Studies on the expansion characteristics of the granular bed present in EGSB bioreactors

Transcription

1 Stdies on the expansion characteristics of the granlar bed present in EGSB bioreactors Yong-Hong Li 1, 2 *, Yan-Ling He 1, Sh-Cheng Yang 1 and Chn-Jiang An 1 1 Department of Environmental Engineering, Xi an Jiaotong University, Xi an, 7149, P.R.China 2 Department of Environmental Engineering, Xi an University of Engineering Science and Technology, Xi an, 7148, P.R.China Abstract In this stdy, the expansion characteristics of an anaerobic granlar bed in EGSB reactors based on terminal settling velocity stdy of the granles and the Richardson-Zaki eqation (1954) have been investigated. The settling velocity stdy shows that the mean settling velocity of the granles is in accordance with the Allen formla becase the settling process falls within the intermediate flow regime range (1< <11) rather than in the laminar flow regime range; the bed expansion stdy based on the above conclsion is fond to be similar to that sggested by the Richardson-Zaki eqation. The validity of the settling velocity model is verified by sing the operating parameters of several fll-scale anaerobic reactors with an average relative error of 3.77%, and the bed expansion data obtained from two laboratory-scale reactors are sed to test the correlation with an average relative error of 2.64% and 4.57% respectively. Moreover, it cold be a theoretical method to propose a sitable vale of Vp dring different operation periods of an EGSB system. Keywords: UASB, EGSB, granlar bed expansion, mean settling velocity, modelling, Richardson-Zaki eqation Notation d p granle diameter by wet sieving (m) g gravitational acceleration (m s -2 ) H bed-expansion height (m) H initial bed height (m) n expansion index Reynolds nmber at terminal velocity mean settling velocity of the granles (m/h) t Vp pward liqid velocity (m/h) μ the viscosity of wastewater (Pa.s) ε preliminary bed voidage Ε bed voidage η bed expansion ρ the density of wastewater (kg m -3 ) ρ p the density of the granles (kg m -3 ) Introdction The expanded granlar sldge bed (EGSB) reactor, a novel variation on the pflow anaerobic sldge blanket reactor (UASB) concept in combination with the anaerobic flidised-bed reactor (AFBR), has recently experienced tremendos growth. Althogh UASB is still the predominant technology in se, EGSB-type processes are now gaining more poplarity driven by economics (Frankin, 21). Similar to UASB technology, the EGSB reactor relies on the self-immobilisation properties of micro-organisms and the * To whom all correspondence shold be addressed. +86 (29) ; fax: +86 (29) ; liyhxa@hotmail.com Received 19 Agst 25; accepted in revised form 1 May 26. development within the reactor of a granlar biomass with good settling properties. Compared with conventional UASB reactors (.5 to 2 m/h), the advantage of the EGSB system (> 4 m/h) is the significantly better contact between sldge and wastewater. So it is widely believed to be a potentially high-rate anaerobic reactor by experts from different contries (Seghezzo et al., 1998; Hlshoff et al., 24). The knowledge of the bed expansion plays an important role in the design and operation of an EGSB reactor, becase it is the key point to reach a compromise between bed expansion and sldge washot, and the stability and performance of the EGSB system wold be sensitive to the degree of expansion (Li et al., 22). Therefore, qantitative research on bed expansion in EGSB reactors is necessary, althogh few reports are fond in the literatre. In fact, the bed expansion is closely related to the settling characteristics of the sldge, hence the settling characteristics of the granles have to be discssed first. The mean settling velocity is generally sed to evalate the settling characteristics of the granles. Almost all previos work pblished sggests that the mean settling velocity of the granles is in accordance with Stokes law becase the settling process is in the laminar flow range (Hlshoff, 1989; H, 23; Field, 25), whilst others sggest that it is in accordance with the Allen formla becase the settling process falls within the intermediate flow range (Wang, 22). However, the settling characteristics of the granles have scarcely been discssed theoretically by the above workers. To the best of or knowledge, there has been no report yet on the qantitative relationship of the bed expansion in EGSB reactors, partly de to the lack of an approved method to determine the settling velocity and of the granles. The objective of this stdy is to focs on the bed expansion characteristics of EGSB reactors on the basis of settling velocity stdy of the granles. Available on website 555

