Comparison of the performance of three different reactors for BioHydrogen production via dark anaerobic fermentations

Proceedings International Hydrogen Energy Congress and Exhibition IHEC 2005
Istanbul, Turkey, 13-15 July 2005
Comparison of the performance of three different reactors for BioHydrogen production via dark anaerobic fermentations
Marcello Camilli* and Paola M. Pedroni&
* EniTecnologie, Biological Science Department, Via E. Ramarini, Monterotondo, Rome, Italy (mcamilli@enitecnologie.eni.it)
& EniTecnologie, Environmental Technologies Research Center, Via F. Maritano, San Donato Milanese, Milan, Italy (ppedroni@enitecnologie.eni.it)
ABSTRACT
Biohydrogen is the microbial production of hydrogen (H2) from water or organic substrates in the presence or absence of light. Among biological H2 production processes, dark anaerobic fermentations using wastes rich in starch and sugars have potential for practical application provided that H2 yields, typically 10% to 20% of stoichiometric, are improved and the range of suitable substrates extended to other wastes and biomass sources. EniTecnologie, the R&D arm of the Italian oil and gas company Eni, has been involved in biological H2 production for over a decade. Our present interest in the field is to evaluate the feasibility of renewable hydrogen production processes in the dark by anaerobic heterotrophic bacteria grown on carbohydrate-rich substrates, with Volatile Fatty Acids (VFA) as fermentation by-products. After the screening and selection on model substrates of inocula optimised for H2 production, both from pure cultures and microbial consortia, current experimental activities deal with the comparison of the H2 evolution performance in three different bioreactors operated at 35°C: a Continuously Stirred Tank Reactor (CSTR), an Upflow Fixed Bed Reactor (UFBR) and an Upflow Anaerobic Sludge Blanket (UASB) reactor. All three cultivation systems were inoculated with a mixed spore-forming microflora from a thermally pre-treated anaerobic methanogenic sludge, fed with a synthetic medium simulating the fermentable carbohydrates from agricultural wastes and operated continuously by increasing the dilution rate up to a minimal hydraulic retention time (HRT).
With the UFBR, the best performance in terms of H2 yield (0.23 L H2/g carbohydrdate) and H2 production rate (2.54 mmol H2/L reactor, h) was obtained at HRT 37.6 h and 20.5 h, respectively. The system failed at HRT 17 h. With the UASB run at HRT 24.6 h, production yield equal to 0.15 L H2/g carbohydrate was reached, corresponding to a mean microbial aggregate concentration of 1.30 g VSS/L. Its best H2 production rate (4.76 mmol H2/L reactor, h) was achieved at HRT 6.7 h and washout of aggregates was observed at HRTs below 5.8 h. The CSTR showed both optimal H2 yield (0.30 L H2/g carbohydrate) and production rate (4.50 mmol H2/L reactor, h) when operated at HRT 32.9 h. The corresponding amount of H2–producing microbial biomass retained was around 0.95 g VSS/L and the system failed at HRT 21 h. In all reactors the H2 content in the gas-phase ranged from 47% to 68%, carbohydrates conversion exceeded 95% and the composition of VFA, both qualitative and quantitative, was according to the substrate degradation. The best performance of CSTR was probably due to the mechanical stirring that promoted both H2 and CO2 removal from the fermentation broth, therefore reducing feed-back inhibition phenomena.
Keywords: biohydrogen; dark fermentations; bioreactor design (CSTR; UASB; Fixed Bed)
1. INTRODUCTION
H2 will be a key component of the clean and sustainable energy system of the future. The development of a carbon-free economy is mainly driven by concerns about the global climate change, due to the greenhouse gas effect, the mismatch between the fossil fuel-based energy production and the increase in energy demand and the depletion of fossil fuel reserves. To foster
1
Camilli and Pedroni
the widespread use of H2 as an energy carrier, the goal is to develop production technologies that are economically feasible, environmentally friendly and possibly renewable.
H2 from biomass and organic wastes is one potentially economical and renewable source of H2, with both thermal and biological conversion processes providing plausible approaches. The latter, including photobiological processes and dark fermentations, are more environmentally friendly and less energy intensive than thermochemical and electrochemical options.
Basic and applied research for the biological production of H2 have been carried out for over 30 years (Benemann 1996 and 1998; Nandi and Sengupta 1998; Das and Veziroglu 2001; Zaborsky 1998; Miyake et al., 2001 and 2004). Early activities were mainly focused on light-driven processes, such as biophotolysis of water (direct or indirect) with microalgae and cyanobacteria and photofermentations of organic acids with photosynthetic bacteria (Asada and Miyake, 1999; Zhu et al., 1999). More recently, the interest has shifted to dark anaerobic fermentations by strict and facultative anaerobes (mesophilic and thermophilic) that combine H2 production with the disposal of organic wastes rich in starch and sugars (Hallenbeck and Benemann 2002).
