- Email: [email protected]
Design of an outdoor stacked – tubular reactor for biological hydrogen production
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e1 0
Available online at www.sciencedirect.com
ScienceDirect journal homepage: www.elsevier.com/locate/he
Design of an outdoor stacked e tubular reactor for biological hydrogen production Emine Kayahan, Inci Eroglu, Harun Koku* Middle East Technical University, Chemical Engineering Department, Ankara, Turkey
article info
abstract
Article history:
Photofermentation is one alternative to produce hydrogen sustainably. The photo-
Received 14 February 2016
bioreactor design is of crucial importance for an economically feasible operation, and an
Received in revised form
optimal design should provide uniform velocity and light distribution, low pressure drop,
13 April 2016
low gas permeability and efficient gaseliquid separation. A glass, stacked tubular biore-
Accepted 13 April 2016
actor aimed at satisfying these criteria has been designed for outdoor photofermentative
Available online xxx
hydrogen production by purple non sulfur bacteria. The design consists of 4 stacked Utubes (tube diameter 3 cm) and 2 vertical manifolds. The hydrodynamics of the 3-dimen-
Keywords:
sional model of this reactor was solved via COMSOL Multiphysics 4.1. The effects of tube
Photofermentative hydrogen pro-
length (1.4, 2.0, 3.8 m), tube pitch (8, 10.5, 13 cm) and volumetric flow rate (25e250 L/h) on
duction
the flow distribution were investigated. The glass stacked tubular reactor design results in
Photobioreactor design
less ground area and longer life time. This design has been constructed and operated with
Flow distribution
using Rhodobacter capsulatus YO3 hup and molasses as the carbon source under outdoor
Computational fluid dynamics (CFD)
conditions.
Stacked U-tube reactor
© 2016 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.
Rhodobacter capsulatus
Introduction Hydrogen is widely accepted as an alternative energy carrier. To be considered an alternative to fossil fuels, however, hydrogen should be produced in a renewable and sustainable manner. Light-driven photofermentation of organic wastes by purple non-sulfur (PNS) bacteria affords a biological alternative to produce hydrogen. The PNS bacteria are able to produce hydrogen using a large variety of wastes as substrate and under mild operating conditions by absorbing light energy at a wide range of the solar spectrum [1]. In order to gain acceptance as a viable route for hydrogen production, photofermentative hydrogen production has to be implemented in large-scale photobioreactors (PBRs) under
natural sunlight. Up to now, many of the problems encountered in large scale outdoor operations have been identified [2e4]. Running PBRs in outdoor conditions introduces complications such as fluctuating temperature and light conditions, which limits productivity. Moreover, large scale applications differ from the small scale ones in terms of light distribution, reactor material, gas collection and mixing. In this work, issues related to the scale up of photobioreactors (PBRs) were reviewed and accordingly, a new reactor design was proposed. Some of these issues are discussed below. The effect of temperature on photofermentative hydrogen production was previously investigated via small-scale experiments under fluctuating temperatures in the range of 15 C and 40 C, using wild type Rhodobacter capsulatus and R.
* Corresponding author. E-mail address: [email protected] (H. Koku). http://dx.doi.org/10.1016/j.ijhydene.2016.04.086 0360-3199/© 2016 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved. Please cite this article in press as: Kayahan E, et al., Design of an outdoor stacked e tubular reactor for biological hydrogen production, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.04.086
2
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e1 0
capsulatus YO3 which is a hup mutant. In the same study, the bacteria were subjected to 16 h light and 8 h dark cycles in order to mimic the diurnal cycle. As control, bacteria were grown under continuous illumination and under a constant temperature of 30 C. The variation in temperature decreased the hydrogen productivity from 1.4 mg H2/(L.h) to 0.5 mg H2/ (L.h). The light/dark cycles further decreased the productivity to 0.4 mg/(L.h). The authors hypothesized that the bacteria spent some of its energy for adaptation to the temperature changes and light/dark cycles [5]. Light/dark cycles are inevitable in outdoor operation, but their effects are relatively minor compared to the impact of temperature extremes, especially during summer months, which are usually preferred for outdoor campaigns due to the abundance of sunlight. With efficient cooling systems, however, the decrease in productivity due to temperature fluctuations could be mitigated. In experiments where the reactor temperature was kept under 30, 33, 35 and 40 C by shading or water spraying, the highest productivity (0.63 mg H2/(L.h)) was obtained when the reactor temperature was kept at 33 C or lower; on the other hand, the lowest productivity was observed as 0.44 mg H2/(L.h), when the reactor temperature was allowed to reach 40 C [5]. Light distribution within the PBR is critical for scale-up. For large culture depths, light penetration to the inner regions of the PBR becomes the limiting factor for productivity. Furthermore, when the available light is below a certain threshold, the bacteria could shift to completely unfavorable metabolic modes such as dark-fermentation. This could be prevented by increasing the surface to volume (or length to thickness) aspect ratio of the photobioreactors [6]. In one study, hydrogen production rates were measured using Roux bottles of varying depths. Hydrogen production was observed mainly at bottles with depths smaller than 1.5 cm [7]. It should be kept in mind, however, that high surface-to-volume ratios will increase the required land area, and may complicate the reactor construction and operation. Therefore a total thickness of 3 cm would be a proper compromise between good light distribution and limited surface area. Anaerobic conditions and limited nitrogen are required during photofermentative hydrogen production [1]. Therefore, another factor to consider is gas and more specifically hydrogen, permeability through the reactor material. Especially in large-scale systems, hydrogen loss through the reactor walls, manifolds and hydrogen gas collectors should be reduced as much as possible due to safety and economic considerations. Since a large surface area is inevitable due to light penetration requirements, it is necessary to decrease the permeability of the material as much as possible. The hydrogen permeability of the candidate PBR materials was assessed and experimentally verified in a previous study. The results of this study were shown in Table 1 [8]. Among these materials, glass was found as the least and LDPE the most permeable material, thus indicating glass is the best choice for the reactor material based on permeability alone. Apart from permeability, however, glass is also more advantageous in terms of mechanical strength and durability; it has been reported that the LDPE reactor tubes used in a previous study [2] had to be changed every
Table 1 e The hydrogen gas permeabilities of the candidate PBR materials. Material Poly (methyl methacrylate) (PMMA) Glass Polyurethane (PU) Polyvinyl chloride-plasticized (PVCplasticized) Low-density polyethylene (LDPE)
Permeability mol/ (m.s.Pa) 1.88 4.38 2.78 4.86
1014 1016 1012 1011
2.30 1011
year and the contribution of these tubes to the operating cost was around 65% [9]. As the reactor volume increases, product-gas collection also becomes a problem. It has been observed that as the reactor headspace pressure increased, hydrogen productivity decreased [10]. Therefore, it is important to maintain the total gas pressure in the headspace as low as possible, which will also promote lower retention of the produced gas, thereby decreasing the loss of hydrogen through the walls due to permeation as discussed above. In general, two reactor types have been preferred in literature for outdoor photofermentative hydrogen production: the panel type and the tubular type. The ratio of illuminated reactor surface to the occupied ground area for panel and tubular reactors was found as 8:1 and 1:1, respectively [11]. Despite this significant advantage in terms of required ground area, panel reactors have several drawbacks. Due to mechanical restrictions arising from the design, mixing becomes a problem in panel type photobioreactors whereas it can easily be achieved in tubular PBRs by a recirculation pump. There is also an upper limit for the volume of panel type photobioreactors; the mechanical integrity of the design begins to suffer beyond a certain size. In one study, severe swelling and deformation due to pressure was observed in a PVC-panel photobioreactor with 1 m 1 m dimensions [8]. Therefore, tubular PBRs also seem to be more advantageous in terms of durability. Mixing is promoted in photobioreactors in order to increase the mass transfer rate, to reduce nutrient gradients, to eliminate cell sedimentation and to facilitate the separation of the produced gas from the liquid culture. On the other hand, there appears to be an optimum mixing power value, above which productivity starts to decline. Though the reasons for the decrease of productivity at elevated mixing power values are unclear, one possibility is that shear stress, which depends on Reynolds number, induces cell damage [12]. In a study aimed to probe the effects of mixing, the performances of reactors shaken at 40, 80, 120 and 160 rpm were monitored and the average productivity was found to be the highest 5.3 mg H2/(L.h) when the shaking velocity was 120 rpm. This value was significantly higher than that of the control experiment with no shaking, reported as 3.72 mg H2/(L.h) [10]. In tubular photobioreactors, it has also been reported that hydrogen production is affected by the volumetric flow rate of recirculation that is used to promote mixing [11]. When the Reynolds number was varied from 10 to 6000, the highest productivity was observed for a Reynolds number of 240. No hydrogen production was observed when the volumetric flow
Please cite this article in press as: Kayahan E, et al., Design of an outdoor stacked e tubular reactor for biological hydrogen production, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.04.086
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e1 0
rate was larger than 2400. However, these experiments were carried out with a manifold type of photobioreactor where the flow distribution among the channels changed significantly with the Reynolds number. Consequently, the optimum Reynolds number for other systems should be obtained individually for other geometries. A flow model is especially useful in this regard, to obtain more insight into how the velocity distribution depends on the geometry. Finally, economics is still the main obstacle, preventing the application of such photofermentative hydrogen production systems in industrial scale. The large ground area required for a manifold type tubular photobioreactor by photofermentation was the reason of capital cost of photofermentation being several times larger than dark fermentation. Therefore, the ground area for the photobioreactor should be reduced [9]. The use of cooling water to keep the reactor temperature below 40 C is another factor that contributes to the hydrogen production cost. Another alternative for temperature control is passive cooling by shading the reactor. However, in a previous small scale outdoor study [5], the use of shading was not enough by itself to keep the reactor temperature at the desired values. Furthermore, as there is one to one correspondence between solar intensity and hydrogen production [4,13], decreasing the irradiance via shading is ultimately detrimental for gas production. As a result, active cooling methods such as water circulation or spraying are generally preferred. Based on the factors discussed up to this point, it can be argued that tubular reactors offer superior characteristics with respect to ease of mixing. Thus, in this work, a pilot-scale manifold type glass stacked U-tube photobioreactor was designed, built and operated. An upright orientation was selected to ensure small ground area to volume ratio with a large illuminated surface area, and the reactor material was chosen as glass for durability and low permeability to hydrogen and air. In designing the tubular photobioreactor geometry, a hydrodynamic model was used to assess whether a low overall pressure drop and a uniform velocity distribution could be achieved throughout the reactor and the dimensions of the reactor was based on the results of the model. The reactor was then constructed and run with and without bacteria, the latter grown using a molasses solution as a complex substrate.
