The kinetics of process dependent ammonia inhibition of methanogenesis from acetic acid

Журнал(книга):

Раздел науки:

Читать онлайн: 
Текст статьи: 
Christopher Allen Wilson a,*, John Novak e, Imre Takacs b, Bernhard Wett c, Sudhir Murthy d a Greeley and Hansen, 9020 Stony Point Parkway, Suite 475, Richmond, VA 23221, United States b Dynamita, Bordeaux, France c University of Innsbruck, Innsbruck, Austria d District of Columbia Water and Sewer Authority, Washington, DC, United States e Virginia Polytechnic Institute and State University, Blacksburg, VA, United States a r t i c l e i n f o Article history: Received 9 August 2011 Received in revised form 16 August 2012 Accepted 19 August 2012 Available online 7 September 2012 Keywords: Acetic acid Ammonia inhibition Kinetic modeling Methanogenesis Thermal hydrolysis a b s t r a c t Advanced anaerobic digestion processes aimed at improving the methanization of sewage sludge may be potentially impaired by the production of inhibitory compounds (e.g. free ammonia). The result of methanogenic inhibition is relatively high effluent concentrations of acetic acid and other soluble organics, as well as reduced methane yields. An extreme example of such an advanced process is the thermal hydrolytic pretreatment of sludge prior to high solids digestion (THD). Compared to a conventional mesophilic anaerobic digestion process (MAD), THD operates in a state of constant inhibition driven by high free ammonia concentrations, and elevated pH values. As such, previous investigations of the kinetics of methanogenesis from acetic acid under uninhibited conditions do not necessarily apply well to the modeling of extreme processes such as THD. By conducting batch ammonia toxicity assays using biomass from THD and MAD reactors, we compared the response of these communities over a broad range of ammonia inhibition. For both processes, increased inhibitor concentrations resulted in a reduction of biomass growth rate (rmax ¼ mmax∙X ) and a resulting decrease in the substrate half saturation coefficient (KS). These two parameters exhibited a high degree of correlation, suggesting that for a constant transport limited system, the KS was mostly a linear function of the growth rate. After correcting for reactor pH and temperature, we found that the THD and MAD biomass were both able to perform methanogenesis from acetate at high free ammonia concentrations (equivalent to 3e5 g/L total ammonia nitrogen), albeit at less than 30% of their respective maximum rates. The reduction in methane production was slightly less pronounced for the THD biomass than for MAD, suggesting that the long term exposure to ammonia had selected for a methanogenic pathway less dependent on those organisms most sensitive to ammonia inhibition (i.e. aceticlastic methanogens). ª 2012 Elsevier Ltd. All rights reserved. 1. Introduction The anaerobic digestion of sludges is, and has been for many years, a broadly applied operation in biological wastewater treatment. Anaerobic digestion has many distinct benefits, including a low biomass yield, a potentially desirable energy balance, and its ability to treat very high concentrations of organic waste. These advantages have long been considered to outweigh the disadvantages of very large high hydraulic retention time (HRT) systems, and relatively poor effluent * Corresponding author. Tel.: þ1 804 929 4040 E-mail address: cwilson@greeley-hansen.com (C.A. Wilson). Available online at www.sciencedirect.com journal homepage: www.elsevier.com/locate/watres w a t e r r e s e a r c h 4 6 ( 2 0 1 2 ) 6 2 4 7 e6 2 5 6 0043-1354/$ e see front matter ª 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.watres.2012.08.028quality. These drawbacks in conjunction with higher costs of biosolids management have fueled the development of processes that can be more heavily loaded (and thus potentially smaller) while outperforming the conventional digestion systems of the past in terms of sludge minimization and energy recovery. One issue that arises in the operation of advanced anaerobic digestion systems is that of methanogenic sensitivity to chemical and environmental stressors. Thermophilic and high solids anaerobic digestion processes have been studied from the standpoint of ammonia inhibition (Angelidaki and Ahring, 1994; Karakashev et al., 2006; Liu and Sung, 2002) and these processes are understood to be wholly different in their biology and biochemistry than conventional anaerobic digestion (Iranpour and Cox, 2006). In order to make use of these new processes, designers must understand the distinctions between advanced and conventional digestion processes in order to apply predictive modeling tools. Compelling evidence exists that the consideration of the unionized forms of both methanogenic substrates and inhibitors (Angelidaki and Ahring, 1994; Eldem et al., 2004; Sung and Liu, 2003) is more determinative of the kinetics of aceticlastic methanogenesis. Specifically, methane production from VFA by acetate and propionate acclimatized anaerobic digester biomass has been observed to be dependent on both total organic acid concentration and pH, or rather, solely dependent on the concentration of undissociated acid species (Fukuzaki et al., 1990a, 1990b). The dependence of methane production rates on nondissociated free acetic (HAc) and propionic (HPr) acid concentrations followed the second order substrate inhibition model originally proposed by Andrews (1978), suggesting that the effects of pH on methanogenic efficiency typically observed for anaerobic systems may be a result of the dependence on either substrate availability or inhibition. Similarly, ammonia inhibition has been related to the presence of free unionized ammonia (NH3) driven by changes in pH (Eldem et al., 2004), thermophilic temperature (Angelidaki and Ahring, 1994), or both (Sung and Liu, 2003). It is interesting that kinetic data derived from a simple nonpH-dependent model for methanogenic degradation of TAA and TPA is still widely employed (Kugelman and Chin, 1971; Lawrence and McCarty, 1969; VanLier et al., 1996). Perhaps separate processes with very similar characteristic pH could be adequately modeled and evaluated without consideration of the unionized substrates and inhibitors. However, the development of advanced anaerobic digestion processes which operate at the margins of the well understood physiologic ranges of anaerobic digestion (e.g. high