2 Materials and methods Experimental The density and diameter of samples sldge were measred experimentally with the methods given elsewhere (Hlshoff, 1989). The sldge particles were seqentially separated by sing five stainless-steel sieves (.2,.45,.63,.9, and 2. mm). Here the mean diameter by wet sieving is considered as the average eqivalent diameter in the settling and bed-expansion processes. To determine the settling velocity of the sldge we sed a modified sedimentation balance which is a standard method developed by Paqes BV, Balk, The Netherlands. The settling rates are determined by sing a 2.5 m high settling colmn filled with tap water and are obtained from the evoltion of the weight of the settled sldge as a fnction of the sedimentation time (Hlshoff, 1989). Bed expansion experiments were condcted in two laboratory-scale EGSB reactors at 3 C, and synthetic wastewater prepared by portable water and glcose was sed to simlate the wastewater. Reactor 1 (R1) had a working volme of 6 l with a height of 2 m and diameter of.2 m, and the original bed height of the reactor was.445 m. Sldge Sample 1 was obtained from a fll-scale EGSB reactor (275 m 3 ) treating starch wastewater in Yi-shi, Shandong province, China. Reactor 2 (R2) had a working volme of 14.8 l with a height of 1.8 m and a diameter of.9 m, and original bed height of the reactor was.78 m. Sldge Sample 2 was obtained from an existing UASB reactor in or laboratory. The different heights of the sldge bed were recorded at the varios pward velocities (1, 2, 3, 4, 5, 6, 7, 8, 9, 1 m/h) throghot the experiments. Model development Terminal settling velocity model The settling velocity of the granles can be determined from the well-known force balance: t 4gd p p 3 In Eq. (1), the drag coefficient ξ is a fnction of : (2) Assming that the settling process of the granles is in the category of intermediate flow regime (.2< <5), ξ is recommended as follows (Perry and Green, 1997): (3) Then sbstitting Eq. (3) into Eq. (1), the settling velocity of granles calclated from Allen s law may be written as (in absence of the wall effects) (Li et al., 25): The vale of shold be verified to check p the rationality of the assmption afterwards: d t (5) p t d p p.714 Bed expansion model There are several papers dealing with sldge bed expansion, sch as EGSB (Kato et al. 1994), AFBR (Setiadi, 1995), (1) (4) biofilm reactors (Nicolella, 1999), UASB (Leitao, 24), and the anaerobic hybrid reactor (Saravanan, 25). Althogh those reactors are actally three-phase systems and the biogas prodced may have some inflence on the bed expansion behavior, it is sggested that those reactors shold still be regarded as classic two-phase (solid-liqid) systems, since the above researchers fond that the effect of the gas flow rate on the bed expansion is insignificant and the effect of biogas can be ignored withot introdcing an important error. On the other hand, it was also fond that the biogas has little effect on expansion behavior in or experiments. In conclsion, it is reasonable to describe the bed as a likely two-phase solid-liqid system. The bed expansion data are generally reported and compared with nmerical relationships in terms of bed voidage ε as a fnction of flid sperficial velocity; convenient as it is, this representation has the disadvantage, particlarly at high bed expansion. Hence, in or stdy these data will be generated by bed expansion of a granlar bed. In crrent stdies, we focs on the bed expansion present in EGSB reactors, as a fnction of pward liqid velocity, with little attention on what is going on in the bed itself, sch as gas prodction. The experimental evidence pblished p to now sggests that bed expansion characteristics of a biological flidised bed reactor can be well represented by the Richardson and Zaki correlation (Richardson and Zaki, 1954). Their relation is written as: n (6) t where the expansion index n was fond to be a fnction of only, whilst Richardson and Zaki recommended the following correlation: n = 4.65 < n = 4.4.2< <1 -.1 n = 4.4 1< <5 n = 2.4 5< Correspondingly assming that the settling process of the granles is in the category of intermediate flow regime, then the correlation for n is written as: n= (7) Sbstitting Eq. (7) into Eq. (6), the bed voidance can be written as: e t.1 The bed expansion η, which is defined as follows: H H H 1 The expanded bed height H data cold be easily transformed to bed voidage by sing the flow eqation (Zhang, 24): 1 1 (8) (9) (1) where: ε is sally from.4 to.45 for nearly spherical particles, here it is taken as.4. Sbstitting Eq. (8) into Eq. (1), the bed expansion can now be written as: 556 Available on website