Materials and methods Modeling The flow distribution inside a pilot-scale manifold type glass stacked U-tube photobioreactor (Fig. 1) was determined by the utilization of the single phase laminar flow module in COMSOL 4.4. Water at 30 C was considered as the working fluid. The laminar flow interface is based on the NaviereStokes equations. In the model, steady-flow was assumed for a Newtonian, incompressible and isothermal liquid medium. With these assumptions, the NaviereStokes equations for mass and momentum balances reduce to Equations (1) and (2), respectively.
3
rV,u ¼ 0
(1)
rðu,VÞu ¼ V PI þ mðVu þ ðVuÞT þ rg
(2)
where r is the density (kg/m3), u is the velocity (m/s), P is the pressure (Pa), I is the identity tensor, m is viscosity (Pa.s) and g is the gravitational acceleration constant in (m/s2). The no-slip boundary condition was applied to all the solid boundaries. At the inlet, the fluid velocity was fixed as shown by Equations (4) and (5). As it can be seen from Equation (4), the flow was assumed to be laminar and fully developed. At the outlet, the gauge pressure was zero, as shown in Equation (6). uwall ¼ 0
(3)
r2 uinlet ¼ 2$uave $ 1 2 R
(4)
uave ¼ u0 $n
(5)
Pout ¼ 0
(6)
where uwall is the velocity at the tube wall (m/s), uinlet is the inlet velocity (m/s), uave is the average velocity (m/s), R is the pipe radius and r is the distance from the center, u0 is the velocity magnitude at the inlet (m/s), n is the normal unit vector pointing out of the domain, and Pout is the outlet gauge pressure (Pa). COMSOL utilizes the finite element method while solving partial differential equations. GMRES (generalized minimum residual method) was used as the solution technique in the model. The tolerance factor was set to 0.001. The model was solved with a mesh number of 609805. A mesh convergence study was carried out by comparing the velocities in all the tubes separately. The total difference was found to be less than 5% when the mesh number was increased from 609805 to 2600023.
Pilot-scale stacked U-tube photobioreactor The photobioreactor (10.4 L) was designed by our group and the glass pieces were custom-built. The diameter of the manifolds and the tubes were 6 cm and 3 cm, respectively. The length of the manifolds and tubes were 0.475 m and 4 m, respectively. The wall-thickness of the glass tubes and the manifolds were 1.5 and 2.2 mm, respectively. All glass pieces were built from Schott borosilicate glass, reported to transmit about 90% of the light within the 350e2000 nm wavelength range (The complete optical properties can be accessed at www.schott.com). The gas was collected from the top of the manifolds by water displacement. Cooling was implemented via circulation of chilled water through spiral glass cooling coils within the manifolds. Moreover, PVC tubes (7 mm in diameter) were inserted into the glass tubes as an additional (and optional) circulatory cooling system, to be activated in case the cooling effect provided by the manifolds alone proved to be inadequate (not shown in process flow diagram). Twelve temperature measurement ports were placed on the tubes as shown in the process flow diagram in Fig. 1.
Please cite this article in press as: Kayahan E, et al., Design of an outdoor stacked e tubular reactor for biological hydrogen production, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.04.086
4
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e1 0
Fig. 1 e Process flow diagram of the stacked U-tube PBR. T1 e T12 are the temperature probes. V1 and V2 are check valves (1/ 3 psi), V4, V5, V6 and V7 are ball valves. CW-in and CW-out are cooling water inlet and outlet, respectively.
Two additional temperature measurement locations were designated on the manifolds. A centrifugal pump was used for recirculation of the liquid culture. The flow rate was adjusted using valves, V6 and V7. A recycle around the pump was necessary for flow rate adjustments. During feeding, recirculation was stopped by closing the valve, V4. Feed was given to the reactor by feed line (V5 is open) and the same amount of content was taken out of the reactor from the discharge line (V3 is open). For ease of gas collection, the tubes were placed with 30 inclination as can be seen from the photograph of the stacked U-tube PBR shown in Fig. 2.
Photobioreactor startup and operation The R. capsulatus YO3 (Hup) strain, which was modified by deleting the gene coding for uptake hydrogenase enzyme (Hup) of the original R. capsulatus MT1131 strain, was used in this study [14]. The bacteria were activated in the Biebl and Pfennig (1981) medium containing 20 mM acetate and 10 mM
glutamate as the carbon and nitrogen sources, respectively. A 22 mM potassium phosphate buffer was used, and the pH was adjusted to 6.4e6.5 by a 5 mM NaOH solution. The bacteria were grown under continuous illumination with 2000 lux and transferred to a second medium for sucrose adaptation, which contained 20 mM acetate, 5 mM sucrose and 10 mM glutamate besides the Biebl and Pfennig (1981) medium. Finally, the bacteria were inoculated to the biological hydrogen production medium, which contained 5 mM sucrose from molasses (obtained from the Ankara Sugar Factory, Ankara, Turkey), and then transferred to the stacked U-tube PBR. 30 mM KH2PO4 solution was used as buffer for the reactor medium and the initial pH was adjusted to 7.5. All inoculations (10% v/ v) were made when the concentration of the bacteria reached around 1 g/L, and argon was flushed to obtain anaerobic conditions. The reactor was sterilized by recirculating a 3% H2O2 solution for 24 h, then washed twice with distilled water. The recirculation rate was set as 43.5 L/h (Re ¼ 160). At the start-up, 10 mL samples were withdrawn from each tube for analysis, and 40 mL of 5 mM sucrose-containing
Fig. 2 e A photograph of the stacked U-tube photobioreactor. Please cite this article in press as: Kayahan E, et al., Design of an outdoor stacked e tubular reactor for biological hydrogen production, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.04.086
5
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e1 0
molasses with 30 mM KH2PO4 buffer solution (pH ¼ 7.5) was fed to the reactor to maintain the original volume (Phase I). After the 4th day, 1 L feed was given to the reactor daily, and 1 L of the reactor content was discharged at the same time. Sucrose content in the feed was changed many times according to changes in pH, cell concentration, organic acid and sucrose concentration. The feeding strategy is summarized in Table 2.
column. 100 mL samples, injected with gas-tight syringes (Hamilton, 22 GA 500 mL) were analyzed by a thermal conductivity detector. Argon with a flow rate of 26 mL/min was used as the carrier gas. The oven, injector and detector temperatures were 140, 160 and 170 C, respectively.
Results and discussion
Analytical methods
Model results
The bacterial cell concentration was measured by a spectrophotometer (Shimadzu UV-1201) at 660 nm. An optical density of 1.0 measured at 660 nm corresponds to 0.47 gdcw/ Lc for R. capsulatus Hup. pH was measured with a Mettler Toledo 3311 transmitter. Averages of 3 measurements taken from each sampling port were reported for bacterial cell concentration and pH. Periodic temperature measurements were taken from the outdoor reactor. Online temperature measurements were obtained from Fe-constant J-type thermocouples connected to a data logger (Ordel UDL100). Air temperature and cooling water inlet and outlet temperatures were also recorded. A pyronometer (HOBO-S-LIBM003), connected to an online weather station (HOBO® U30 ETH), was used for solar radiation measurements. Organic acid analyses were carried out using a High Performance Liquid Chromatography system (Shimadzu 20A series) equipped with an Alltech IOA-1000 column (300 mm 7.8 mm). The mobile phase (0.0085 M H2SO4 solution) was pumped with a low gradient pump (Shimatzu LC-10AT) with a flow rate of 0.4 mL/min. The oven temperature was fixed at 66 C. 10 mL samples were analyzed with a UV detector (Shimadzu FCV-10AT) whose absorbance was set at 210 nm [8,13]. The measured organic acids were lactic acid, formic acid, acetic acid, propionic acid and butyric acid. The sucrose content was analyzed by a commercial kit based on glucoseeoxidase reactions (Biyozim kit, Biasis, Turkey). The kit determines the sucrose content first by hydrolyzing sucrose to fructose and glucose and then reacting the latter to form a red-colored immunoquinone solution. The concentration of immunoquinone can be determined at 505 nm by a spectrophotometer (Shimadzu UV-1201), and in turn used for back-calculation of sucrose. The calibration curves were determined using standard sucrose solutions and the sucrose concentrations in the samples were determined accordingly. The gas composition was determined by a gas chromatograph (Agilent Technologies 6890N) equipped with a Supelco Carboxen 1010
Within the scope of this work, the effect of tube length, tube pitch (i.e., the spacing between the parallel tubes measured from the center of one tube to the center of the other one), and volumetric flow rate on the flow distribution were studied. For convenience, the results were interpreted by introducing the fractional volumetric flow rate, defined for the ith tube as shown in Equation (7). If the flow distribution is uniform among the tubes, then bi would be 0.25 for all the tubes for a manifold with 4 channels.