temperature, high/low pH, high concentrations of chemical inhibitors) make the inclusion of substrate/inhibitor speciation critical. The objective of this study is to investigate the kinetics of methanogenesis from HAc under a condition of processdependent inhibitory conditions. Thermal hydrolytic pretreatment followed by mesophilic anaerobic digestion (THD) was studied, as well as a side by side conventional mesophilic digestion process (MAD). THD was chosen since abnormally high solids loading to the anaerobic digester is facilitated by reduced viscosity during thermal pretreatment (Kepp et al., 2000; Potts and Jolly, 2004). Destruction of protein and other complex organics results in high characteristic reactor pH and stoichiometric production of ammonia during THD (Kepp et al., 2001). Thus, this process is particularly amenable to this study. By presenting kinetic data from these different processes, the importance of ammonia and acetic acid speciation for the performance of diverse anaerobic digestion processes is recognized. 2. Materials and methods 2.1. Anaerobic digester operation Conical high-density polyethylene reactors supplied by Hobby Beverage Equipment Company (Temecula, California) were used as pilot scale anaerobic digesters. The conical bottom of these vessels was thought to be advantageous in terms of mixing of dense high solids sludge feed to THD. The nominal volume of the reactor vessels was 25 L. Reactors were operated with active volumes of 15 L, and daily batch feeding was performed at a rate of 1 L/day. The reactors were modified to accept a threaded stainless steel thermometer. Reactor temperatures were 37 C and 42 C for MAD and THD, respectively. Reactors were continuously mixed by digester gas recirculation from the headspace through the conical reactor bottom. Conventional mesophilic anaerobic digesters (MAD) were initially seeded with sludge from Pepper’s Ferry Regional Wastewater Treatment Plant (Radford, VA). Thermally hydrolyzed sludge digesters (THD) were initially seeded with reconstituted dewatered biosolids from Ringsend Wastewater Treatment Works (Dublin, Ireland) which was commissioned in 2003 and operates THD. Digester feed was obtained from DCWASA Blue Plains and consisted of a 1:1 ratio of primary and secondary solids on a total solids basis. Sludge to be subjected to MAD was shipped on ice directly to Virginia Tech (Blacksburg, VA). These solids had a concentration of approximately 6% total solids (TS). Sludge that was thermally pretreated was dewatered to approximately 15% TS by DCWASA and shipped on ice to RDP Technologies, Inc. (Norristown, PA) for thermal treatment. Thermal treatment took place at 150 C and 4.8 bar for approximately 30 min. Sludge hydrolyzate was then shipped on ice to Virginia Tech. These solids had concentration of approximately 12% TS. Both reactors had been operating for greater than a year at the time of this study. Other operating data for the digesters is shown in Table 1. 2.2. Batch ammonia toxicity assays Ammonia toxicity assays were conducted on biomass from semi-continuous flow lab-scaled digesters. Incubations with total ammonia nitrogen (TAN) concentrations of 16e230 mM (220e3220 mg/L) as N for THD and 36e357 mM (500e5000 mg/ L) as N for MAD were performed. Ammonia was added as a 7.15 M stock solution of NH4HCO3 in deionized water. Initial acetic acid concentrations for MAD incubations were 80e100 mM, and were 220 mM for THD. Acetate was added as a 10% w/w glacial acetic acid stock. Such different ranges of ammonia and acetic acid for MAD and THD were necessary in order to achieve similar ranges in free unionized species in 6248 w a t e r r e s e a r c h 4 6 ( 2 0 1 2 ) 6 2 4 7 e6 2 5 6reactors exhibiting differing pH levels. Each serum bottle received 10 mL of digester biomass, 10 mL 1.0 M phosphate buffered saline solution (prepared according to the natural pH of the parent digesters), appropriate volumes of ammonia and acetic acid stock solutions, and filled to 100 mL with distilled water. Measurement of pH showed that natural buffering and buffering provided externally by addition of phosphate buffer was sufficient to prevent deviation in pH in any serum bottle by more than 0.10 pH units over the course of the assays. Based on the small degree of variation of monitored pH, it is assumed that the partitioning constants described in Table 1 for acetic acid and ammonia are accurate throughout the course of the assays. Biomass samples were effectively diluted 10 in order to reduce the carryover of TAN and volatile fatty acids (VFA) from the parent digesters so that low ammonia batch incubations could be performed. Bottle headspaces were purged with nitrogen gas before sealing with a butyl rubber stopper. Serum bottles were submerged in a shaking water bath at the either 37 C (MAD) or 42 C (THD) according to the operating temperature of the parent digester. Biogas production was periodically monitored via a liquid column manometer filled with pH indicating acid/salt solution as described in Standard Methods (APHA, 2005). Bottles were sampled periodically for acetic acid concentration and pH. Samples for VFA analysis were passed through a 0.45 mm nitrocellulose membrane filter and frozen prior to analysis. Thawed samples were acidified by adding 85% phosphoric acid at a rate of 1% v/v and analyzed via HP5890 GC-FID and Supelco Nukol column with the following gas flow rates: (Nitrogen) 14 mL/min, (Air) 450 mL/min, (Hydrogen) 44 mL/ min, (Helium) 16 mL/min. The initial oven temperature was set to 80 C and ran isothermally for 3 min, then increased at 6 C/min for 10 min. Triplicate serum bottles were operated for each assay condition, and analytical triplicates were conducted for acetic acid analysis from each serum bottle at each sampling time. Progress curves, reflecting measured total acetic acid concentrations multiplied by the appropriate conjugate acid fraction for acetic acid shown in Table 1, and corresponding model fits (described below) for each of the batch assays are included within the supplemental material. 