The maximal theoretical yield for these fermentative processes operating in the absence of light is calculated to be 4 mol of H2 per mole of glucose (33% of stoichiometric), but in practice only 10%-20% of the glucose energy content is recoverable as H2 because of the limited metabolic energy derived from H2 fermentations. Several approaches proved to drive the reaction against the metabolic barriers and to increase H2 yields: operation at non-standard conditions (at higher temperatures, lower pH or lower H2 partial pressure), metabolically engineering the H2 evolution pathway and competing reactions and decoupling ATP generation from H2 evolution to reduce the energy requirements (Claassen et al., 2000; Woodward et al., 1996; Kumar et al., 2001). Key advantages of dark anaerobic fermentations for H2 production are the high production rates, the efficient catalyst (hydrogenase) that requires no direct ATP supply, the integration with the disposal of organic wastes and the lower costs of the fermentation technology, widely used commercially in the production of ethanol fuel from starch and in the conversion of organic waste to methane. The economics and the sustainability of these processes can be further improved by combining H2 fermentations with the production of higher value co-products, such as organic acids and bioplastics and the treatment of a wider range of wastes, agricultural, industrial and municipal (Hawkes et al., 2002).
Our company has been involved in biological H2 production for over a decade, starting with a research project promoted by the Japanese MITI (Ministry of International Trade and Industry) and carried out with industrial partners such as Kubota, Kajima and IHI. The objectives were the development of a biohydrogen production process combined with the disposal of food wastes using photosynthetic bacteria and the optimization of their H2 evolution performance by genetic engineering techniques (Fascetti et al., 1998; Franchi et al., 2004). Our present interest in the field is to evaluate the feasibility of H2 production processes based on dark anaerobic fermentations using carbohydrate-rich wastes as growth substrates. In these fermentative conversions H2 is produced by heterotrophic anaerobic and acidogenic bacteria producing respectively Volatile Fatty Acids (VFA), alcohols and CO2 as soluble and gaseous by-products (Hallenbeck and Benemann, 2002).
Our research activities were firstly focused on the screening and selection of mesophilic inocula optimised for H2 production, both from pure cultures and environmental samples, using synthetic media as model substrates. Biological H2 production via dark fermentations, operated under mesophilic or thermophilic conditions, traditionally uses pure cultures of strict and facultative anaerobes, such as Clostridia, Enterobacteriaciae and Thermatogales, for inocula preparation. More recently, microbial consortia such as acclimated sewage sludges or anaerobic digested sludges, after proper adaptation, were also employed as H2 producers (Lay, 2000; Lee et al., 2004). Our current experiments deal with process design and operations, on lab-scale, of single-stage fermentations using synthetic media and actual wastes. In particular, we are comparing the H2 2
Camilli and Pedroni.
evolution performance in three different bioreactors: a Continuously Stirred Tank Reactor (CSTR), an Upflow Fixed Bed Reactor (UFBR) and an Upflow Anaerobic Sludge Blanket (UASB) reactor. Immobilized cells represent a common alternative to suspended cell systems in continuously operated fermentors since they are more efficient in both solid/liquid separation and H2 production rates. In fact they can be operated at high dilution rates (low HRTs) without washout of bacterial cells (Chang et al., 2002).
All three cultivation systems were inoculated with a thermally pre-treated (100°C for 20 min.) anaerobic methanogenic sludge, continuously fed with a synthetic medium simulating the fermentable carbohydrates from agricultural wastes (10 g/L of soluble starch/xylose 1/1w) and operated at 35°C by increasing the dilution rate to reach a minimal HRT that still allowed the activity and stability of each cultivation system while maximising the substrate loading rate.
2. MATERIAL AND METHODS
2.1 Hydrogen-producing Inoculum
The inoculum was obtained from an experimental anaerobic digester producing methane from sewage sludge and pre-treated via acid thermo-hydrolysis (pH 1.0; 160°C) (Pappa et al., 2002).
The H2-producing capacity of this mixed microbial biomass was improved by a thermal pre-treatment (100°C for 20 min.) that enriched the spore-forming microflora (selectively the clostridial component) and displaced the H2-utilizing (oxidizing) bacteria such as methanogens, characterized by their thermo-sensitivity (Lay, 2000 ). Before use in the bioreactors, this inoculum was characterised as follows: spore density 109/mL; Total Suspended Solids (TSS) and Volatile Suspended solids (VSS) 51.8 g/L and 16.3 g/L, respectively. The ratio of the inoculum at the start of each fermentation was 2% of the operative liquid volume of the bioreactors.
2.2 Growth Medium
The synthetic medium used as a growth substrate in the fermentative H2 production tests carried out in the three bioreactors was prepared according to the following composition (g/L): soluble starch 5.0, xylose 5.0, ammonium nitrate 0.285 or 0.340, sodium thioglycolate 0.5, KH2PO4 0.25 (total COD ~ 10 g/L). These chemicals were added (final pH 7.0) to an Owen-defined solution (Owen et al. 1979) before sterilization (121°C x 30 min.). The resulting C/N ratio suitably ranged from 33 to 40 (Lin, 2004), while P concentration was not limiting for the bacterial growth (C/N/P from 100/2.5/1.4 to 100/3.0/1.4). The carbohydrate composition employed in the medium simulates the fermentable sugars resulting from the hydrolysis of agricultural wastes and the selected concentration derives from preliminary experiments in batch aimed to assess carbohydrates/hydrogen conversion yield (Lucarelli, 2004).