Table 2 e Feeding strategy. Phase I II III IV V VI VII
Sucrose concentration in molasses (feed) 5 mM 5 mM 100 mM e 50 mM e 50 mM
pH Feed rate (feed) 7.5 7.5 7.5 e 7.5 e 11e12
40 mL/day 1 L/day 1 L/day No feeding 1 L/day No feeding 1 L/day
bi ¼
Qi Q0
(7)
where Qi and Q0 represents the volumetric flow rate (m3/s) in the ith tube and total volumetric flow rate. The flow non-uniformity parameter (F), shown in Equation (8) was defined to quantify the flow distribution. This parameter is analogous to standard deviation and could be defined as the fractional volumetric flow rate when the flow is distributed equally to all channels [15]. sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi P4 2 i¼1 ðbi 1=nÞ $n$100 F¼ n
(8)
where n denotes the number of tubes, bi is the fractional volumetric flow rate in the ith tube. The summary of the design parameters taken into consideration and the corresponding non-uniformity parameters estimated from the model are listed in Table 3. Fig. 3 illustrates the variation of fractional volumetric flow rate (bi) for a 4 channel stacked tubular PBR for three different Table 3 e The summary of the design parameters studied and the corresponding non-uniformity parameter estimated from Equation (8). Tube length (m) 1.4 2.0 3.8 1.4 1.4 1.4 1.4 1.4 1.4 1.4 1.4 1.4 1.4 1.4
Tube pitch Overall volumetric flow Ret F (cm) rate (L/h) (%) 10.5 10.5 10.5 8.0 13.0 10.5 10.5 10.5 10.5 10.5 10.5 10.5 10.5 10.5
25.0 25.0 25.0 25.0 25.0 50.0 75.0 100.0 125.0 150.0 175.0 200.0 225.0 250.0
92.0 92.0 92.0 92.0 92.0 184.0 276.0 368.0 460.0 552.0 644.0 736.0 828.0 920.0
2.1 2.1 1.5 2.1 3.1 2.2 2.2 2.2 2.3 2.3 2.3 2.3 2.3 2.3
Please cite this article in press as: Kayahan E, et al., Design of an outdoor stacked e tubular reactor for biological hydrogen production, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.04.086
6
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e1 0
Fig. 3 e Effect of tube length on the fractional volumetric flow rate (bi) for a total volumetric flow rate (Q0) of 25.0 L/h and a tube pitch of 10.5 cm. Tube 1 is the closest to the ground.
tube lengths 1.4 m, 2.0 m and 3.8 m, while maintaining the volumetric flow rate at 25.0 L/h, and the tube pitch as 10.5 cm. Tube channels were counted from the bottom (closest to the ground is Tube 1) to the top (Tube 4). The flow non-uniformity parameter was calculated as 2.1% for both 1.4 m and 2.0 m long tubes, but found to be 1.5% for 3.8 m long tubes, indicating more uniform flow distribution in the latter. As the tube length is increased, resistance to flow increases in all of the tubes. The flow distribution is considerably affected by the tube length when the frictional forces in the tubes are larger than the frictional forces in the manifolds. This behavior was also noted in a recent study; although there, the flow resistance due to channel length was found to be small compared to the flow resistance in the manifolds, thus reducing the effect of tube length [16]. Usually, tube length is increased to scale-up tubular PBRs, therefore, it may be useful to further increase the tube length, as long tubes result in high residence times; however this in turn may increase the likelihood of leakage and it is difficult to maintain such systems at the desired temperature. The pressure drop between the inlet and outlet was found as 0.30, 0.34 and 0.49 Pa for tube lengths of 1.4 m, 2.0 and 3.6 m, respectively, and thus can be considered as almost negligible. The second parameter investigated in the model was the tube pitch. Three tube pitches, 8.0, 10.5 and 13.0 cm, were studied while other parameters were kept constant. The volumetric flow rate and tube length were 25.0 L/h and 1.4 m, respectively. The most uniform flow distribution was obtained for a tube pitch of 10.5 cm (Fig. 4). As the tube pitch increases, the diameter to height ratio of the distribution and collection manifolds decreases in this geometry. The effect of this ratio on the flow distribution was previously studied for other manifold models [16,17]. The diameter to height ratio for the distribution manifold has previously been stated as the ‘main governing parameter’ in design, whereas the same ratio for the combination manifold was found as a controlling parameter [16]. This behavior could be easily interpreted via the effective forces in such manifolds. In general, for horizontal manifolds, the flow distribution is determined by the wall friction and momentum change. As the manifold diameter to height ratio decreases for the
Fig. 4 e Effect of tube pitch on the fractional volumetric flow rate (bi) for a total volumetric flow rate (Q0) of 25.0 L/h and a tube length of 1.4 m. Tube 1 is the closest to the ground.
distribution manifold, friction becomes the dominant force, whereas for larger diameter to length ratios, momentum effects are the dominant factor. In the collection manifold, both the momentum and friction effects work in the same direction, as the fluid moves towards the outlet [17,18]. In addition to these forces, for vertical manifolds, gravity also affects the flow distribution. Therefore, there is an optimum diameter to length ratio when these forces balance each other. In this work, the optimum tube pitch was found as 10.5 cm. The pressure drops were found as 0.30, 0.30 and 0.29 Pa for 8, 10.5 and 13 cm tube pitches respectively. The volumetric flow rate was increased from 25 L/h to 250 L/h with 25 L/h increments. For these cases, the tube pitch was 10.5 cm and tube length was 1.4 m. It is more convenient to report the volumetric flow rates in terms of Reynolds number (Equation (9)). The tube velocity used in the calculation of the Reynolds number is shown in Equation (10). Ret ¼
ut ¼
r$ut $Dt m
4$Q0 n$p$D2t
(9)
(10)
where r is the density (kg/m3), ut is the tube velocity if the flow was distributed to all 4 tubes uniformly, Dt is the tube diameter (m), m is the viscosity (Pa.s), Q0 is the total volumetric flow rate (m3/s) and n is the number of tubes (4 in our case). In this study, fractional volumetric flow rate was compared using the tube Reynolds number, Ret. It should be noted that Ret is not the actual Reynolds number in the tubes computed from the model, but rather a hypothetical number, that gives a sense about what would have been the Reynolds number in the tubes if the flow was distributed uniformly to all the tubes. Fig. 5 indicates that increasing the volumetric flow rate (Ret) did not change the flow distribution. More fluid tended to go from the lowest tube for all the volumetric flow rates. This result revealed that both gravitational and momentum effects were higher compared to frictional effects in the distribution manifold. Still, looking at the flow non-uniformity parameter changing between 2.1% and 2.3% for different flow rates, it could be concluded that a uniform flow distribution was achieved.
Please cite this article in press as: Kayahan E, et al., Design of an outdoor stacked e tubular reactor for biological hydrogen production, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.04.086
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e1 0
Fig. 5 e Effect of Reynolds number (Ret) on the fractional volumetric flow rate (bi) for a tube pitch of 10.5 and tube lengths of 1.4 m. Tube 1 is the closest to the ground.
7
tubes accumulates on top of the manifolds, while liquid flows downwards and is recirculated back to the tubes by means of a pump. Vertical manifolds also have another advantage; tubular reactors have been claimed to be disadvantageous compared to panel reactors due to their lower illuminated surface area to ground area ratio [11]. However, with the current design, the ratio of the illuminated surface area to ground area was increased by a factor of almost 5 compared to a fully horizontal design [2], with approximately 5 m2 (4.8 m2) of the illuminated surface area found to occupy 1 m2 ground area. Moreover, this ratio could be further increased by increasing number of tubes connected to the manifold.
Photobioreactor operation results The pressure difference between the inlet and outlet of the reactor with respect to Ret is plotted in Fig. 6. The pressure drop increased as Ret increased, as expected. The values were low; however, even for the highest volumetric flow rate (250 L/ h) for which the pressure difference between the inlet and outlet of the reactor was only 20.3 Pa. As can be seen in Table 3, the most uniform flow distribution (the non-uniformity parameter is 1.5%) is obtained for a tube length of 3.8 m, a tube pitch of 10.5 cm and a volumetric flow rate of 25 L/h. Accordingly, the tube length for the final design of the stacked U-tube photobioreactor was chosen as 4 m, and the tube pitch was chosen as 10.5 cm.
The reactor was initially tested without bacteria to verify that the cooling and circulation systems worked properly without leakage. The experiment with R. capsulatus YO3 was conducted between September 9 and September 29, 2014. The reactor was cooled intermittently, only on days 2, 3, 4 and 6 at noon time, to keep the culture temperature below 40 C. Temperature did not change significantly along the length of the reactor (Fig. 7a) and for different tubes (Fig. 7b). Therefore, it can be concluded that effective cooling was achieved even if the system was cooled just by using the manifolds. Even without cooling, the temperature variation according to tube position and length was not appreciable (Fig. 8). It can be
Construction of the reactor The stacked tubular photobioreactor was constructed, based on the results of the simulations discussed above, and the findings of previous studies. Fig. 2 displays a photograph of stacked U-tube photobioreactor. Uniform velocity distribution with small pressure drop was achieved with this design as validated by our model. Glass was chosen as the reactor material, since it has low hydrogen and air permeability compared to other candidate PBR materials as mentioned before. The tube diameter was selected as 3 cm [7], based on the results of the recent studies to achieve good light distribution. For ease of gaseliquid separation, the manifolds were placed vertically. With this design, gas formed within the
Fig. 6 e Pressure drop with respect to Reynolds number (Ret). The tube pitch is 10.5 cm and the tube length is 1.4 m.
Fig. 7 e Temperature variation with time on the 5th day of the experiment: (a) along the length of tube 4, at the inlet (T1), midpoint (T5) and exit (T9) of the tube; (b) at the exits of tube 4 (T9), tube 3 (T10) and tube 1 (T11). See Fig. 1 for the locations of the tubes and measurement probes. The arrows indicate the operation of the cooling system (↑: Cooling on, ↓: cooling off).