2.3. Kinetic parameter estimation The Monod expression relating the rate of substrate utilization (rs) to substrate concentration can be written as:  dS dt ¼ rs ¼  mmax Y $ S KS þ S$X (1) where S is the concentration of the limiting substrate in its bioavailable form (referred to herein as CL), t is time in days, mmax, Y, and X are the maximum specific growth rate, yield coefficient, and concentration of the active population degrading CL, and KS is the half-saturation constant for CL, each expressed in terms of chemical oxygen demand (COD). This expression can be integrated as: t ¼  Y rmax KS$lnC C LO L  þ CL  CLO where rmax ¼ mmax$X (2) The use of the modeled parameter rmax to represent the product of mmax and X is necessary since an accurate estimate of the concentration of the microbial population actively degrading acetic acid cannot be readily obtained. This parameter has the units of mg CODBiomass/L-day. In this way, the actual concentration of active biomass was not directly measured or modeled within this study. The decision to use Equation (2), which does not account for the impact of biomass growth throughout the ammonia toxicity assays, to model acetic acid degradation over time was based primarily on the shape of the progress curves (included within the supplemental material). It is acknowledged that a significant mass of acetic acid was degraded throughout the course of batch assays and this would naturally result in a corresponding growth of acetic acid degrading biomass. However, the initial slope to the acetic acid progress curves was essentially constant while substrate is not limiting, as opposed to the increasing negative slope that one would expect if (a) biomass growth were contributing to the removal rate of acetic acid from the bulk liquid, or (b) if substrate toxicity, as described by Andrews (1978), were occurring during the early period of the assay and was later relieved via acetic acid degradation. It is unclear by which mechanism, or combination of mechanisms, that the rate of acetic acid removal as shown in the progress curves remained relatively constant. Equation (2), which does not account for these factors, was chosen within this study for parameter estimation since it was descriptive of the data collected and provided a means to quantify the impact of free ammonia on acetic acid degradations rates within these two cultures. Non-linear least squares regression was used to fit acetic acid progress curve data to the integrated Monod expression via Microsoft Excel’s solver add-in. The inclusion of CLO as a fitting parameter provides a measure of quality assurance, as the fitted parameter can be directly compared against analytical data. An estimate of the yield coefficient for anaerobic growth on acetic acid of 0:04 gbiomass=gacetic acid was applied as a constant for both MAD and THD assays, which is consistent with previous literature estimates of biomass yields of acetate enrichment cultures (Lawrence and McCarty, 1969; Kugelman and Chin, 1971; Bhattacharya, 1986). Estimates of uncertainties of fitted parameters from progress curve data were performed using a method described in detail elsewhere (Smith et al., 1997, 1998). In general, the Table 1 e General operating parameters of parent (labscaled) anaerobic digesters. Parameter MAD THD General SRT 15 days 15 days Temp 37 C 42 C pH 7.35 7.85 Ammonia TAN, mg/L as N 1400  80 2380  170 fA, ammonia 0.972 0.890 pKa, ammonia 8.93 8.79 VFA VFATOTAL, mg/L as HAc 230  50 5700  160 fA, acetic acid 2.617  103 8.461  104 pKa, acetic acid 4.76 4.76 w a t e r r e s e a r c h 4 6 ( 2 0 1 2 ) 6 2 4 7 e6 2 5 6 6249uncertainty estimation calculates an approximate 95% confidence interval for each fitting parameter based on the effect on the model fit of a small change in that fitting parameter’s value (i.e. specific parameter sensitivity), and also the sum of squared errors (i.e. overall model fit quality) across the entire progress curve. A joint (3-parameter) 95% percent confidence interval was calculated for each ammonia concentration based on a one-way ANOVA (F-test). The resulting joint confidence interval represents all points within a 3- dimensional (3D) ellipsoid which lack statistical difference from the model-fitted parameters (e.g. CLO,FIT). Because we were primarily interested in the values of rmax, and KS, the joint confidence intervals presented in this study are shown as 2D ellipses that represent a slice through the 3-parameter joint confidence interval at CLO ¼ CLO,FIT. Initial slopes of cumulative biogas production curves were used to determine the degree of inhibition imposed by unionized ammonia. These data are plotted as the fraction of initial biogas production rate remaining versus the concentration of unionized ammonia. For example, a reduction in the initial biogas generation rate by 20% would be plotted as an inhibition factor INH3 of 0.80, thus a high INH3 value (i.e. close to 1.0) would be indicative of little methanogenic inhibition. Previous research shows that ammonia inhibition of aceticlastic methane generation is relatively mild at low unionized ammonia concentrations and follows a logistic reduction in methanogenesis as ammonia is increased, theoretically leading to full inhibition of methanogenesis at sufficiently high concentrations (Angelidaki and Ahring, 1994). As such, the following logistic model has been applied to calculated LINH3 values (the L prefix referring to the logistic model applied measured INH3 values) via least-squares regression (Wett et al., 2009): LINH3 ¼ 1 1 þ eBðKI;NH3Þ½NH3 (3) where KI;NH3 is the molar unionized ammonia concentration at which INH3 ¼ 0.50 and B is equal to the slope of the function at a molar unionized ammonia concentration equal to KI;NH3. Acid-base partitioning was calculated based on published pKa values for ammonia and acetic acid, and corrected for temperature using the constant enthalpy form of the van’t Hoff equation (CRC, 2009). 