2.3 Design of the Bioreactors
a) Continuously Stirred Tank Reactor (CSTR)
Figure 1 schematically describes the lab-scale CSTR apparatus. It consists of a cylindrical glass vessel with a working volume of 1.50 L, an internal diameter of 10 cm, a height of 22 cm. It is equipped with a temperature control (T was kept at 35°C) and a mechanical stirrer working at around 500-600 rpm and assuring a complete mix condition (New Multigen system, New Brunswick Scientific CO. Inc. N.J., USA). The pH was controlled by a pH-stat system (CLAIND, Como-Italy). If requested, a sodium hydroxide (NaOH 2N) flow, provided by a peristaltic pump (Gilson Minipuls 2), was automatically combined with influent feed flow to adjust the pH on the pre-imposed value of 5.60. The amount of biogas produced was recorded daily using a wet gas meter (TRITON-WR C model 181) acting also as water seal. A biogas sampling port was installed between the meter and the reactor to allow direct gas sampling by a syringe for gas-chromatographic analyses. The substrate reservoir was kept under N2 atmosphere and both the influent and the effluent reactor flows were regulated by the same peristaltic pump (Gilson 3
Camilli and Pedroni
Minipuls 2 multi-channel). Tubing, adapters and gaskets were made of 100% fluorine-elastomer material (Iso-Versinic) to avoid H2 loss. LiquideffluentLiquideffluent
Figure 1 – Schematic of CSTR system
b) Upflow Fixed Bed Reactor (UFBR)
By utilising the same ancillary accessories, the CSTR vessel was substituted by a glass column reactor (working volume 0.95 L) that includes a water jacket (kept at 35°C) and was packed with pumice stones (mean size diameter < 2.0 cm; bed height 35 cm; bed porosity or void bed fraction 65-70% ) (Figure 2). Sampling ports were located along the 10 cm height interval (the highest connected to a 0.02 bar water seal and acting as a liquid effluent discharge) and an additional pump (Watson Marlow mod. 501 U) assured the recycling of the internal fermentation broth at a flow rate of 8.0 L/h. water jacketsampleportspumicestonesALKALIPUMPFEEDINGPUMPupflowinfluentBiogasoutpressurizerFTrecyclegasLiquideffluentpHCLiquidand gaswater gas
Figure 2 – Schematic of UFBR system
c) Upflow Anaerobic Sludge Blanket (UASB) reactor
A copy of the glass column assembled for the Fixed Bed was used for the design of the UASB reactor (working volume 1.20 L). In this case, a gas/liquid/solid separator device was set up in the upper zone of the liquid volume (Figure 3). The above-mentioned recycling flow rate and the mesophilic conditions (35°C) were applied for the UASB reactor as well.
4
Camilli and Pedroni.
blanketzonewater jacketALKALIPUMPFEEDINGPUMPupflowinfluentBiogasoutpressurizerFTrecyclegasLiquideffluentpHCgas/solid/liquidseparatorLiquidand gasblanketzonewater gas
Figure 3 – Schematic of UASB system
2.4 Start-up and Operations of the Bioreactors
At the start of the H2 fermentation tests, each system was managed in a fed-batch mode by changing 1/3 of the working volume of the exhaust substrate with fresh medium. This replacement has been carried out for 10 days, whenever the residual carbohydrate concentration was around 1.0 g/L. For the UFBR, fed-batch operations were prolonged for three weeks in order to facilitate the initial bacterial adhesion to the pumice carrier.
During H2 fermentations operated in continuous, in all bioreactors HRTs were reduced stepwise, starting from an average of 70 h down to a specific retention time that corresponded, for each system, to the loss of equilibrium and functionality. At each different HRT, the bioreactors were operated for a given time period sufficient to ensure reproducible steady-state conditions that were established when the volume and composition of the produced biogas were constant. After the collection of the analytical parameters, the retention time was shortened (dilution rate increased).
2.5 Monitoring the Performance of the Bioreactors
The bioreactors were monitored at each steady-state condition (i.e. every stable HRT) by analysing the concentration, in the liquid effluent, of: Total Carbohydrates, COD (discontinuously), VFA, TSS and VSS. Bacterial biomass content was determined as Mixed Liquor Volatile Suspended Solids (MLVSS) on samples of the fermentation broth, periodically taken from the sampling ports along the column of the bioreactors or, in the case of CSTR, from the head ports. The composition of every new batch of influent growth substrate (10 L reservoir capacity) was analysed for the same parameters as above. Gas production and composition were evaluated on a daily basis. Mean H2 volumes for each HRT were calibrated to 25°C and 760 mmHg and the end results expressed as H2 yield (L H2/g carbohydrdate converted) and H2 production rate (mmol H2/L reactor, h).