Please cite this article in press as: Kayahan E, et al., Design of an outdoor stacked e tubular reactor for biological hydrogen production, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.04.086
8
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e1 0
Fig. 8 e Temperature variation with time, with cooling switched off, on the 9th day of the experiment: (a) along the length of tube 4, at the inlet (T1), midpoint (T5) and exit (T9) of the tube; (b) at the exits of tube 4 (T9), tube 3 (T10), tube 2 (T11) and tube 1 (T12). See Fig. 1 for the locations of the tubes and measurement probes.
observed that the reactor temperature followed the same trend with the ambient air temperature, but was slightly higher. The daily temperature variation within the reactor is illustrated in Fig. 9. As expected, the liquid culture temperature was highly affected by the solar radiation. Since most of the time there was no cooling, the temperature of the reactor was mainly driven by solar radiation. We based the feeding strategy on our previous indoor studies. The highest hydrogen production was found with R. capsulatus YO3 with 5 mM sucrose containing molasses in a
batch photobioreactor [19]. However, due to the fact that the observed sucrose consumption rate of the present study was found to deviate considerably from indoor results, a more variable strategy had to be adopted based on the culture response in order to maintain the nutrients and reactor conditions close to optimal conditions. The ‘phases’ of this feeding strategy were given in Table 2 and are explained further below. The initial sucrose concentration of molasses was adjusted to 5 mM. The cell concentration stabilized at the 3rd day of the experiment (Phase I in Fig. 10) and the sucrose concentration (Fig. 11) decreased to very low values (around 0.5 mM). To increase the cell concentration, Phase II, involving continuous feeding of molasses was started on the 4th day. During Phase II, the organic acid content remained below 4 mM (Fig. 12). At the end of 10 days, cell concentration dropped to a low value (0.12 gdcw/L) due to the rapid depletion of carbon source, requiring a substantial addition of sucrose to maintain cell viability. Thus, in Phase III, 100 mM sucrose containing molasses were fed to the reactor. This led to a significant increase in the organic acid content, in turn, resulting in a decrease of pH (Fig. 12). On the 12th day, the bacteria finally reached optimal levels of cell concentration for hydrogen production (around 0.4 gdcw/L). A total of 454 mL H2 (corresponding to a productivity of 0.327 mg H2/(L.h)) was produced in this phase, on the 12th day of the experiment. The gas composition was 60.9% H2, with CO2 as the remainder. As the solubility of CO2 decreases at low pH values, its percentage was higher compared to previous photofermentative hydrogen production studies [20,21] where pH was kept close to its optimum value of 7 [22]. After the 12th day (Phase IV), hydrogen production ceased, possibly due to the sustained decrease in pH. It could be speculated that the metabolism of the bacteria had shifted to other modes following Phase III, based on the increase in lactic and acetic acid concentrations. To recover steady operation, feeding was stopped for two days, resulting in stabilized organic acid content in the reactor on the 13th and 14th days of the experiment. As a result, the cell concentration increased to its highest value (1.0 gdcw/L) on the 14th day of the experiment. Subsequently, one liter of 50 mM sucrose-containing molasses was fed to the reactor on the 15th day (Phase V). In the following day, as elevated levels
Fig. 9 e Comparison of the temperature variation in the reactor with that of the ambient air, and the daily solar radiation (T9 is measured at the exit port of tube 4). Please cite this article in press as: Kayahan E, et al., Design of an outdoor stacked e tubular reactor for biological hydrogen production, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.04.086
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e1 0
9
Fig. 10 e Daily cell concentration variation in the reactor.
whole too unfavorable for hydrogen production or growth, hence the run was terminated on the 20th day.
Conclusions
Fig. 11 e Daily sucrose concentration variation in the reactor.
of organic acids were observed again, feeding was stopped once more (Phase VI). However, this led to a decrease in cell concentration levels, and consequently, during the last phase (Phase VII), 1 L of 50 mM sucrose containing molasses was fed to the reactor daily. In the feed, pH was adjusted to 11e12 with addition of NaOH solution to counter the drop in pH. During this phase, all organic acid concentrations decreased; however, pH stabilized around 5.0. As cell concentration continued to decrease, the culture conditions were decided to be on the
A new stacked-tubular design was proposed in this study with the goal of addressing problems with scale up of PBRs for hydrogen production. A hydrodynamic model for this design was analyzed, and it was found that the most uniform flow distribution was obtained for a tube pitch of 10.5 cm, a tube length of 3.6 m and a volumetric flow rate of 25 L/h. The design was then constructed with dimensions based on the model results. In view of its lower long-term material costs and low permeability to hydrogen and air, glass was chosen as the reactor material. The illuminated surface area to ground area of the reactor was found as 5:1, a high value for tubular-type reactors [11]. To test the new reactor, an experiment was carried out using molasses as the carbon source and R. capsulatus YO3 (Hup) as the microorganism. During the experiment, the feeding strategy was varied with the purpose of finding the best feeding strategy for single stage biological hydrogen production. It can be argued that maintaining the sucrose content in the reactor around 5 mM, and keeping the acetic acid and lactic acid content below 40 mM could be a
Fig. 12 e Daily organic acid variation in the reactor. Please cite this article in press as: Kayahan E, et al., Design of an outdoor stacked e tubular reactor for biological hydrogen production, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.04.086
10
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e1 0
good strategy for photofermentative hydrogen production in outdoor conditions. Many of the indoor studies on biological hydrogen production are batch; whereas in outdoor studies, fed batch operation could be more advantageous. Ultimately, the choice of a proper feeding strategy is an issue that warrants further investigation for scale up.
Acknowledgments _ ¨ BITAK This work was part of the ongoing project funded by TU (114M436) and by the METU BAP project (BAP-07-02-2014-007538).
Nomenclature Symbol Dt F g I n n P Pout Q Qi Q0 R r Ret u uave uinlet ut uwall u0 bi m r
tube diameter, m non-uniformity parameter gravitational acceleration, m/s2 identity tensor number of tubes normal unit vector pointing out of the domain pressure, Pa outlet gauge pressure, Pa volumetric flow rate, m3/s volumetric flow rate in the ith tube, m3/s total volumetric flow rate, m3/s pipe radius, m distance from the center, m tube Reynolds number velocity, m/s average velocity, m/s inlet velocity, m/s tube velocity if the flow was distributed to all 4 tubes uniformly velocity at the tube wall, m/s velocity magnitude at the inlet, m/s dimensionless volumetric flow rate in the ith tube viscosity, Pa.s density, kg/m3
Note: Symbols with single and double underlines indicate vector and tensor quantities, respectively.
references € ¨ r E, Eroglu E, Yu¨cel M, Gu¨ndu¨z U. Applications [1] Eroglu I, Ozgu of photofermentative hydrogen production. In: Zannoni D, De Philippis R, editors. Microbial bioenergy: hydrogen production. Springer; 2014. p. 237e67. € ¨ r E, Yu¨cel M, Gu¨ndu¨z U, Eroglu I. “Biohydrogen [2] Boran E, Ozgu production by Rhodobacter capsulatus Hup mutant in pilot solar tubular photobioreactor. Int J Hydrogen Energy Nov. 2012;37(21):16437e45.
[3] Adessi A, Torzillo G, Baccetti E, De Philippis R. Sustained outdoor H2 production with Rhodopseudomonas palustris cultures in a 50 L tubular photobioreactor. Int J Hydrogen Energy 2012;37(10):8840e9. € [4] Avcıoglu SG, Ozgur E, Eroglu I, Yu¨cel M, Gu¨ndu¨z U. “Biohydrogen production in an outdoor panel photobioreactor on dark fermentation effluent of molasses. Int J Hydrogen Energy Aug. 2011;36(17):11360e8. € ¨ r E, Uyar B, Oztu € ¨ rk Y, Yu¨cel M, Gu¨ndu¨z U, Eroǧlu I. [5] Ozgu Biohydrogen production by Rhodobacter capsulatus on acetate at fluctuating temperatures. Resour Conserv Recycl Mar. 2010;54(5):310e4. € ¨ r E, Eroglu I, Gu¨ndu¨z U, Yu¨cel M. [6] Androga DD, Ozgu Photofermentative hydrogen production in outdoor conditions. In: Minic D, editor. Hydrogen energy e challenges and perspectives. InTech; 2012. [7] Nakada E, Asada Y, Arai T, Miyake J. Light penetration into cell suspensions of photosynthetic bacteria and relation to hydrogen production. J Ferment Bioeng 1995;80(1):53e7. [8] Avcioglu SG. Scale up of panel photobioreactors for hydrogen production by PNS bacteria [M.Sc. thesis]. Middle East Technical University; 2010. [9] Urbaniec K, Grabarczyk R. Hydrogen production from sugar beet molasses e a techno-economic study. J Clean Prod 2014;65:324e9. [10] Li X, Wang Y, Zhang S, Chu J, Zhang M, Huang M, et al. Effects of light/dark cycle, mixing pattern and partial pressure of H2 on biohydrogen production by Rhodobacter sphaeroides ZX5. Bioresour Technol 2011;102(2):1142e8. [11] Gebicki J, Modigell M, Schumacher M, Van Der Burg J, Roebroeck E. Comparison of two reactor concepts for anoxygenic H2 production by Rhodobacter capsulatus. J Clean Prod 2010;18(Suppl. 1):S36e42. [12] Flickinger MC. Upstream industrial biotechnology, 2 volume set. John Wiley & Sons; 2013. [13] Androga DD, Ozgur E, Gunduz U, Yucel M, Eroglu I. Factors affecting the longterm stability of biomass and hydrogen productivity in outdoor photofermentation. Int J Hydrogen Energy 2011;36(17):11369e78. € ¨ rk Y, Yu¨cel M, Daldal F, Mandaci S, Gu¨ndu¨z U, Tu¨rker L, [14] Oztu et al. Hydrogen production by using Rhodobacter capsulatus mutants with genetically modified electron transfer chains. Int J Hydrogen Energy 2006;31(11):1545e52. [15] Fan J, Shah LJ, Furbo S. Flow distribution in a solar collector panel with horizontally inclined absorber strips. Sol Energy 2007;81:1501e11. [16] Ahn H, Lee S, Shin S. Flow distribution in manifolds for low Reynolds number flow. KSME Int J 1998;12(1):87e95. [17] Wang J. Theory of flow distribution in manifolds. Chem Eng J 2011;168(3):1331e45. [18] Acrivos A, Babcock BD, Pigford RL. Flow distributions in manifolds. Chem Eng Sci 1959;10(1e2):112e24. ır E. Photobiological hydrogen production from sugar beet [19] Sag molasses [M.Sc. thesis]. Middle East Technical University; 2012. [20] Koku H, Eroǧlu I, Gu¨ndu¨z U, Yu¨cel M, Tu¨rker L. Kinetics of biological hydrogen production by the photosynthetic bacterium Rhodobacter sphaeroides O.U. 001. Int J Hydrogen Energy 2003;28(4):381e8. € ¨ r E, Yu¨cel M, Gu¨ndu¨z U, Eroglu I. Biohydrogen [21] Boran E, Ozgu production by Rhodobacter capsulatus in solar tubular photobioreactor on thick juice dark fermenter effluent. J Clean Prod 2012;31:150e7. [22] Koku H, Eroglu I, Gu¨ndu¨z U, Yu¨cel M, Tu¨rker L. Aspects of the metabolism of hydrogen production by Rhodobacter sphaeroides. Int J Hydrogen Energy 2002;27(11e12):1315e29.