3. Results and discussion 3.1. Values and trends of kinetic fitting parameters Kinetic parameters obtained from non-linear least squares regression, relative to HAc, are presented in Table 2. The low magnitude of KS values is striking in comparison with previously published TAA-based data. Furthermore, the reported HAc half-saturation values between 0.45 and 6.8 mg/L for anaerobic digester communities in this study are of the same order of magnitude as literature reported values for various microbial communities pertinent to wastewater treatment (Button, 1985). Single parameter confidence intervals shown in Table 2 reveal a higher degree of certainty for MAD data, which were largely driven by the higher frequency of data collection for MAD ATA experiments. However, characteristic model fits to HAc progress curve data for both processes suggest high-quality fitting parameter estimates. In addition, model-fitted CLO values varied minimally from intended spiked HAc concentrations. Progress curves showing averages of triplicate assays, and corresponding model curves are included in the supplemental materials. A direct linear correlation exists between corresponding model-estimated values for KS and rmax (Fig. 1). Because rmax is dependent both on the maximum specific growth rate, mmax, and the active HAc-degrading population concentration, X, it is concluded that KS is not a constant, but rather, dependent on the microbial growth rate. Such phenomena in which KS is observed to vary linearly with biomass growth rate have been described in the past in both aerobic and anaerobic processes (Lawrence and McCarty, 1969; Novak, 1974). As NH3 concentration increases, rmax is observed to decrease as a result of kinetic inhibition, suggesting non-competitive inhibition as has been described by previous researchers (Karakashev et al., 2005, 2006). Considering a family of Monod curves in which rmax is observed to vary, the different shapes of the Monod curves (and as such, the different values of rmax) can be obtained either by shifts in KS or by shifts in the initial slopes of the Monod curves. This slope has been identified elsewhere as a particular community’s specific affinity for a substrate of interest (Button, 1985; Laanbroek and Gerards, 1993). It is noteworthy to state that Button (1985) have defined the parameter describing specific affinity, denoted as aO A, as having the units of L/g cells-hr while we have defined aO A as in Laanbroek and Gerards which is mathematically equivalent to the ratio of the parameters rmax/KS, as shown below in Equation (4). Specific Affinity aO A ¼ mg substrate mg biomass  day  mg biomass L volume  L volume mg substrate (4) This parameter is particularly well suited for describing the affinity of a biomass as it is indicative of the effect of increased substrate concentrations on growth at very low substrate concentrations (CLO z 0). The linear correlation observed between KS and rmax for THD and MAD assays suggests a constant affinity over various levels of NH3, meaning that the inhibition of rmax rather than variations in specific affinity dictate the shapes of the family of Monod curves obtained from these kinetic values (Fig. 2). While the implication of the covariance of rmax and KS on the constant substrate affinity of THD and MAD communities is clear, the cause of this covariance is not well described in the literature. For sludges comprised of floc-forming particles that exhibit somewhat viscous bulk fluid properties, it is expected that diffusion limited transport of substrate to actively degrading cells dictates the concentration of substrate at the cellular interface that drives biological reactions. For a diffusion limited system, this substrate concentration is inherently lower than that measured from the bulk fluid and used for kinetic determinations. Previous observations have suggested that unstirred boundary layers adjacent to a biological membrane tend to place artificial upward bias on measured 6250 w a t e r r e s e a r c h 4 6 ( 2 0 1 2 ) 6 2 4 7 e6 2 5 6half-saturation coefficients (Winne, 1973). It is possible that the reduction in rmax at increasing inhibitor concentration reduces the effect of diffusion limitation across unstirred boundary layers near the sludge floc surface, and this the artificial upward bias on KS at high rmax is alleviated. This finding suggests that under for diffusion limited constant transport of bulk-phase substrate to the surface of a biological floc where uptake occurs, the magnitude of KS is dependent, as its name implies, on saturation kinetics. Hence, the values of kinetic parameters obtained through modeling of biological systems are specific to the controlled experimental conditions, especially those that dictate substrate transport and utilization to and by the biomass (e.g. inhibitor and substrate concentration, incubation temperature, biomass and particulate matter densities, etc.). The result of this relationship is that parameter estimates of the half saturation constant at high (i.e. less inhibited) values of rmax are likely indicative of higher “apparent” KS values. The estimates of KS at lower (more inhibited) rmax values are less subject to this mass transfer limitation, as the lower rate of substrate utilization at the cell surface results in a relatively smaller DCL across the diffusion layer, and thus are likely closer estimates to the mass transfer limitation free half saturation value. 3.2. NH3 inhibition of maximum growth rates, rmax A plot of rmax values for both THD and MAD against either TAN or NH3 concentrations shows that increased nitrogen concentration has a clear negative effect on apparent substrate affinity. Growth rate reductions of 63% and 76% were achieved at the highest ammonia concentrations tested in this study for MAD and THD, respectively. Based on data in Fig. 3a, one would conclude that the impact of increased ammonia loads HAc utilization is more dramatic for THD compared with MAD. For example at a TAN concentration of 1220 mg/L as N, the THD process is approximately 50% A B Fig. 1 e Correlation between modeled KS and rmax (mmax∙X) for the (A) THD and (B) MAD process at various ammonia concentrations. Gray ellipses represent thin slices of 95% joint confidence intervals for the three curve fitting parameters: KS, rmax, CO. Displayed slice of each 3-dimensional ellipsoid taken at CO [ CO (Modeled). Both the overall correlation between KS and rmax, as well as the inclination of each ellipsoid in the xey plane suggest as high degree of dependence between the values obtained for KS and rmax. Due to the tightness of fit to MAD data, graphical display of joint confidence intervals is not practical and is thus omitted. Table 2 e Summary of modeled kinetic parameters for MAD and THD processes as effected by TAN concentration. Reactor Ammonia Concentration mg/L as N KS (mg/L as HAc) rmax (mg/L-day) C0 (mg/L as HAc) R2 Free NH3 TAN MAD 15 500 6.8  0.19 0.593  0.009 13.8  0.20 0.998 35 1250 5.7  0.39 0.492  0.018 13.5  0.48 0.990 56 2000 5.2  0.05 0.455  0.003 13.3  0.07 1.00 77 2750 4.0  0.12 0.377  0.006 13.5  0.21 0.999 77 2750 4.1  0.06 0.384  0.003 16.2  0.11 0.999 98 3500 3.2  0.03 0.309  0.001 16.1  0.07 1.00 119 4250 2.5  0.03 0.265  0.002 16.2  0.09 0.999 140 5000 1.6  0.11 