2.6 Analytical Parameters and Methods
Analytical parameters such as pH, Eh, TSS, VSS and COD were determined according to standard procedures (APHA, AWWA, WEF, 1992). The gas composition was determined using a gas cromatograph (Agilent Technologies 6890N) equipped with a thermal conductivity detector (TCD) 280°C, a capillary column GS-Gaspro J&W Scientific (15 m x 0.32 mm id), oven 30°C isotherm, injection temperature 180°C, carrier gas N2. VFA composition was quantified by the same instrument, with a flame ionization detector (FID) 230°C, capillary column Nukol TM Supelco (30 m x 0.25 mm id, 0.25 μm film thickness), oven 100°C to 200°C (5°C/min.), injection temperature 210°C, carrier gas N2. For the GC analysis of VFA, samples of the fermentation broth were first acidified (pH 1.0) with H2SO4 (10% vol) and then extracted with ethyl ether (2/1 vol/vol). Total carbohydrates concentration was determined colorimetrically both in the influent and effluent liquids by phenol-sulphuric acid method (Anonymous, 1956).
5
Camilli and Pedroni
3. RESULTS AND DISCUSSION
Table 1 summarizes the experimental data of the fermentation tests carried out in the three bioreactors operated at different HRTs and obtained under steady state conditions. The carbohydrate Organic Loading Rates (OLRs, g/L, d) are reported as a mean of values of each reactor for a specific range of HRT.
Table 1. Data under steady state conditions CSTRUASBUFBR HRT OLRVSSCarbohy.degr.OLRVSSCarbohy.degr.OLR VSSCarbohy.degr. (h)(g/L d)(g/L)(%)(g/L d)(g/L)(%)(g/L d)(g/L)(%)69.7 + 3.73.381.9992.903.330.7098.853.470.7297.4261.7 + 2.53.741.9591.76ndndnd3.920.7297.0652.2 + 0.44.551.9789.804.511.1198.634.451.0997.9444.8 + 1.75.371.2088.955.311.2098.055.291.1187.8835.1 + 2.97.070.9591.07ndndnd7.431.0388.3527.7 + 0.99.020.9496.49ndndnd9.601.0588.9522.7 + 1.510.541.0984.0710.031.3087.0211.111.0587.7617.4 + 0.813.11washout66.6013.921.5784.7213.430.8562.8213.1 + 0.918.851.1382.8918.20washout56.007.4 + 0.733.820.9961.605.8 + 0.142.781.0356.625.246.575washout31.20
The UASB reactor turned out to be manageable at the shortest treatment time (HRT 5.8 h, corresponding to an OLR of 42.78 g carbohydrates/L, d), whereas the CSTR and the UFBR showed a remarkable biomass washout and a decay in equilibrium and performance at HRTs around 21 h (OLR 11.11 g/L, d) and 17 h (OLR 11.24 g/L, d), respectively. In all bioreactors the redox potential, after some initial fluctuations during the start-up phase, resulted stable and ranged from -300 to -400 mV. During the continuous operations, carbohydrate degradation decreased from 93% to 84% in the CSTR, from 99% to 56.5% in the UASB and from 97.5% to 63% in the UFBR. A partial increase of the substrate degradation (from 89% to 96.5%) occurred only in the CSTR at HRTs from 35.1 h to 22.7 h. This was probably due to a change of the bacterial dominant morphotypes endowed with a different specific metabolism. A microbial shift from long-chain to cluster rod-shaped forms was indeed observed by optical microscope analysis in the bacterial suspension of the CSTR.
In the CSTR, the microbial biomass concentration regularly decreased from an average of 2.0 gVSS/L, at the beginning of the test, to 1.0 gVSS/L at lower HRTs. A different performance was shown both by the UASB aggregates and the suspended flocs in the UFBR. In the UASB, a progressive increase occurred from 0.7 gVSS/L to a maximum of 1.6 gVSS/L at HRT of 18 h with a final value of 1.0 gVSS/L before biomass washout at HRT of 5.2 h. In the UFBR, an increase was recorded from 0.7 gVSS/L to an average value of 1.0 gVSS/L that was stably maintained until HRT below 17 h, at which the performance of the bioreactor failed.
For the CSTR, according to the equation HRT = S (microbial cells) RT, typical of the system configuration, the decrease in biomass concentration was directly related to the decrease of the treatment time.
In the case of the UFBR, the selected packing material (porous pumice) turned out not to be efficiently colonized at the end of the fermentation test by the microbial consortium used as an inoculum. Taking into account that the irreversible microbial attachment to a suitable carrier increases with OLR and it is directly related to the consequent increase in bulk liquid VSS concentration (Shapiro and Switzenbaum, 1984), our data are in disagreement with the correlation of the parameters mentioned. It is likely that most of the H2 was produced by suspended microorganisms, rather than by those adhering to the support and that the pumice might even have exerted a detrimental effect on the adhesion mechanisms or on the physical release of the biogas produced. Additional experiments using different carrier materials for bacterial immobilization
6
Camilli and Pedroni.
