigan dining halls. We also thank Yuang Guo, Victor Luk, Jenna Kutscher, Evan Zalek, Brianna Jimenez, Olivia Sherman, Sofia Martinez Cantu, and Alisson Villarreal Hurtado for their contributions to this work. This work was funded by the Department of Energy (DE-FOA-0002203_FY20_BETO_Multi-Topic).
This paper was presented at the WEF Residuals & Biosolids and Innovations in Treatment Technology Joint Conference, May 6-9, 2025.
Author(s)Karki, Renisha, Starostka, Renata, Lakshminarasimman, Narasimman, Puente, Pedro, Fairley-Wax, Timothy, Zhu, Kuang, Skerlos, Steven, Raskin, Lutgarde
Author(s)R. Karki1, R. Starostka1, N. Lakshminarasimman1, P. Puente1, T. Fairley-Wax1, K. Zhu1, S. Skerlos1, L. Raskin1
Author affiliation(s)University of Michigan, 1University of Michigan Dept of Civil & Env Eng, 1University of Michigan, 1University of Michigan, 1University of Michigan, 1University of Michigan Dept of Civil & Env Eng, 1University of Michigan, 1
SourceProceedings of the Water Environment Federation
Document typeConference Paper
Print publication date May 2025
DOI10.2175/193864718825159766
Volume / Issue
Content sourceResiduals and Biosolids Conference
Word count14