0.217  0.008 16.8  0.37 0.996 THD 24 220 5.3  2.0 0.125  0.039 11.6  1.2 0.995 79 720 3.8  1.8 0.093  0.033 11.4  1.5 0.987 134 1220 2.3  1.1 0.063  0.019 10.9  1.3 0.980 189 1720 1.6  0.72 0.052  0.013 11.2  1.1 0.979 299 2720 0.93  0.44 0.040  0.007 11.2  0.85 0.980 354 3220 0.45  0.24 0.030  0.004 10.6  0.54 0.986 w a t e r r e s e a r c h 4 6 ( 2 0 1 2 ) 6 2 4 7 e6 2 5 6 6251inhibited, whereas the MAD process shows 17% inhibition at a TAN concentration of 1250 mg/L as N. However, high solids loading to THD resulted in a stoichiometric production of bicarbonate alkalinity based on volatile solids destroyed (Kepp et al., 2001; Wilson et al., 2011) and resulted in a characteristic pH of 7.85, compared with 7.35 for MAD. Such different pH conditions within these separate anaerobic digestion processes support the necessity for considering the substrate and inhibitor speciation. Taking pH and temperature effects into account, the conjugate acid fractions of ammonia ( fA) for the THD and MAD processes are 0.890 and 0.972, respectively. This means that at characteristic process pH and any given TAN concentration, the functional inhibitor concentration is approximately four times higher in THD. A plot of growth rate reduction due to NH 3 inhibition reveals that biological growth on acetate is very similarly affected by increased NH3 concentration for both digestion processes (Fig. 3b). However, an important distinction for the operation of these processes is made by considering not only the partitioning of total ammonia to the conjugate base form, but also the typical TAN concentration of the two digesters. Applying the calculated fA values to the typical TAN concentrations for both THD and MAD, in situ NH3 concentrations are approximately 260 and 40 mg/L N, respectively. At these NH3 concentrations, we would expect that biological growth on acetate would be inhibited by 15% for MAD, and by 65% for THD. While these numbers are exacerbated by the high solids loading to THD, it is clear that comparative evaluations of A B Fig. 2 e Derived Monod curves using model-estimated kinetic parameters for THD (A) and MAD (B). Data labels refer to TAN concentrations. The initial slope of each curve, aO A, is equivalent to rmax/KS. A B Fig. 3 e Maximum growth rates (rmax) for acetic acid degradation expressed as a fraction of rmax at the lowest ammonia level. The apparent effect of TAN concentration on growth rate inhibition is greater for THD than MAD (A); however, considering only the unionized form of ammonia (NH3), it is clear that THD and MAD respond similarly to increasing inhibitor concentrations (B). Vertical lines denote in-situ nitrogen species concentrations for THD (dashed line) and MAD (solid line). 6252 w a t e r r e s e a r c h 4 6 ( 2 0 1 2 ) 6 2 4 7 e6 2 5 6MAD and THD could not be accurately conducted without consideration for free ammonia inhibition. 3.3. Implications of KS values for THD operation Optimization of reactor operation requires knowledge of factors limiting biological growth, and by association, substrate utilization. High VFA concentrations after THD represent a significant loss of recoverable energy as methane gas. Additionally, undegraded soluble organics must receive further treatment, either as a dewatered centrate side stream process or as a return flow to the liquid side treatment plant. Unlike with NH3, the lower pH of MAD tends to increase the bioavailable fraction of HAc compared with the higher pH THD process; however, the magnitude of VFA accumulation results in higher reactor substrate concentrations (CL) for THD (4.8 mg/L HAc) than MAD (0.60 mg/L HAc). Compared with respective modeled values for KS, digester HAc concentrations are much higher than KS for THD at 260 mg/L NH3eN (0.95 mg/ L HAc) and much less than KS for MAD at 40 mg/L NH3eN (5.8 mg/L HAc). Practically, THD is operated in the portion of its Monod growth curve where the substrate utilization rate (dCL/dT ) is governed by rmax, whereas the substrate utilization rate for MAD is more strongly dependent on KS. Achieving lower effluent concentrations of VFA from THD requires either (a) providing a longer SRT to achieve more complete stabilization of HAc (O’Rourke, 1968), or (b) implementing measures to alleviate NH3 inhibition of rmax. Reduction in NH3 concentrations by reduced pH could be achieved by introduction of a rapidly acidifiable substrate, such as in the codigestion of fats, oils, and grease (FOG), or by the metered addition of a mineral acid to the digester. At a concentration of 2380 mg/L TAN-N, reducing the pH from 7.85 to 7.50 would cause a reduction of NH3eN concentration from 260 to 125 mg/L, and result in an approximate 25% increase in rmax. Additionally, loading of proteinaceous sludges can be reduced in order to achieve the combined effects of lower TAN concentrations and lower pH. The yield and concentration of soluble products such as TAN and bicarbonate alkalinity ðHCO 3 Þ are predictable based on sludge composition and organic loading rates, respectively (Wett et al., 2006). Since a widely cited advantage of THD is reactor volume reduction due to increased biologically and mechanically attainable solid loading rates, major reductions in feed solids as a method for lowering alkalinity and TAN concentration are undesirable. However, the nature of feed solids, whether proteinaceous or not, significantly affects yields of HCO 3 and TAN during anaerobic digestion. For example, methanogenic degradation of glycine theoretically yields 181 mg TAN-N and 650 mg alkalinity as CaCO3 per gram glycine degraded. Comparatively, methanogenic degradation of palmitic acid (HPa) theoretically yields 125 mg alkalinity as CaCO3 and consumes approximately 19 mg TAN-N per gram HPa (Rittmann and McCarty, 2001); see supplemental material for calculations. This previous analysis has broad implications for the operation of THD. It is clear that the THD process is not necessarily solids loading limited, but more specifically organic-nitrogen loading limited. This suggests that THD is amenable to codigestion of low-nitrogen wastes (e.g. FOG). Previous studies have reported volatile solids total Kjeldahl nitrogen fraction for FOG of 1.3% compared to 6e9% for combined primary and secondary sludge (Kabouris et al., 2008). While mixing and incorporation of FOG within an anaerobic digester poses a significant operational challenge for MAD, the fragmentation of lipids and conversion to VFA during thermal hydrolytic pretreatment (Wilson and Novak, 2009) is likely to mitigate lipid aggregation during digestion. Additionally, FOG exists as a high-solids liquid at mesophilic temperatures. By taking advantage of this distinction between a solids loading limited and an organic nitrogen loading limited process as revealed by the in-situ HAc saturation during THD, we propose that solids loading to and energy recovery from the THD process can be maximized. 