(i.e. activated carbon) are planned in the future. In the UASB reactor, microbial aggregates (granulation) appeared after operating the system for 60 days (carbohydrate OLR > 14.0 g/L, d; up-flow velocity > 0.41 m/h; HRT below 17 h). At that time, small (0.5-2.0 mm) aggregates were visible as a dimensional gradient along the height of the UASB column. The biomass, in the form of aggregates or suspended flocs, grew slowly, but was retained longer if short HRTs are compared (see Table 1, VSS data in UASB and UFBR vs. CSTR).
In all bioreactors only two peaks, corresponding to H2 and CO2, were present in the TCD-GC profiles of the gas-phase and the specific H2 content ranged from 47% to 68 %.
In Figures 4 and 5, H2 yields and H2 production rates for the three bioreactors are reported as a function of the HRTs actually operated in each specific case. As shown in Figure 4, for the CSTR H2 yields peaked with 0.30 L H2/g carbohydrate at HRT of 32.9 h, while for the UFBR and the UASB reactor with 0.23 L H2/g carbohydrate at HRT of 37.6 h and 0.15 L H2/g carbohydrate at HRT of 24.6 h, respectively. Figure 4. Reactor Hydrogen Yield CSTR0.0000.0500.1000.1500.2000.2500.3000.3500.010.020.030.040.050.060.070.080.0L H2 /
g
carbohydrate UASB0.0000.0500.1000.1500.2000.010.020.030.040.050.060.070.080.0L H2/ g
carbohydrate UFBR0.0000.0500.1000.1500.2000.2500.010.020.030.040.050.060.070.080.0 HRT (h)L H2 / g carbohydrateFigure 5. Hydrogen production rateCSTR 0.001.002.003.004.005.000.010.020.030.040.050.060.070.080.0mmol H2 /
L
react. UASB0.001.002.003.004.005.006.000.010.020.030.040.050.060.070.080.0mmol H2 /
L
react. hUFBR0.0001.0002.0003.0000.010.020.030.040.050.060.070.080.0HRT (h)mmol H2/L react. hFigure hFigure h
In the UASB reactor, a maximal H2 production rate of 4.76 mmol H2/L reactor, h (HRT 6.7 h) was reached, while in the CSTR and in the UFBR peak values were 4.50 mmol H2/L reactor, h (HRT 32.9 h) and 2.54 mmol H2/L reactor, h (HRT 20.5 h), respectively (Figure 5).
The data indicate that HRTs are related to H2 yields somewhat differently compared to production rates. In fact, while in the case of the CSTR both optimal H2 yield (the best value among all bioreactors) and production rate are achieved operating at the same treatment time (HRT 32.9 h), in the UASB and UFBR, H2 production rates were maximal when HRTs near to the lower limits of tolerance were applied. In these two bioreactors, however, H2 yields were lower compared to those of the CSTR. The concentration and the physical status of the biomass (aggregates, flocs and attached bacteria), that may affect the substrate availability for the inner cells of the microbial clusters and the biofilm and their consequent different metabolism, might have played a significant role and explain the different performance of the bioreactors.
7
Camilli and Pedroni
In the CSTR, the HRT-coincident best performance, in terms of H2 yield and productivity, was probably due to the mechanical stirring that promoted both H2 and CO2 removal from the fermentation broth, therefore reducing feed-back inhibition phenomena. In both UASB and UFBR, the higher biogas generation rates and the hydrodynamic shear forces, typical of high volumetric loading rates (short HRTs), led to a similar mixing effect of the cultures.
In summary, these results point out that H2 can be generated in a short period of time (fraction of day) by selecting the UASB configuration, the best performer in productivity with 4.76 mmol H2/L reactor, h at an HRT of 6.7 h.
In future experiments, the three bioreactors will be operated under vacuum (de-pressurized) conditions, in order to facilitate the release of dissolved H2 from the liquid phase of the systems.
Table 2 shows the composition, qualitative and quantitative, of VFAs analyzed in the bioreactor effluents during the operations in continuous mode. Traces of iso-butyric and valeric acids were also sometimes detected.