Please cite this article in press as: Kayahan E, et al., Design of an outdoor stacked e tubular reactor for biological hydrogen production, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.04.086
Available online at www.sciencedirect.com
ScienceDirect journal homepage: www.elsevier.com/locate/he
Design of an outdoor stacked e tubular reactor for biological hydrogen production Emine Kayahan, Inci Eroglu, Harun Koku* Middle East Technical University, Chemical Engineering Department, Ankara, Turkey
article info
abstract
Article history:
Photofermentation is one alternative to produce hydrogen sustainably. The photo-
Received 14 February 2016
bioreactor design is of crucial importance for an economically feasible operation, and an
Received in revised form
optimal design should provide uniform velocity and light distribution, low pressure drop,
13 April 2016
low gas permeability and efficient gaseliquid separation. A glass, stacked tubular biore-
Accepted 13 April 2016
actor aimed at satisfying these criteria has been designed for outdoor photofermentative
Available online xxx
hydrogen production by purple non sulfur bacteria. The design consists of 4 stacked Utubes (tube diameter 3 cm) and 2 vertical manifolds. The hydrodynamics of the 3-dimen-
Keywords:
sional model of this reactor was solved via COMSOL Multiphysics 4.1. The effects of tube
Photofermentative hydrogen pro-
length (1.4, 2.0, 3.8 m), tube pitch (8, 10.5, 13 cm) and volumetric flow rate (25e250 L/h) on
duction
the flow distribution were investigated. The glass stacked tubular reactor design results in
Photobioreactor design
less ground area and longer life time. This design has been constructed and operated with
Flow distribution
using Rhodobacter capsulatus YO3 hup and molasses as the carbon source under outdoor
Computational fluid dynamics (CFD)
conditions.
Stacked U-tube reactor
© 2016 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.
Rhodobacter capsulatus
Introduction Hydrogen is widely accepted as an alternative energy carrier. To be considered an alternative to fossil fuels, however, hydrogen should be produced in a renewable and sustainable manner. Light-driven photofermentation of organic wastes by purple non-sulfur (PNS) bacteria affords a biological alternative to produce hydrogen. The PNS bacteria are able to produce hydrogen using a large variety of wastes as substrate and under mild operating conditions by absorbing light energy at a wide range of the solar spectrum [1]. In order to gain acceptance as a viable route for hydrogen production, photofermentative hydrogen production has to be implemented in large-scale photobioreactors (PBRs) under
natural sunlight. Up to now, many of the problems encountered in large scale outdoor operations have been identified [2e4]. Running PBRs in outdoor conditions introduces complications such as fluctuating temperature and light conditions, which limits productivity. Moreover, large scale applications differ from the small scale ones in terms of light distribution, reactor material, gas collection and mixing. In this work, issues related to the scale up of photobioreactors (PBRs) were reviewed and accordingly, a new reactor design was proposed. Some of these issues are discussed below. The effect of temperature on photofermentative hydrogen production was previously investigated via small-scale experiments under fluctuating temperatures in the range of 15 C and 40 C, using wild type Rhodobacter capsulatus and R.
* Corresponding author. E-mail address: [email protected] (H. Koku). http://dx.doi.org/10.1016/j.ijhydene.2016.04.086 0360-3199/© 2016 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved. Please cite this article in press as: Kayahan E, et al., Design of an outdoor stacked e tubular reactor for biological hydrogen production, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.04.086
2
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e1 0
capsulatus YO3 which is a hup mutant. In the same study, the bacteria were subjected to 16 h light and 8 h dark cycles in order to mimic the diurnal cycle. As control, bacteria were grown under continuous illumination and under a constant temperature of 30 C. The variation in temperature decreased the hydrogen productivity from 1.4 mg H2/(L.h) to 0.5 mg H2/ (L.h). The light/dark cycles further decreased the productivity to 0.4 mg/(L.h). The authors hypothesized that the bacteria spent some of its energy for adaptation to the temperature changes and light/dark cycles [5]. Light/dark cycles are inevitable in outdoor operation, but their effects are relatively minor compared to the impact of temperature extremes, especially during summer months, which are usually preferred for outdoor campaigns due to the abundance of sunlight. With efficient cooling systems, however, the decrease in productivity due to temperature fluctuations could be mitigated. In experiments where the reactor temperature was kept under 30, 33, 35 and 40 C by shading or water spraying, the highest productivity (0.63 mg H2/(L.h)) was obtained when the reactor temperature was kept at 33 C or lower; on the other hand, the lowest productivity was observed as 0.44 mg H2/(L.h), when the reactor temperature was allowed to reach 40 C [5]. Light distribution within the PBR is critical for scale-up. For large culture depths, light penetration to the inner regions of the PBR becomes the limiting factor for productivity. Furthermore, when the available light is below a certain threshold, the bacteria could shift to completely unfavorable metabolic modes such as dark-fermentation. This could be prevented by increasing the surface to volume (or length to thickness) aspect ratio of the photobioreactors [6]. In one study, hydrogen production rates were measured using Roux bottles of varying depths. Hydrogen production was observed mainly at bottles with depths smaller than 1.5 cm [7]. It should be kept in mind, however, that high surface-to-volume ratios will increase the required land area, and may complicate the reactor construction and operation. Therefore a total thickness of 3 cm would be a proper compromise between good light distribution and limited surface area. Anaerobic conditions and limited nitrogen are required during photofermentative hydrogen production [1]. Therefore, another factor to consider is gas and more specifically hydrogen, permeability through the reactor material. Especially in large-scale systems, hydrogen loss through the reactor walls, manifolds and hydrogen gas collectors should be reduced as much as possible due to safety and economic considerations. Since a large surface area is inevitable due to light penetration requirements, it is necessary to decrease the permeability of the material as much as possible. The hydrogen permeability of the candidate PBR materials was assessed and experimentally verified in a previous study. The results of this study were shown in Table 1 [8]. Among these materials, glass was found as the least and LDPE the most permeable material, thus indicating glass is the best choice for the reactor material based on permeability alone. Apart from permeability, however, glass is also more advantageous in terms of mechanical strength and durability; it has been reported that the LDPE reactor tubes used in a previous study [2] had to be changed every
Table 1 e The hydrogen gas permeabilities of the candidate PBR materials. Material Poly (methyl methacrylate) (PMMA) Glass Polyurethane (PU) Polyvinyl chloride-plasticized (PVCplasticized) Low-density polyethylene (LDPE)
Permeability mol/ (m.s.Pa) 1.88 4.38 2.78 4.86
1014 1016 1012 1011
2.30 1011
year and the contribution of these tubes to the operating cost was around 65% [9]. As the reactor volume increases, product-gas collection also becomes a problem. It has been observed that as the reactor headspace pressure increased, hydrogen productivity decreased [10]. Therefore, it is important to maintain the total gas pressure in the headspace as low as possible, which will also promote lower retention of the produced gas, thereby decreasing the loss of hydrogen through the walls due to permeation as discussed above. In general, two reactor types have been preferred in literature for outdoor photofermentative hydrogen production: the panel type and the tubular type. The ratio of illuminated reactor surface to the occupied ground area for panel and tubular reactors was found as 8:1 and 1:1, respectively [11]. Despite this significant advantage in terms of required ground area, panel reactors have several drawbacks. Due to mechanical restrictions arising from the design, mixing becomes a problem in panel type photobioreactors whereas it can easily be achieved in tubular PBRs by a recirculation pump. There is also an upper limit for the volume of panel type photobioreactors; the mechanical integrity of the design begins to suffer beyond a certain size. In one study, severe swelling and deformation due to pressure was observed in a PVC-panel photobioreactor with 1 m 1 m dimensions [8]. Therefore, tubular PBRs also seem to be more advantageous in terms of durability. Mixing is promoted in photobioreactors in order to increase the mass transfer rate, to reduce nutrient gradients, to eliminate cell sedimentation and to facilitate the separation of the produced gas from the liquid culture. On the other hand, there appears to be an optimum mixing power value, above which productivity starts to decline. Though the reasons for the decrease of productivity at elevated mixing power values are unclear, one possibility is that shear stress, which depends on Reynolds number, induces cell damage [12]. In a study aimed to probe the effects of mixing, the performances of reactors shaken at 40, 80, 120 and 160 rpm were monitored and the average productivity was found to be the highest 5.3 mg H2/(L.h) when the shaking velocity was 120 rpm. This value was significantly higher than that of the control experiment with no shaking, reported as 3.72 mg H2/(L.h) [10]. In tubular photobioreactors, it has also been reported that hydrogen production is affected by the volumetric flow rate of recirculation that is used to promote mixing [11]. When the Reynolds number was varied from 10 to 6000, the highest productivity was observed for a Reynolds number of 240. No hydrogen production was observed when the volumetric flow
Please cite this article in press as: Kayahan E, et al., Design of an outdoor stacked e tubular reactor for biological hydrogen production, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.04.086
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e1 0
rate was larger than 2400. However, these experiments were carried out with a manifold type of photobioreactor where the flow distribution among the channels changed significantly with the Reynolds number. Consequently, the optimum Reynolds number for other systems should be obtained individually for other geometries. A flow model is especially useful in this regard, to obtain more insight into how the velocity distribution depends on the geometry. Finally, economics is still the main obstacle, preventing the application of such photofermentative hydrogen production systems in industrial scale. The large ground area required for a manifold type tubular photobioreactor by photofermentation was the reason of capital cost of photofermentation being several times larger than dark fermentation. Therefore, the ground area for the photobioreactor should be reduced [9]. The use of cooling water to keep the reactor temperature below 40 C is another factor that contributes to the hydrogen production cost. Another alternative for temperature control is passive cooling by shading the reactor. However, in a previous small scale outdoor study [5], the use of shading was not enough by itself to keep the reactor temperature at the desired values. Furthermore, as there is one to one correspondence between solar intensity and hydrogen production [4,13], decreasing the irradiance via shading is ultimately detrimental for gas production. As a result, active cooling methods such as water circulation or spraying are generally preferred. Based on the factors discussed up to this point, it can be argued that tubular reactors offer superior characteristics with respect to ease of mixing. Thus, in this work, a pilot-scale manifold type glass stacked U-tube photobioreactor was designed, built and operated. An upright orientation was selected to ensure small ground area to volume ratio with a large illuminated surface area, and the reactor material was chosen as glass for durability and low permeability to hydrogen and air. In designing the tubular photobioreactor geometry, a hydrodynamic model was used to assess whether a low overall pressure drop and a uniform velocity distribution could be achieved throughout the reactor and the dimensions of the reactor was based on the results of the model. The reactor was then constructed and run with and without bacteria, the latter grown using a molasses solution as a complex substrate.