3.4. Evidence of an alternate methanogenic pathway in THD Methanogenic inhibition factors ðINH3Þ related to the initial slope of cumulative biogas production curves from batch ammonia toxicity assays are presented in Fig. 4a. It is clear that the degree of methanogenic inhibition attributable to NH3 is less for THD than MAD at equivalent NH3 concentrations. The logistic model that was applied to these data exhibited good statistical fit. Coefficients of determination (R2) of LINH3 for THD and MAD were 0.947 and 0.951, respectively. Values of KI;NH3 for MAD and THD communities were calculated to be 250 mg/L and 505 mg/L of unionized ammonia as N, respectively. While ammonia toxicity assays were not conducted for MAD at NH3 concentrations greater than 150 mg/L, plotted ratios of LINH3 THDeLINH3 MAD over the concentration range of THD experiments show that overall methanogenesis during MAD is much more sensitive to ammonia inhibition than THD (Fig. 4b). This is especially true as NH3 levels are increased above the characteristic in-situ ammonia concentrations of MAD. These trends are distinct from previously discussed data depicting similar responses of the kinetics of HAc degradation by THD and MAD biomass over a range of NH3 concentrations. Acclimation of the methanogenic community may partially explain the relatively low sensitivity methane production by THD biomass to high NH3 concentrations. However, the contrasting responses of acetate utilization kinetics and methane production rate to increased NH3 levels suggests that the causeeeffect relationship between acetate degradation and methane production differs among THD and MAD reactors. In addition, methanogenic 16S rDNA clone library analyses performed on the THD and MAD reactor communities show a difference in the dominant methanogenic communities, favoring obligate hydrogenotrophic species in the THD biomass and favoring species capable of performing aceticlastic methanogenesis in the MAD reactor community (Wilson, 2010; Wett et al., 2012). Based on the differing sensitivities to ammonia exhibited by these biomass samples and the differences in microbial community, it is thus hypothesized that methanogenesis during THD is less dependent on ammonia-sensitive aceticlastic methanogenesis, and rather, non-aceticlastic methanogenesis from HAc is important for THD. w a t e r r e s e a r c h 4 6 ( 2 0 1 2 ) 6 2 4 7 e6 2 5 6 6253High NH3 concentrations as a result of thermophilic temperatures or high solids loading have been associated with a methanogenic pathway shift from aceticlastic to nonaceticlastic methanogenesis. Non-aceticlastic methanogenesis refers to HAc metabolism by a syntrophic co-culture containing a hydrogenotrophic methanogenic species and an organism capable of oxidizing acetate to CO2 plus H2 (Petersen and Ahring, 1991; Zinder and Koch, 1984). Examples of such previously identified organisms include the thermophilic Thermoacetegenium phaeum (Hattori et al., 2000) and mesophilic Clostridum ultunense (Schnurer et al., 1999), suggesting that non-aceticlastic methanogenesis is dependent on chemical rather than physical environmental conditions. Previous research has shown that the dominance of nonaceticlastic methanogenesis is related to the inhibition of aceticlastic methanogens resulting in an available ecological niche, rather than competitive advantage of syntrophic HAc oxidation. Aceticlastic methanogenesis is exergonic (DG00 ¼ 31.0 kJ/mol), thus proceeds in the absence of syntrophic associations. Acetate oxidation however, is highly endergonic (DG00 ¼ 104.6 kJ/mol) and requires consumption of reaction products, primarily gaseous hydrogen (similar to fatty acid oxidation), in order to proceed (Rothfuss and Conrad, 1993). It has been calculated elsewhere that given the following in-situ conditions for a methanogenic coculture with C. ultunense: 25 mM TAA, 100 mM HCO 3 , 31 kPa methane; the total energy release available for syntrophic acetate oxidation and hydrogenotrophic methanogenesis is approximately 17 kJ/mol (Ahring and Westermann, 1988; Hattori, 2008; Schink, 1997; Scholten and Conrad, 2000). Several studies have suggested that the minimum quantum of energy required to sustain microbial life corresponds approximately 1/3 the quantum of energy required for the production of 1 mol ATP (i.e. 70 kJ/mol ATP/3 protons/mol ATP ¼ 23 kJ/ proton; equal to the energy obtained from the translocation of one proton across a charged cellular membrane) (Ahring and Westermann, 1988; Hattori, 2008; Schink, 1997; Scholten and Conrad, 2000). Lower minimum free energy requirements for growth of hydrogenotrophic methanogens of approximately 10 kJ/mol have been observed in either stressed physiologic states (Jackson and McInerney, 2002) or in highly competitive environments (Hoehler et al., 2001), suggesting that the growth of a syntrophic acetate oxidizing coculture can be supported these low energy yields. In addition, hydrogenotrophic methanogens having high hydrogen affinity such as Methanoculleus sp. are most commonly associated with syntrophic acetate oxidation (Schnurer et al., 1999; Shigematsu et al., 2004), a characteristic that supports the low hydrogen concentrations that are required for the maintenance of acetate oxidation near its thermodynamic limit. Because acetate oxidation is not thermodynamically advantageous relative to aceticlastic methanogenesis at characteristic HAc levels of anaerobic digestion, it is generally believed that inhibition of aceticlastic methanogens is necessary to promote acetate oxidation as a dominant methanogenic pathway. In the context of THD and the kinetic data presented herein, it is clear that elevated NH3 concentrations impose meaningful inhibition of aceticlastic methanogenesis such that the alternate, less thermodynamically preferred metabolism may dominate. Past research has showed that hydrogenotrophic methanogenic species (including Methanobacterium bryantii, Methanobacterium