Table 2. VFA concentration under steady state conditions CSTRUASBUFBRHRT Acetic PropionicButyricCaproicAceticPropionic ButyricAcetic Propion Butyric (h)(mg/L)(mg/L)(mg/L)(mg/L)(mg/L)(mg/L)(mg/L)(mg/L)(mg/L)(mg/L)69.7 + 3.74434285565ndndndnd88828367661.7 + 2.526781121674ndndndnd84937263952.2 + 0.41248nd2408126258595256596023267044.8 + 1.7964nd29601512480882555167996243235.1 + 2.9890nd3280185ndndnd1838174202927.7 + 0.9965nd33252541035962338ndndnd22.7 + 1.51250nd303920912737723621600106212917.4 + 0.8washout9191132684700120188513.1 + 0.9ndndndwashout7.4 + 0.77103422055.8 + 0.16902020055.2washout
Since H2 yield is reduced to 50% when butyrate, instead of acetate, is the main fermentation end product of glucose (or of its isomers hexoses or of its polymers) (Hawkes et al., 2002), as shown in the following equations:
C6H12O6 + 2H2O → 2CH3COOH + 4H2 + 2CO2 (0.50 LH2/g carbohydrate) (eq. 1)
C6H12O6 → CH3CH2CH2COOH + 2H2 + 2CO2 (0.25 LH2/g carbohydrate) (eq. 2)
the same homo-acidic transformations can be assumed for xylose (pentose):
2C5H10O5 + 4H2O → 3CH3COOH +8H2 + 4CO2 (0.60 LH2/g carbohydrate) (eq. 3)
6C5H10O5 → 5CH3CH2CH2COOH +10H2 + 10CO2 (0.30 LH2/g carbohydrate) (eq. 4)
In the case of the mixed substrate we employed as a growth medium (starch/xylose 10 g/L, 1/1 w), the following theoretical conversions can be assumed:
1.2C5H10O5 + C6H12O6 + 4H2O → 4CH3COOH + 8H2 + 4CO2 (eq. 5)
1.2C5H10O5 + C6H12O6 → 2CH3CH2CH2COOH + 4H2 + 4CO2 (eq. 6)
where, with respect to the eq. 5, H2 yield of eq. 6 is 50% (0.25 LH2/g carbohydrate).
8
Camilli and Pedroni.
For each HRT and for each bioreactor, the quantitative composition of VFAs (Table 2) was elaborated according to the above equations 5 and 6. The results were then verified applying the equation 7 below and they resulted in agreement with carbohydrate degradation values (Table 1).
1.0 g substrate converted → max. 0.67 g acetate + 0.49 g butyrate (eq.7)
COD quantification in the bioreactor effluents was occasionally carried out, confirming the supposed theoretical maximum of 33% COD reduction (4 moles CO2/12 moles C), as expressed by both eq. 5 and 6. The experimental values of soluble COD reduction, in fact, were around 23.9% in all bioreactors where a hetero-acidic conversion of carbohydrates (acetate, butyrate and other VFAs, such as propionate) occurred (Table 2).
In addition, due to COD reductions below the value of 33%, a bacterial production of alcohols, such as ethanol, propanol or butanol, competing and parallel to VFAs generation, is probably present.
In the CSTR the concentration of acetate decreased (from 84% to 22%) and that of butyrate progressively increased (from 10% to 73%) as HRT decreased (and OLR increased). Propionate was detected operating at initial HRTs, while caproate was present and increased in parallel with OLRs.
The UASB performed differently: equal amounts of acetate and butyrate were present until a HRT of 27.7 h, then a dominance of butyrate (up to 72%) was observed. Caproate was not detected, whereas propionate was always present in low amount.
Finally in the UFBR butyrate was always predominant (up to 80%) until an HRT of 44.8 h. Then the production of acetate increased to a maximal value of 45%. Caproate was not detected and propionate was always produced in small amount.
In conclusion, the results reported in Figures 4, 5 and in Table 2 show that a dominant and efficiently controlled H2/acetate/(mainly)butyrate fermentation pathway has been induced and maintained in all bioreactors tested using anaerobic H2-producing spore formers as an inoculum. Clostridium species were probably the dominant microorganisms of the consortium as they are involved in butyrate fermentation (Nandi and Sengupta, 1998).
The comparison of our experimental data with those reported in the literature for similar fermentation processes shows that the maximal H2 yield we obtained with the CSTR is in the range of the best value reported for the same cultivation system (0.28 LH2/g carbohydrate) inoculated with an anaerobic digested sludge, using soluble starch (20 g/L) as a growth substrate, operating at 30°C, 600 rpm and HRT of 17 h. (Lay, 2000). With UASB reactors, a higher H2 production rate (11.2 mmol H2/L reactor, h) was reported using a single model substrate, sucrose (20 g/L), a heat-treated (100°C x 45 min.) sewage sludge as an inoculum and running the bioreactor at HRT of 8 h (Chang and Lin, 2004). Finally in UFBR operated under mesophilic conditions, an optimal H2 production rate of 18.5 mmol H2/L reactor, h at HRT of 2 h was achieved, always using sucrose (20g/L) as a single substrate, activated carbon as bacterial carrier and acidified (pH 3.0) activated sludge as an inoculum. The use of other supports, such as loofah sponge and expanded clay, proved to perform less efficiently (Chang et al., 2002).
4. CONCLUSIONS
Effective H2 production via dark anaerobic fermentation of starch/xylose was carried out using pH-controlled CSTR, UASB and UFBR (pumice packed) bioreactors operated under mesophilic conditions and inoculated with a thermally pre-treated anaerobic methanogenic consortium. The enriched community of H2-producing spore formers showed metabolic stability (mainly H2/butyrate pathway) during the entire duration of continuous operation.