Materials and methods Modeling The flow distribution inside a pilot-scale manifold type glass stacked U-tube photobioreactor (Fig. 1) was determined by the utilization of the single phase laminar flow module in COMSOL 4.4. Water at 30 C was considered as the working fluid. The laminar flow interface is based on the NaviereStokes equations. In the model, steady-flow was assumed for a Newtonian, incompressible and isothermal liquid medium. With these assumptions, the NaviereStokes equations for mass and momentum balances reduce to Equations (1) and (2), respectively.
3
rV,u ¼ 0
(1)
rðu,VÞu ¼ V PI þ mðVu þ ðVuÞT þ rg
(2)
where r is the density (kg/m3), u is the velocity (m/s), P is the pressure (Pa), I is the identity tensor, m is viscosity (Pa.s) and g is the gravitational acceleration constant in (m/s2). The no-slip boundary condition was applied to all the solid boundaries. At the inlet, the fluid velocity was fixed as shown by Equations (4) and (5). As it can be seen from Equation (4), the flow was assumed to be laminar and fully developed. At the outlet, the gauge pressure was zero, as shown in Equation (6). uwall ¼ 0
(3)
r2 uinlet ¼ 2$uave $ 1 2 R
(4)
uave ¼ u0 $n
(5)
Pout ¼ 0
(6)
where uwall is the velocity at the tube wall (m/s), uinlet is the inlet velocity (m/s), uave is the average velocity (m/s), R is the pipe radius and r is the distance from the center, u0 is the velocity magnitude at the inlet (m/s), n is the normal unit vector pointing out of the domain, and Pout is the outlet gauge pressure (Pa). COMSOL utilizes the finite element method while solving partial differential equations. GMRES (generalized minimum residual method) was used as the solution technique in the model. The tolerance factor was set to 0.001. The model was solved with a mesh number of 609805. A mesh convergence study was carried out by comparing the velocities in all the tubes separately. The total difference was found to be less than 5% when the mesh number was increased from 609805 to 2600023.
Pilot-scale stacked U-tube photobioreactor The photobioreactor (10.4 L) was designed by our group and the glass pieces were custom-built. The diameter of the manifolds and the tubes were 6 cm and 3 cm, respectively. The length of the manifolds and tubes were 0.475 m and 4 m, respectively. The wall-thickness of the glass tubes and the manifolds were 1.5 and 2.2 mm, respectively. All glass pieces were built from Schott borosilicate glass, reported to transmit about 90% of the light within the 350e2000 nm wavelength range (The complete optical properties can be accessed at www.schott.com). The gas was collected from the top of the manifolds by water displacement. Cooling was implemented via circulation of chilled water through spiral glass cooling coils within the manifolds. Moreover, PVC tubes (7 mm in diameter) were inserted into the glass tubes as an additional (and optional) circulatory cooling system, to be activated in case the cooling effect provided by the manifolds alone proved to be inadequate (not shown in process flow diagram). Twelve temperature measurement ports were placed on the tubes as shown in the process flow diagram in Fig. 1.
Please cite this article in press as: Kayahan E, et al., Design of an outdoor stacked e tubular reactor for biological hydrogen production, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.04.086
4
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e1 0
Fig. 1 e Process flow diagram of the stacked U-tube PBR. T1 e T12 are the temperature probes. V1 and V2 are check valves (1/ 3 psi), V4, V5, V6 and V7 are ball valves. CW-in and CW-out are cooling water inlet and outlet, respectively.
Two additional temperature measurement locations were designated on the manifolds. A centrifugal pump was used for recirculation of the liquid culture. The flow rate was adjusted using valves, V6 and V7. A recycle around the pump was necessary for flow rate adjustments. During feeding, recirculation was stopped by closing the valve, V4. Feed was given to the reactor by feed line (V5 is open) and the same amount of content was taken out of the reactor from the discharge line (V3 is open). For ease of gas collection, the tubes were placed with 30 inclination as can be seen from the photograph of the stacked U-tube PBR shown in Fig. 2.
Photobioreactor startup and operation The R. capsulatus YO3 (Hup) strain, which was modified by deleting the gene coding for uptake hydrogenase enzyme (Hup) of the original R. capsulatus MT1131 strain, was used in this study [14]. The bacteria were activated in the Biebl and Pfennig (1981) medium containing 20 mM acetate and 10 mM
glutamate as the carbon and nitrogen sources, respectively. A 22 mM potassium phosphate buffer was used, and the pH was adjusted to 6.4e6.5 by a 5 mM NaOH solution. The bacteria were grown under continuous illumination with 2000 lux and transferred to a second medium for sucrose adaptation, which contained 20 mM acetate, 5 mM sucrose and 10 mM glutamate besides the Biebl and Pfennig (1981) medium. Finally, the bacteria were inoculated to the biological hydrogen production medium, which contained 5 mM sucrose from molasses (obtained from the Ankara Sugar Factory, Ankara, Turkey), and then transferred to the stacked U-tube PBR. 30 mM KH2PO4 solution was used as buffer for the reactor medium and the initial pH was adjusted to 7.5. All inoculations (10% v/ v) were made when the concentration of the bacteria reached around 1 g/L, and argon was flushed to obtain anaerobic conditions. The reactor was sterilized by recirculating a 3% H2O2 solution for 24 h, then washed twice with distilled water. The recirculation rate was set as 43.5 L/h (Re ¼ 160). At the start-up, 10 mL samples were withdrawn from each tube for analysis, and 40 mL of 5 mM sucrose-containing
Fig. 2 e A photograph of the stacked U-tube photobioreactor. Please cite this article in press as: Kayahan E, et al., Design of an outdoor stacked e tubular reactor for biological hydrogen production, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.04.086
5
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e1 0
molasses with 30 mM KH2PO4 buffer solution (pH ¼ 7.5) was fed to the reactor to maintain the original volume (Phase I). After the 4th day, 1 L feed was given to the reactor daily, and 1 L of the reactor content was discharged at the same time. Sucrose content in the feed was changed many times according to changes in pH, cell concentration, organic acid and sucrose concentration. The feeding strategy is summarized in Table 2.
column. 100 mL samples, injected with gas-tight syringes (Hamilton, 22 GA 500 mL) were analyzed by a thermal conductivity detector. Argon with a flow rate of 26 mL/min was used as the carrier gas. The oven, injector and detector temperatures were 140, 160 and 170 C, respectively.
Results and discussion
Analytical methods
Model results
The bacterial cell concentration was measured by a spectrophotometer (Shimadzu UV-1201) at 660 nm. An optical density of 1.0 measured at 660 nm corresponds to 0.47 gdcw/ Lc for R. capsulatus Hup. pH was measured with a Mettler Toledo 3311 transmitter. Averages of 3 measurements taken from each sampling port were reported for bacterial cell concentration and pH. Periodic temperature measurements were taken from the outdoor reactor. Online temperature measurements were obtained from Fe-constant J-type thermocouples connected to a data logger (Ordel UDL100). Air temperature and cooling water inlet and outlet temperatures were also recorded. A pyronometer (HOBO-S-LIBM003), connected to an online weather station (HOBO® U30 ETH), was used for solar radiation measurements. Organic acid analyses were carried out using a High Performance Liquid Chromatography system (Shimadzu 20A series) equipped with an Alltech IOA-1000 column (300 mm 7.8 mm). The mobile phase (0.0085 M H2SO4 solution) was pumped with a low gradient pump (Shimatzu LC-10AT) with a flow rate of 0.4 mL/min. The oven temperature was fixed at 66 C. 10 mL samples were analyzed with a UV detector (Shimadzu FCV-10AT) whose absorbance was set at 210 nm [8,13]. The measured organic acids were lactic acid, formic acid, acetic acid, propionic acid and butyric acid. The sucrose content was analyzed by a commercial kit based on glucoseeoxidase reactions (Biyozim kit, Biasis, Turkey). The kit determines the sucrose content first by hydrolyzing sucrose to fructose and glucose and then reacting the latter to form a red-colored immunoquinone solution. The concentration of immunoquinone can be determined at 505 nm by a spectrophotometer (Shimadzu UV-1201), and in turn used for back-calculation of sucrose. The calibration curves were determined using standard sucrose solutions and the sucrose concentrations in the samples were determined accordingly. The gas composition was determined by a gas chromatograph (Agilent Technologies 6890N) equipped with a Supelco Carboxen 1010
Within the scope of this work, the effect of tube length, tube pitch (i.e., the spacing between the parallel tubes measured from the center of one tube to the center of the other one), and volumetric flow rate on the flow distribution were studied. For convenience, the results were interpreted by introducing the fractional volumetric flow rate, defined for the ith tube as shown in Equation (7). If the flow distribution is uniform among the tubes, then bi would be 0.25 for all the tubes for a manifold with 4 channels.