formicicum, Methanobrevibacter smithii, and Methanobrevibacter arboriphilus) were less sensitive to ammonia toxicity than aceticlastic species (including Methanosarcina barkeri, and Methanosaeta concilii). The hydrogenotrophic species were found to be insensitive to TANN concentrations of 5600 mg/L (Sprott and Patel, 1986) and recovered more quickly from long-term incubation at TAN-N concentrations as high as 7700 mg/L (Hajarnis and Ranade, 1993, 1994). Other researchers have found a direct correlation between anaerobic digester TAN-N concentration, the absence of Methanosaeta sp. from the digester community, and the corresponding dominance of non-aceticlastic methanogenesis as the preferred degradation pathway for HAc (Karakashev et al., 2006). Such findings directly implicate inhibition of aceticlastic methanogenesis in an observed pathway shift to nonaceticlastic methanogenesis, and a relative tolerance to high NH3 concentrations by THD. A B Fig. 4 e (A) Inhibition factors ðINH3Þ related to the initial slope of cumulative methane production curves from batch ammonia toxicity assays. The rate of methane generation from the THD process is less affected by increased ammonia concentration, especially at high NH3 concentrations (B). This suggests that methanogenesis (from acetate) is less dependent on the activity of aceticlastic methanogens during THD (i.e. dominance of non-aceticlastic methanogenesis is apparent). 6254 w a t e r r e s e a r c h 4 6 ( 2 0 1 2 ) 6 2 4 7 e6 2 5 64. Conclusions This study provided kinetic parameter estimates for acetic acid utilization by two anaerobic digester communities, specifically looking at the impact of free ammonia inhibition on methanogenesis. The following conclusions can be drawn from this research: 1) The half saturation constant for acetic acid was calculated to be between 1.6 and 5.8 mg/L as unionized acetic acid for MAD and between 0.5 and 5.3 mg/L as unionized acetic acid for THD, depending on the concentration of unionized ammonia present. 2) The concentrations of unionized species of ammonia and acetic acid were better predictors of acetic acid kinetics (i.e. calculated biomass growth rate and half saturation constant) than total ammonia nitrogen and total acetic acid. 3) The calculated values of KS were found to vary as a function of rmax. This finding suggests that for these digester samples, the rate of acetic acid utilization is controlled by diffusion limited constant transport of bulk-phase substrate to the surface of a biological floc where uptake occurs. 4) Based on parameter estimates obtained through the batch ammonia toxicity assays and the operating conditions of the parent reactors, THD and MAD operate in distinctly different portions of their respective Monod curves. The MAD reactor operates in a portion of the Monod curve that is dictated by KS, meaning that hydrolysis rates, rather than methanogenesis, are the rate limiting metabolism in this digester. Conversely, THD operates in a portion of the Monod curve that is dictated by rmax, meaning that methanogenesis is likely the rate limiting metabolism in this digester. 5) Based on estimates of KI;NH3, the concentration of free ammonia that resulted in a 50% reduction in maximum methane production rate in THD (505 mg/L as NH3eN) was more than double that for MAD (250 mg/L as NH3eN). In combination with 16S rDNA clone library analyses of archeal communities of the parent reactors (described elsewhere), these findings suggest the presence of an alternative HAc degradation pathway in THD that is less sensitive to ammonia inhibition than aceticlastic methanogenesis which appears to prevalent in MAD. Appendix A. Supplementary material Supplementary data related to this article can be found online at http://dx.doi.org/10.1016/j.watres.2012.08.028. r e f e r e n c e s Ahring, B.K., Westermann, P., 1988. Product inhibition of butyrate metabolism by acetate and hydrogen in a thermophilic coculture. Applied and Environmental Microbiology 54 (10), 2393e2397. American Public Health Association, American Water Works Association and The Water Environment Federation, 2005. Standard Methods for the Examination of Water and Wastewater, twentyfirst ed. Andrews, J.F., 1978. A mathematical model for the continuous culture of microorganisms utilizing inhibitory substrates. Biotechnology and Bioengineering 10, 707e710. Angelidaki, I., Ahring, B.K., 1994. Anaerobic thermophilic digestion of manure at different ammonia loads e effect of temperature. Water Research 28 (3), 727e731. Bhattacharya, S.K., 1986. Toxic Substances in Methane Fermentation Systems; Fate and Effects on Process Kinetics. Ph.D dissertation, Drexel University, Department of Environmental Studies. Button, D.K., 1985. Kinetics of nutrient-limited transport and microbial-growth. Microbiology Reviews 49 (3), 270e297. CRC, 2009. Handbook of Chemistry and Physics, ninetieth ed. Eldem, N.O., Ozturk, I., Soyer, E., Calli, B., Akgiray, O., 2004. Ammonia and pH inhibition in anaerobic treatment of wastewaters, part I: experimental. Journal of Environmental Science and Health, Part A. Toxic/Hazardous Substances and Environmental Engineering 39 (9), 2405e2420. Fukuzaki, S., Nishio, N., Nagai, S., 1990a. Kinetics of the methanogenic fermentation of acetate. Applied and Environmental Microbiology 56 (10), 3158e3163. Fukuzaki, S., Nishio, N., Shobayashi, M., Nagai, S., 1990b. Inhibition of the fermentation of propionate to methane by hydrogen, acetate, and propionate. Applied and Environmental Microbiology 56 (3), 719e723. Hajarnis, S.R., Ranade, D.R., 1993. Revival of ammonia inhibited cultures of Methanobacterium-bryantii and Methanosarcinabarkeri. Journal of Fermentation and Bioengineering 76 (1), 70e72. Hajarnis, S.R., Ranade, D.R., 1994. Revival of Methanobacterium-formicicum after its inhibition by highconcentrations of ammonia. Letters in Applied Microbiology 18 (5), 254e256. Hattori, S., Kamagata, Y., Hanada, S., Shoun, H., 2000. Thermacetogenium phaeum gen. nov., sp nov., a strictly anaerobic, thermophilic, syntrophic acetate-oxidizing bacterium. International Journal of Systematic and Evolutionary Microbiology 50, 1601e1609. Hattori, S., 2008. Syntrophic acetate-oxidizing microbes in methanogenic environments. Microbiological Environment 23 (2), 118e127. Hoehler, T.M., Alperin, M.J., Albert, D.B., Martens, C.S., 2001. Apparent minimum free energy requirements for methanogenic Archaea and sulfate-reducing