9
Camilli and Pedroni
The pumice, selected as an inert carrier in the UFBR, was not efficiently colonised, on the surface, by the microbial cells and most of the H2 was produced by suspended microorganisms, rather than by those adhering to the support. In the future, other materials (i.e. activated carbon) will be tested as supports.
The best performance, overall, of the CSTR, with a H2 yield of 0.30 L H2/g carbohydrate and a productivity rate of 4.50 mmol H2/L reactor, h and their co-occurrence at the same HRT of 39.2 h, was probably due to the mechanical stirring that promoted both H2 and CO2 removal from the fermentation broth, therefore reducing feed-back inhibition phenomena.
In continuous H2 generation, the UASB reactor performed well too with a high productivity rate (4.76 mmol H2/L reactor, h) and showing a better stability at higher organic loading rates (33.82 g/L, d), at to the lowest HRT (6.7 h) applicable. In the case of this cultivation system, inducers of granulation (i.e. powder of activated carbon) could be applied in order to speed biomass aggregation.
In future experiments, always using the same synthetic medium, the three bioreactors will be modified as suggested above and operated under vacuum (de-pressurized) conditions, in order to facilitate the release of dissolved H2 from the liquid phase of the systems. The next step will be the use of actual wastes, such as agro-industrial waste waters or thermo-hydrolyzed organic solid wastes.
NOMENCLATURE
Reactor operating (working) volume: accounts for only the portion of the reactor occupied by the fermentation liquid.
Influent flow rate (Qi): The influent flow rate is the average volumetric flow into the reactor and does not include the flow from recycling of effluent.
Hydraulic Retention Time (HRT): The HRT is a measure of the average amount of time which an ideal soluble compound would remain in the reactor and is calculated by dividing the liquid volume by the average feed volume flow rate and is normally presented in either days or hours.
Solids Retention Time (SRT): The SRT is a measure of the average amount of time which an ideal particulate would remain in the reactor.
Biological Oxygen Demand (BOD): The BOD is a biological measure of the organic strength of a wastewater stream and is used to measure influent BOD (BODi) and effluent BOD (BODe). BOD can be proportioned into total BOD and soluble BOD.
Chemical Oxygen Demand (COD): The COD is a chemical measure of the organic strength of a wastewater stream and is used to measure influent COD (CODi), effluent COD (CODe) and may also be fractionated into total COD (TCOD) and soluble COD (CODs). The COD is generally higher than the BOD measure of a given sample by the amount of refractory organics in the sample.
Suspended Solids (SS): SS is a measure of the weight of a filtered suspended solids sample which is dried at 105°C. SS may be reported as a % of wet sample or as a concentration. For dilute wastewaters the SS may be used as a measure of the microbial biomass concentration.
Volatile suspended solids (VSS): VSS is a measure of the organic fraction of a filtered suspended solids of a sample which can be burned at 550°C. It is found by removing the weight of the ash. VSS may be reported as a % of wet sample or as a concentration. For dilute wastewaters the VSS may be used as a measure of the active microbial biomass concentration in the sample.
10
Camilli and Pedroni.
Substrate concentration (S): For wastewaters, either BOD or COD is commonly used to measure the influent concentration. For sludge, manure, and solid wastes VS is more often used. For special wastewaters containing principally a single organic compound such as carbohydrates, acetic acid, etc. the measured concentration of the compound may be used for the substrate concentration.
Volatile Organic Acids (VOA): The VOA is a measure of the organic acid content (carboxylic acids) of the effluent and is often expressed as mg/l as acetic acid. VOA may often be termed Volatile Fatty Acids (VFA) and may be measured on a GC where concentrations of acetic, propionic, butyric, and valeric acids (and isomers) are found.
Volumetric Organic Loading Rate (OLR): The OLR is the average rate at which substrate is introduced into a unit volume of the reactor and is normally reported as kg S, Kg VS or Kg COD added per m3 reactor per day. The OLR accounts for both the influent flow rate and the influent substrate concentration.
Treatment efficiency (%Red): The treatment efficiency is the % of substrate destroyed to the amount of substrate processed. It is usually reported a %S, %VS or %COD destroyed.
Specific Hydrogen Yield (Y): Y is the volume (usually calibrated STP) of hydrogen produced for a given weight of substrate removed or destroyed. The substrate units may be expressed as S, VS or COD. Theoretically (homo-acetic conversion) exactly 0.50 L H2 are produced in dark fermentation processes for every gram of glucose removed (corresponding to 4.0 moles H2/mole glucose).
Specific Hydrogen Productivity (Rate, R): R is the volume of hydrogen produced per reactor volume per time unit (day or hour). Can be expressed as L (or mmol) H2/L reactor, day (or h) or as L (or mmol) H2/g VSS microbial biomass, day (or h). Both Y and R are a measure of the fermentative system performance.
REFERENCES
Anonymous (1956) Phenol sulphuric acid method for total carbohydrates assay.