Table 2 e Feeding strategy. Phase I II III IV V VI VII
Sucrose concentration in molasses (feed) 5 mM 5 mM 100 mM e 50 mM e 50 mM
pH Feed rate (feed) 7.5 7.5 7.5 e 7.5 e 11e12
40 mL/day 1 L/day 1 L/day No feeding 1 L/day No feeding 1 L/day
bi ¼
Qi Q0
(7)
where Qi and Q0 represents the volumetric flow rate (m3/s) in the ith tube and total volumetric flow rate. The flow non-uniformity parameter (F), shown in Equation (8) was defined to quantify the flow distribution. This parameter is analogous to standard deviation and could be defined as the fractional volumetric flow rate when the flow is distributed equally to all channels [15]. sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi P4 2 i¼1 ðbi 1=nÞ $n$100 F¼ n
(8)
where n denotes the number of tubes, bi is the fractional volumetric flow rate in the ith tube. The summary of the design parameters taken into consideration and the corresponding non-uniformity parameters estimated from the model are listed in Table 3. Fig. 3 illustrates the variation of fractional volumetric flow rate (bi) for a 4 channel stacked tubular PBR for three different Table 3 e The summary of the design parameters studied and the corresponding non-uniformity parameter estimated from Equation (8). Tube length (m) 1.4 2.0 3.8 1.4 1.4 1.4 1.4 1.4 1.4 1.4 1.4 1.4 1.4 1.4
Tube pitch Overall volumetric flow Ret F (cm) rate (L/h) (%) 10.5 10.5 10.5 8.0 13.0 10.5 10.5 10.5 10.5 10.5 10.5 10.5 10.5 10.5
25.0 25.0 25.0 25.0 25.0 50.0 75.0 100.0 125.0 150.0 175.0 200.0 225.0 250.0
92.0 92.0 92.0 92.0 92.0 184.0 276.0 368.0 460.0 552.0 644.0 736.0 828.0 920.0
2.1 2.1 1.5 2.1 3.1 2.2 2.2 2.2 2.3 2.3 2.3 2.3 2.3 2.3
Please cite this article in press as: Kayahan E, et al., Design of an outdoor stacked e tubular reactor for biological hydrogen production, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.04.086
6
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e1 0
Fig. 3 e Effect of tube length on the fractional volumetric flow rate (bi) for a total volumetric flow rate (Q0) of 25.0 L/h and a tube pitch of 10.5 cm. Tube 1 is the closest to the ground.
tube lengths 1.4 m, 2.0 m and 3.8 m, while maintaining the volumetric flow rate at 25.0 L/h, and the tube pitch as 10.5 cm. Tube channels were counted from the bottom (closest to the ground is Tube 1) to the top (Tube 4). The flow non-uniformity parameter was calculated as 2.1% for both 1.4 m and 2.0 m long tubes, but found to be 1.5% for 3.8 m long tubes, indicating more uniform flow distribution in the latter. As the tube length is increased, resistance to flow increases in all of the tubes. The flow distribution is considerably affected by the tube length when the frictional forces in the tubes are larger than the frictional forces in the manifolds. This behavior was also noted in a recent study; although there, the flow resistance due to channel length was found to be small compared to the flow resistance in the manifolds, thus reducing the effect of tube length [16]. Usually, tube length is increased to scale-up tubular PBRs, therefore, it may be useful to further increase the tube length, as long tubes result in high residence times; however this in turn may increase the likelihood of leakage and it is difficult to maintain such systems at the desired temperature. The pressure drop between the inlet and outlet was found as 0.30, 0.34 and 0.49 Pa for tube lengths of 1.4 m, 2.0 and 3.6 m, respectively, and thus can be considered as almost negligible. The second parameter investigated in the model was the tube pitch. Three tube pitches, 8.0, 10.5 and 13.0 cm, were studied while other parameters were kept constant. The volumetric flow rate and tube length were 25.0 L/h and 1.4 m, respectively. The most uniform flow distribution was obtained for a tube pitch of 10.5 cm (Fig. 4). As the tube pitch increases, the diameter to height ratio of the distribution and collection manifolds decreases in this geometry. The effect of this ratio on the flow distribution was previously studied for other manifold models [16,17]. The diameter to height ratio for the distribution manifold has previously been stated as the ‘main governing parameter’ in design, whereas the same ratio for the combination manifold was found as a controlling parameter [16]. This behavior could be easily interpreted via the effective forces in such manifolds. In general, for horizontal manifolds, the flow distribution is determined by the wall friction and momentum change. As the manifold diameter to height ratio decreases for the
Fig. 4 e Effect of tube pitch on the fractional volumetric flow rate (bi) for a total volumetric flow rate (Q0) of 25.0 L/h and a tube length of 1.4 m. Tube 1 is the closest to the ground.
distribution manifold, friction becomes the dominant force, whereas for larger diameter to length ratios, momentum effects are the dominant factor. In the collection manifold, both the momentum and friction effects work in the same direction, as the fluid moves towards the outlet [17,18]. In addition to these forces, for vertical manifolds, gravity also affects the flow distribution. Therefore, there is an optimum diameter to length ratio when these forces balance each other. In this work, the optimum tube pitch was found as 10.5 cm. The pressure drops were found as 0.30, 0.30 and 0.29 Pa for 8, 10.5 and 13 cm tube pitches respectively. The volumetric flow rate was increased from 25 L/h to 250 L/h with 25 L/h increments. For these cases, the tube pitch was 10.5 cm and tube length was 1.4 m. It is more convenient to report the volumetric flow rates in terms of Reynolds number (Equation (9)). The tube velocity used in the calculation of the Reynolds number is shown in Equation (10). Ret ¼
ut ¼
r$ut $Dt m
4$Q0 n$p$D2t
(9)
(10)
where r is the density (kg/m3), ut is the tube velocity if the flow was distributed to all 4 tubes uniformly, Dt is the tube diameter (m), m is the viscosity (Pa.s), Q0 is the total volumetric flow rate (m3/s) and n is the number of tubes (4 in our case). In this study, fractional volumetric flow rate was compared using the tube Reynolds number, Ret. It should be noted that Ret is not the actual Reynolds number in the tubes computed from the model, but rather a hypothetical number, that gives a sense about what would have been the Reynolds number in the tubes if the flow was distributed uniformly to all the tubes. Fig. 5 indicates that increasing the volumetric flow rate (Ret) did not change the flow distribution. More fluid tended to go from the lowest tube for all the volumetric flow rates. This result revealed that both gravitational and momentum effects were higher compared to frictional effects in the distribution manifold. Still, looking at the flow non-uniformity parameter changing between 2.1% and 2.3% for different flow rates, it could be concluded that a uniform flow distribution was achieved.
Please cite this article in press as: Kayahan E, et al., Design of an outdoor stacked e tubular reactor for biological hydrogen production, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.04.086
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e1 0
Fig. 5 e Effect of Reynolds number (Ret) on the fractional volumetric flow rate (bi) for a tube pitch of 10.5 and tube lengths of 1.4 m. Tube 1 is the closest to the ground.
7
tubes accumulates on top of the manifolds, while liquid flows downwards and is recirculated back to the tubes by means of a pump. Vertical manifolds also have another advantage; tubular reactors have been claimed to be disadvantageous compared to panel reactors due to their lower illuminated surface area to ground area ratio [11]. However, with the current design, the ratio of the illuminated surface area to ground area was increased by a factor of almost 5 compared to a fully horizontal design [2], with approximately 5 m2 (4.8 m2) of the illuminated surface area found to occupy 1 m2 ground area. Moreover, this ratio could be further increased by increasing number of tubes connected to the manifold.
Photobioreactor operation results The pressure difference between the inlet and outlet of the reactor with respect to Ret is plotted in Fig. 6. The pressure drop increased as Ret increased, as expected. The values were low; however, even for the highest volumetric flow rate (250 L/ h) for which the pressure difference between the inlet and outlet of the reactor was only 20.3 Pa. As can be seen in Table 3, the most uniform flow distribution (the non-uniformity parameter is 1.5%) is obtained for a tube length of 3.8 m, a tube pitch of 10.5 cm and a volumetric flow rate of 25 L/h. Accordingly, the tube length for the final design of the stacked U-tube photobioreactor was chosen as 4 m, and the tube pitch was chosen as 10.5 cm.
The reactor was initially tested without bacteria to verify that the cooling and circulation systems worked properly without leakage. The experiment with R. capsulatus YO3 was conducted between September 9 and September 29, 2014. The reactor was cooled intermittently, only on days 2, 3, 4 and 6 at noon time, to keep the culture temperature below 40 C. Temperature did not change significantly along the length of the reactor (Fig. 7a) and for different tubes (Fig. 7b). Therefore, it can be concluded that effective cooling was achieved even if the system was cooled just by using the manifolds. Even without cooling, the temperature variation according to tube position and length was not appreciable (Fig. 8). It can be
Construction of the reactor The stacked tubular photobioreactor was constructed, based on the results of the simulations discussed above, and the findings of previous studies. Fig. 2 displays a photograph of stacked U-tube photobioreactor. Uniform velocity distribution with small pressure drop was achieved with this design as validated by our model. Glass was chosen as the reactor material, since it has low hydrogen and air permeability compared to other candidate PBR materials as mentioned before. The tube diameter was selected as 3 cm [7], based on the results of the recent studies to achieve good light distribution. For ease of gaseliquid separation, the manifolds were placed vertically. With this design, gas formed within the
Fig. 6 e Pressure drop with respect to Reynolds number (Ret). The tube pitch is 10.5 cm and the tube length is 1.4 m.
Fig. 7 e Temperature variation with time on the 5th day of the experiment: (a) along the length of tube 4, at the inlet (T1), midpoint (T5) and exit (T9) of the tube; (b) at the exits of tube 4 (T9), tube 3 (T10) and tube 1 (T11). See Fig. 1 for the locations of the tubes and measurement probes. The arrows indicate the operation of the cooling system (↑: Cooling on, ↓: cooling off).