bacteria in an anoxic marine sediment. FEMS Microbiology Ecology 38 (1), 33e41. Iranpour, R., Cox, H., 2006. Retrospective and perspectives of thermophilic anaerobic digestion: part II. Water Environment Research 78 (3), 219e220. Jackson, B.E., McInerney, M.J., 2002. Anaerobic microbial metabolism can proceed close to thermodynamic limits. Nature 415 (6870), 454e456. Kabouris, J.C., Tezel, U., Pavlostathis, S.G., Engelmann, M., Todd, A.C., Gillette, R.A., 2008. The anaerobic biodegradability of municipal sludge and fat, oil, and grease at mesophilic conditions. Water Environment Research 80 (3), 212e221. Karakashev, D., Batstone, D.J., Angelidaki, I., 2005. Influence of environmental conditions on methanogenic compositions in anaerobic biogas reactors. Applied and Environmental Microbiology 71 (1), 331e338. Karakashev, D., Batstone, D.J., Trably, E., Angelidaki, I., 2006. Acetate oxidation is the dominant methanogenic pathway w a t e r r e s e a r c h 4 6 ( 2 0 1 2 ) 6 2 4 7 e6 2 5 6 6255from acetate in the absence of Methanosaetaceae. Applied and Environmental Microbiology 72 (7), 5138e5141. Kepp, U., Machenbach, I., Weisz, N., Solheim, O.E., 2000. Enhanced stabilisation of sewage sludge through thermal hydrolysis e three years of experience with full scale plant. Water Science and Technology 42 (9), 89e96. Kepp, U., Panter, K. and Solheim, O.E., 2001. High dry solids digestion. CIWEM/Aqua Enviro 6th European Biosolids and Organic Residiuals Conference, UK. Kugelman, I.J., Chin, K.K., 1971. In: Pohland, F.G. (Ed.), Anaerobic Biological Treatment Processes. American Chemical Society, Washington, D.C, pp. 55e90. Laanbroek, H.J., Gerards, S., 1993. Competition for limiting amounts of oxygen between Nitrosomonas-europaea and Nitrobacter-winogradskyi grown in mixed continuous cultures. Archives of Microbiology 159 (5), 453e459. Lawrence, W.D., McCarty, P.L., 1969. Kinetics of methane fermentation in anaerobic treatment. Journal of the Water Pollution Control Federation 41 (2), R1eR17. Liu, T., Sung, S., 2002. Ammonia inhibition on thermophilic aceticlastic methanogens. Water Science and Technology 45 (10), 113e120. Novak, J., 1974. Temperature-substrate interactions in biological treatment. Journal of the Water Pollution Control Federation 46 (8), 1984e1994. O’Rourke, J.T., 1968. Kinetics of Anaerobic Treatment at Reduced Temperatures. Stanford University. Petersen, S.P., Ahring, B.K., 1991. Acetate oxidation in a thermophilic anaerobic sewage-sludge digester e the importance of non-aceticlastic methanogenesis from acetate. FEMS Microbiology Ecology 86 (2), 149e157. Potts, L.G.A., Jolly, M., 2004. Controlling and monitoring anaerobic digesters fed with thermally hydrolysed sludge. Water and Environment Journal 18 (2), 68e72. Rittmann, B.E., McCarty, P.L., 2001. Environmental Biochemistry. Principles and Applications McGraw-Hill, New York, NY. Rothfuss, F., Conrad, R., 1993. Thermodynamics of methanogenic intermediary metabolism in Littoral sediment of lake Constance. FEMS Microbiology Ecology 12 (4), 265e276. Schink, B., 1997. Energetics of syntrophic cooperation in methanogenic degradation. Microbiology and Molecular Biology Reviews 61 (2), 262e280. Schnurer, A., Zellner, G., Svensson, B.H., 1999. Mesophilic syntrophic acetate oxidation during methane formation in biogas reactors. FEMS Microbiology Ecology 29 (3), 249e261. Scholten, J.C.M., Conrad, R., 2000. Energetics of syntrophic propionate oxidation in defined batch and chemostat cocultures. Applied and Environmental Microbiology 66 (7), 2934e2942. Shigematsu, T., Tang, Y.Q., Kobayashi, T., Kawaguchi, H., Morimura, S., Kida, K., 2004. Effect of dilution rate on metabolic pathway shift between aceticlastic and nonaceticlastic methanogenesis in chemostat cultivation. Applied and Environmental Microbiology 70 (7), 4048e4052. Smith, L.H., Kitanidis, P.K., McCarty, P.L., 1997. Numerical modeling and uncertainties in rate coefficients for methane utilization and TCE cometabolism by a methane-oxidizing mixed culture. Biotechnology and Bioengineering 53 (3), 320e331. Smith, L.H., McCarty, P.L., Kitanidis, P.K., 1998. Spreadsheet method for evaluation of biochemical reaction rate coefficients and their uncertainties by weighted nonlinear least-squares analysis of the integrated monod equation. Applied and Environmental Microbiology 64 (6), 2044e2050. Sprott, G.D., Patel, G.B., 1986. Ammonia toxicity in pure cultures of methanogenic bacteria. Systematic and Applied Microbiology 7 (2e3), 358e363. Sung, S.W., Liu, T., 2003. Ammonia inhibition on thermophilic anaerobic digestion. Chemosphere 53 (1), 43e52. VanLier, J.B., Martin, J.L.S., Lettinga, G., 1996. Effect of temperature on the anaerobic thermophilic conversion of volatile fatty acids by dispersed and granular sludge. Water Research 30 (1), 199e207. Wett, B., Eladawy, A., Ogurek, M., 2006. Description of nitrogen incorporation and release in ADM1. Water Science and Technology 54 (4), 67e76. Wett, B., Murthy, S.N., Taka ´cs, I., Wilson, C.A., Novak, J.T., Panter, K. and Bailey, W., 2009. Simulation of Thermal Hydrolysis at the Blue Plains AWT: A New Toolkit Developed for Full-Plant Process Design. Water Environment Federation 82nd Annual Technical Exhibition and Conference, Orlando. Wett, B., Taka ´cs, I., Bastone, D., Wilson, C.A., Murthy, S.N., 2012. Anaerobic model for high-loaded or high-temperature digestion e additional pathway of acetate oxidation. 3rd IWA/ WEF Wastewater Treatment Modeling Seminar, Mont-SainteAnne, Quebec, Canada. Wilson, C.A., Novak, J.T., 2009. Hydrolysis of macromolecular components of primary and secondary wastewater sludge by thermal hydrolytic pretreatment. Water Research 43 (18), 4489e4498. Wilson, C.A., 2010. Mechanisms of Methanogenic Inhibition in Advanced Anaerobic Digestion. Virginia Polytechnic institute and State University. Wilson, C.A., Tanneru, C.T., Banjade, S., Murthy, S.N., Novak, J.T., 2011. Anaerobic digestion of raw and thermally hydrolyzed sludge under various operational conditions. Water Environment Research 83 (9), 815e825. Winne, D., 1973. Unstirred layer source of biased Michaelis constants in membrane transport. Biochimica et Biophysica Acta 298, 27e31. Zinder, S.H., Koch, M., 1984. Non-Aceticlastic methanogenesis from acetate - Acetate oxidation by a thermophilic syntrophic coculture. Archives of Microbiology 138 (3), 263e272. 6256 w a t e r r e s e a r c h 4 6 ( 2 0 1 2 ) 6 2 4 7 e6 2 5 6

Комментировать