Anal. Chem. 28: 350-356
APHA, AWWA, WEF (1992) Standard methods for the examination of water and wastewater.
18th edition, NY, USA
Asada Y. and Miyake J. (1999). Photobiological hydrogen production. J Biosci Bioeng 88: 1-6
Benemann JR (1996) Hydrogen Biotechnology: Progress and Prospects. Nature Biotechnol.,
14: 1101-1103
Benemann JR (1998) The Technology of Biohydrogen. In: Zaborsky OR (ed) BioHydrogen,
Plenum Press, New York, pp 19-30
Chang JS, Lee KS and Lin PJ (2002) Biohydrogen production with fixed bed bioreactors.
Int. J. Hydrogen Energy 27: 1167-1174
Chang FY and Lin CY (2004) Biohydrogen production using an upflow anaerobic sludge blanket
reactor. Int. J. Hydrogen Energy 29: 33-39
Claassen P.A.M.., J.W. van Groenestijn, A.J.H. Janssen, E.W.J. van Niel and R.H. Wijffels (2000)
Feasibility of biological hydrogen production from biomass for utilization in fuel cells. Proceedings of the 1st World Conference and Exhibition on Biomass for Energy and Industry, Sevilla, Spain.
Das D. and Veziroglu TN. (2001) Hydrogen production by biological processes: a survey of
literature. Int. J. Hydrogen Energy 26: 13-28
Fascetti E., D’addario E., Todini O. and Robertiello A. (1998) Photosynthetic hydrogen evolution
with volatile organic acids derived from the fermentation of source selected municipal solid wastes. Int. J. Hydrogen Energy 23: 753-760
Franchi E, Tosi C, Scolla G, Della Penna G, Rodriguez F and Pedroni PM (2004) Metabolically
11
Camilli and Pedroni
engineered Rhodobacter sphaeroides RV strains for improved biohydrogen photoproduction combined with disposal of food wastes. Marine Biotechnol., 6: 552-565
Hallenbeck, P.C. and Benemann J.R. (2002) Biological hydrogen production; fundamentals and
limiting process. Int. J. Hydrogen Energy 27: 1185-1193
Hawkes, F.R., Dinsdale, R., Hawkes, D.L. and Hussy, I. (2002) Sustainable fermentative hydrogen
production: challenges for process optimisation. Int. J. Hydrogen Energy 27: 1339-1347
Kumar, N.A., A. Ghosh and D. Das (2001) Redirection of biochemical pathways for the
enhancement of H2 production by Enterobacter cloacae. Biotechnol. Letters. 23: 537-541
Lay J. (2000) Modeling and optimization of anaerobic digested sludge converting starch to
hydrogen. Bioechnol. Bioeng. 68: 269-278
Lee, K.S., Wu, J.F., Lo, Y.S., Lo, Y.C., Lin, P.J. and Chang, J.S. (2004) Anaerobic hydrogen
production with an efficient carrier-induced granular sludge bed bioreactor. Biotechnol. Bioeng. 87: 648-657
Lin, C.Y. and Lay C.H .(2004) Carbon/nitrogen ratio effect on fermentative hydrogen production
by mixed microflora. Int .J. Hydrogen Energy 29: 41-45
Lucarelli C. (2004) La “dark fermentation” per la produzione di Bioidrogeno. Prove in scala di
laboratorio. Thesis of Environmental Engineering, Univ. “La Sapienza” Rome-Italy
Miyake, J., T. Matsunaga and A. San Pietro (eds.) 2001. Biohydrogen II, Pergamon Press,
Amsterdam
Miyake, J. Y. Igarashi and M. Roegner (eds.) 2004 Biohydrogen III, Elsevier
Nandi, R. and Sengupta S. (1998) Microbial production of hydrogen an overview.
Crit. Rev. Microbiol., 24: 61-84.
Owen W.F., Stuckey J.B., Healy J.R., Young L.Y. and McCarty P.L. (1979) Bioassay for
monitoring biochemical methane potential and anaerobic toxicity. Wat. Res., 13: 485-492
Pappa R., Massetti F., Robertiello A., Molinari M., Tornaboni P. (2002) Obiettivo”fango zero”.
T Point 4(1): 13-17, EniTecnologie-Via F. Maritano 26-20097 SD Milanese
Shapiro M. and Switzenbaum M.S. (1984) Initial anaerobic biofilm development. Biotech. Lett.,
6: 729-734
Woodward, J., Mattingly, S.M., Danson, M., Hough, D. Ward, N. and Adams M. (1996) In vitro
hydrogen production by glucose dehydrogenase and hydrogenase. Nature Biotechnol., 14:872-874
Zaborsky, O. (ed) 1998. Biohydrogen, Plenum Press, New York
Zhu H.G., Wakayama T., Suzuki T., Asada Y. and Miyake J. (1999) Entrapment of Rhodobacter
sphaeroides RV in cationic polymer/agar gels for hydrogen production in the presence of NH4+. J. Biosci. Bioeng., 88: 507-512
12