Please cite this article in press as: Kayahan E, et al., Design of an outdoor stacked e tubular reactor for biological hydrogen production, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.04.086
8
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e1 0
Fig. 8 e Temperature variation with time, with cooling switched off, on the 9th day of the experiment: (a) along the length of tube 4, at the inlet (T1), midpoint (T5) and exit (T9) of the tube; (b) at the exits of tube 4 (T9), tube 3 (T10), tube 2 (T11) and tube 1 (T12). See Fig. 1 for the locations of the tubes and measurement probes.
observed that the reactor temperature followed the same trend with the ambient air temperature, but was slightly higher. The daily temperature variation within the reactor is illustrated in Fig. 9. As expected, the liquid culture temperature was highly affected by the solar radiation. Since most of the time there was no cooling, the temperature of the reactor was mainly driven by solar radiation. We based the feeding strategy on our previous indoor studies. The highest hydrogen production was found with R. capsulatus YO3 with 5 mM sucrose containing molasses in a
batch photobioreactor [19]. However, due to the fact that the observed sucrose consumption rate of the present study was found to deviate considerably from indoor results, a more variable strategy had to be adopted based on the culture response in order to maintain the nutrients and reactor conditions close to optimal conditions. The ‘phases’ of this feeding strategy were given in Table 2 and are explained further below. The initial sucrose concentration of molasses was adjusted to 5 mM. The cell concentration stabilized at the 3rd day of the experiment (Phase I in Fig. 10) and the sucrose concentration (Fig. 11) decreased to very low values (around 0.5 mM). To increase the cell concentration, Phase II, involving continuous feeding of molasses was started on the 4th day. During Phase II, the organic acid content remained below 4 mM (Fig. 12). At the end of 10 days, cell concentration dropped to a low value (0.12 gdcw/L) due to the rapid depletion of carbon source, requiring a substantial addition of sucrose to maintain cell viability. Thus, in Phase III, 100 mM sucrose containing molasses were fed to the reactor. This led to a significant increase in the organic acid content, in turn, resulting in a decrease of pH (Fig. 12). On the 12th day, the bacteria finally reached optimal levels of cell concentration for hydrogen production (around 0.4 gdcw/L). A total of 454 mL H2 (corresponding to a productivity of 0.327 mg H2/(L.h)) was produced in this phase, on the 12th day of the experiment. The gas composition was 60.9% H2, with CO2 as the remainder. As the solubility of CO2 decreases at low pH values, its percentage was higher compared to previous photofermentative hydrogen production studies [20,21] where pH was kept close to its optimum value of 7 [22]. After the 12th day (Phase IV), hydrogen production ceased, possibly due to the sustained decrease in pH. It could be speculated that the metabolism of the bacteria had shifted to other modes following Phase III, based on the increase in lactic and acetic acid concentrations. To recover steady operation, feeding was stopped for two days, resulting in stabilized organic acid content in the reactor on the 13th and 14th days of the experiment. As a result, the cell concentration increased to its highest value (1.0 gdcw/L) on the 14th day of the experiment. Subsequently, one liter of 50 mM sucrose-containing molasses was fed to the reactor on the 15th day (Phase V). In the following day, as elevated levels
Fig. 9 e Comparison of the temperature variation in the reactor with that of the ambient air, and the daily solar radiation (T9 is measured at the exit port of tube 4). Please cite this article in press as: Kayahan E, et al., Design of an outdoor stacked e tubular reactor for biological hydrogen production, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.04.086
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e1 0
9
Fig. 10 e Daily cell concentration variation in the reactor.
whole too unfavorable for hydrogen production or growth, hence the run was terminated on the 20th day.
Conclusions
Fig. 11 e Daily sucrose concentration variation in the reactor.
of organic acids were observed again, feeding was stopped once more (Phase VI). However, this led to a decrease in cell concentration levels, and consequently, during the last phase (Phase VII), 1 L of 50 mM sucrose containing molasses was fed to the reactor daily. In the feed, pH was adjusted to 11e12 with addition of NaOH solution to counter the drop in pH. During this phase, all organic acid concentrations decreased; however, pH stabilized around 5.0. As cell concentration continued to decrease, the culture conditions were decided to be on the
A new stacked-tubular design was proposed in this study with the goal of addressing problems with scale up of PBRs for hydrogen production. A hydrodynamic model for this design was analyzed, and it was found that the most uniform flow distribution was obtained for a tube pitch of 10.5 cm, a tube length of 3.6 m and a volumetric flow rate of 25 L/h. The design was then constructed with dimensions based on the model results. In view of its lower long-term material costs and low permeability to hydrogen and air, glass was chosen as the reactor material. The illuminated surface area to ground area of the reactor was found as 5:1, a high value for tubular-type reactors [11]. To test the new reactor, an experiment was carried out using molasses as the carbon source and R. capsulatus YO3 (Hup) as the microorganism. During the experiment, the feeding strategy was varied with the purpose of finding the best feeding strategy for single stage biological hydrogen production. It can be argued that maintaining the sucrose content in the reactor around 5 mM, and keeping the acetic acid and lactic acid content below 40 mM could be a
Fig. 12 e Daily organic acid variation in the reactor. Please cite this article in press as: Kayahan E, et al., Design of an outdoor stacked e tubular reactor for biological hydrogen production, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.04.086
10
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e1 0
good strategy for photofermentative hydrogen production in outdoor conditions. Many of the indoor studies on biological hydrogen production are batch; whereas in outdoor studies, fed batch operation could be more advantageous. Ultimately, the choice of a proper feeding strategy is an issue that warrants further investigation for scale up.
Acknowledgments _ ¨ BITAK This work was part of the ongoing project funded by TU (114M436) and by the METU BAP project (BAP-07-02-2014-007538).
Nomenclature Symbol Dt F g I n n P Pout Q Qi Q0 R r Ret u uave uinlet ut uwall u0 bi m r
tube diameter, m non-uniformity parameter gravitational acceleration, m/s2 identity tensor number of tubes normal unit vector pointing out of the domain pressure, Pa outlet gauge pressure, Pa volumetric flow rate, m3/s volumetric flow rate in the ith tube, m3/s total volumetric flow rate, m3/s pipe radius, m distance from the center, m tube Reynolds number velocity, m/s average velocity, m/s inlet velocity, m/s tube velocity if the flow was distributed to all 4 tubes uniformly velocity at the tube wall, m/s velocity magnitude at the inlet, m/s dimensionless volumetric flow rate in the ith tube viscosity, Pa.s density, kg/m3
Note: Symbols with single and double underlines indicate vector and tensor quantities, respectively.
references € ¨ r E, Eroglu E, Yu¨cel M, Gu¨ndu¨z U. Applications [1] Eroglu I, Ozgu of photofermentative hydrogen production. In: Zannoni D, De Philippis R, editors. Microbial bioenergy: hydrogen production. Springer; 2014. p. 237e67. € ¨ r E, Yu¨cel M, Gu¨ndu¨z U, Eroglu I. “Biohydrogen [2] Boran E, Ozgu production by Rhodobacter capsulatus Hup mutant in pilot solar tubular photobioreactor. Int J Hydrogen Energy Nov. 2012;37(21):16437e45.
[3] Adessi A, Torzillo G, Baccetti E, De Philippis R. Sustained outdoor H2 production with Rhodopseudomonas palustris cultures in a 50 L tubular photobioreactor. Int J Hydrogen Energy 2012;37(10):8840e9. € [4] Avcıoglu SG, Ozgur E, Eroglu I, Yu¨cel M, Gu¨ndu¨z U. “Biohydrogen production in an outdoor panel photobioreactor on dark fermentation effluent of molasses. Int J Hydrogen Energy Aug. 2011;36(17):11360e8. € ¨ r E, Uyar B, Oztu € ¨ rk Y, Yu¨cel M, Gu¨ndu¨z U, Eroǧlu I. [5] Ozgu Biohydrogen production by Rhodobacter capsulatus on acetate at fluctuating temperatures. Resour Conserv Recycl Mar. 2010;54(5):310e4. € ¨ r E, Eroglu I, Gu¨ndu¨z U, Yu¨cel M. [6] Androga DD, Ozgu Photofermentative hydrogen production in outdoor conditions. In: Minic D, editor. Hydrogen energy e challenges and perspectives. InTech; 2012. [7] Nakada E, Asada Y, Arai T, Miyake J. Light penetration into cell suspensions of photosynthetic bacteria and relation to hydrogen production. J Ferment Bioeng 1995;80(1):53e7. [8] Avcioglu SG. Scale up of panel photobioreactors for hydrogen production by PNS bacteria [M.Sc. thesis]. Middle East Technical University; 2010. [9] Urbaniec K, Grabarczyk R. Hydrogen production from sugar beet molasses e a techno-economic study. J Clean Prod 2014;65:324e9. [10] Li X, Wang Y, Zhang S, Chu J, Zhang M, Huang M, et al. Effects of light/dark cycle, mixing pattern and partial pressure of H2 on biohydrogen production by Rhodobacter sphaeroides ZX5. Bioresour Technol 2011;102(2):1142e8. [11] Gebicki J, Modigell M, Schumacher M, Van Der Burg J, Roebroeck E. Comparison of two reactor concepts for anoxygenic H2 production by Rhodobacter capsulatus. J Clean Prod 2010;18(Suppl. 1):S36e42. [12] Flickinger MC. Upstream industrial biotechnology, 2 volume set. John Wiley & Sons; 2013. [13] Androga DD, Ozgur E, Gunduz U, Yucel M, Eroglu I. Factors affecting the longterm stability of biomass and hydrogen productivity in outdoor photofermentation. Int J Hydrogen Energy 2011;36(17):11369e78. € ¨ rk Y, Yu¨cel M, Daldal F, Mandaci S, Gu¨ndu¨z U, Tu¨rker L, [14] Oztu et al. Hydrogen production by using Rhodobacter capsulatus mutants with genetically modified electron transfer chains. Int J Hydrogen Energy 2006;31(11):1545e52. [15] Fan J, Shah LJ, Furbo S. Flow distribution in a solar collector panel with horizontally inclined absorber strips. Sol Energy 2007;81:1501e11. [16] Ahn H, Lee S, Shin S. Flow distribution in manifolds for low Reynolds number flow. KSME Int J 1998;12(1):87e95. [17] Wang J. Theory of flow distribution in manifolds. Chem Eng J 2011;168(3):1331e45. [18] Acrivos A, Babcock BD, Pigford RL. Flow distributions in manifolds. Chem Eng Sci 1959;10(1e2):112e24. ır E. Photobiological hydrogen production from sugar beet [19] Sag molasses [M.Sc. thesis]. Middle East Technical University; 2012. [20] Koku H, Eroǧlu I, Gu¨ndu¨z U, Yu¨cel M, Tu¨rker L. Kinetics of biological hydrogen production by the photosynthetic bacterium Rhodobacter sphaeroides O.U. 001. Int J Hydrogen Energy 2003;28(4):381e8. € ¨ r E, Yu¨cel M, Gu¨ndu¨z U, Eroglu I. Biohydrogen [21] Boran E, Ozgu production by Rhodobacter capsulatus in solar tubular photobioreactor on thick juice dark fermenter effluent. J Clean Prod 2012;31:150e7. [22] Koku H, Eroglu I, Gu¨ndu¨z U, Yu¨cel M, Tu¨rker L. Aspects of the metabolism of hydrogen production by Rhodobacter sphaeroides. Int J Hydrogen Energy 2002;27(11e12):1315e29.
Please cite this article in press as: Kayahan E, et al., Design of an outdoor stacked e tubular reactor for biological hydrogen production, International Journal of Hydrogen Energy (2016), http://dx.doi.org/10.1016/j.ijhydene.2016.04.086