3 e t.1 t.1 1 e.4 1% (11) Therefore, the bed expansion in EGSB reactors with different granlar sldge nder different liqid sperficial velocities can be investigated. Sbstitting Eq. (11) into Eq. (9), the expanded bed height nder different operating conditions can also be calclated: t.1 t.1.4 H H 1 e 1% 1 e Reslts and discssions (12) Firstly, the theoretical method to determine the settling velocity and of the granles is developed. Secondly, experimental data of the bed expansion presented in EGSB reactors and the application of the bed expansion model are presented. Testing the terminal settling velocity model of anaerobic granlar sldge Since the settling tests are condcted in a settling colmn filled with tap water, the viscosity of water at room temperatre (1-3 Pa.s) will be sed in the present calclations. Simlation analysis of the settling velocity model In general, the granlar diameter ranges between.5 and 5 mm, bt the mean diameter of sldge granles in fll-scale installations typically ranges between 1 and 3 mm. So the diameter of the granles ranges between.3 and 2.5 mm in or simlation analysis. On the other hand, larger sldge granles have low density whose vales range typically between 1 and 1 5 kg m -3 (Lens et al., 1998), so they are respectively 1 1, 1 2,1 3, 1 4, 1 5 kg m -3 in or calclation. Additionally, ρ is taken as 1 kg m -3. t (m/h) dp(mm) dp(mm) Figre 1 Calclated settling velocity and of the granles The calclated settling velocities and of the granles with different density verss mean diameter obtained from Eqs. (4) and (5) are shown in Fig. 1. Figre 1 shows that settling velocities are in the range of 4 ~ 14 m/h. For most sldge granles, the calclated settling velocities are above 1 m/h so that high hydralic loads can be applied withot considerable sldge washot in high-rate anaerobic reactors. On the other hand, for most sldge granles, the corresponding are from 1 to 11, as shown in Fig. 1. Briefly, it demonstrates theoretically that the settling process of the sldge is in the intermediate flow regime category while the mean settling velocity shold be calclated with Eq. (4). Verification of terminal settling velocity model In the present stdy, in order to verify the validity of the above model, settling velocities and of nine samples between calclated and measred vales are shown in Table 1 (Li et al., 25b; Hlshoff, 1989). TABLE 1 Measred settling velocities and Reynolds nmber at terminal velocity of nine samples Types of wastewater Granle diameter Density (kg m -3 ) Settling velocity Reynolds nmber (mm) (m h -1 ) Sample a Beet sgar factory b Sample c Sample d Sample e Distillery wastewater f Potato processing g Beet sgar factory 1 h Wastepaper plant i a. Starch wastewater (EGSB-granles granles in Zh-cheng, Shandong Province, China) b. Beet sgar factory (Siker Unie, Roosendaal, The Netherlands) c. Starch wastewater (EGSB-granles in Yi-shi, Shandong Province, China) d. Starch wastewater (UASB-granles in Yi-shi, Shandong Province, China) e. Starch wastewater (UASB-granles in Zh-cheng, Shandong Province, China) f. Distillery wastewater (Nedalco, Bergen opzoom, The Netherlands) g. Potato processing plant (Aviko, Steenderen, The Netherland) h. Beet sgar factory (Central Siker Maatschappij, Breda, The Netherlands) i. Wastepaper plant (Papierfabrick Roermond, The Netherlands) Available on website 557

4 bed expansion (%) bed expansion ( %) mm 1.5mm 2mm V p (m/h) Figre 2 The mathematical analysis between bed expansion and pward liqid velocity mm 3.5mm 4.mm V p (m/h) Figre 3 The mathematical analysis between bed expansion and Vp in a Biobed EGSB reactor Table 1 shows that the average relative errors of t and between calclated and experimental vales were 3.77% and 3.2% respectively. The error of the proposed model may be de to the wall effects, gas prodction dring settling processes and so on. It implies that those factors cold be disregarded when calclating the mean settling velocity of the granles. Bed expansion present in EGSB reactors Based on the aforementioned reslts and stdies, the bed expansion present in EGSB reactors can be stdied by Eq. (11). The viscosity of wastewater μ varies with the temperatre and it has been reported that the vale is approximately Pa.s at 3 C in EGSB reactors (EC Pireset al., 2; Mahmod et al., 23), ths we se the vale of Pa.s in this section for the prpose of consistency. Simlation analysis of the expansion model Here we select the pward liqid velocity from to 1 m/h in or simlation analysis. Moreover, while d p is 1 mm, 1.5 mm and 2. mm, the corresponding density of the granles is 1 3 kg m -3, 1 2 kg m -3, 1 1 kg m -3 respectively. Additionally, ρ is taken as 1 kg m -3. The bed expansion data nder different pward liqid velocities with different granle size calclated by Eq. (11) are shown in Fig. 2. The bed expansion with increasing velocity is observed in Fig. 2. The predicted bed expansion is very low when Vp V p (m/h) TABLE 2 Bed expansion vales calclated from the model compared with actal vales is lower than 2 m/h, and then the bed behaves as a static bed, e.g. those in UASB reactors (<2 m/h); the predicted η is nearly ~5% when Vp ranges from 2 to 6 m/h and the predicted η is nearly 5 ~1% when Vp ranges from 6 to1 m/h. It implies that the bed will begin to expand when Vp exceeds 2 m/h. Figre 2 also reveals that the predicted η will exceed 2% when Vp is above 4 m/h. It is consistent with the main characteristics of EGSB reactors (>4 m/h) (Seghezzoet al., 1998). In the design and start-p of AFBR, Marin sggested that the degree of bed expansion shold be controlled in the range of 2~4 % (Marin et al., 1999). If we accept his proposal here, one conclsion that can be drawn is that the vale of Vp necessary to maintain an expansion of 2~4% is in the range of 3~7 m/h (as shown in Fig. 2). These reslts are in good agreement with the pblished reslts of many previos stdies on EGSB reactors (Kato et al. 1994; Hw et al., 1998; Jeison et al., 1999). Verification of expansion model sing experimental data According to the methods mentioned above, the mean diameter of Sample 1 is mm, ρ p 1 4 kg m -3, and ρ kg m -3. The mean diameter of Sample 2 is.9696 mm, ρ p 1 45 kg m -3, and ρ kg m -3. Based on those physical parameters and the variations of the bed height, the comparison of the bed expansion between calclated and measred vales is shown in Table 2. As shown in Table 2, calclated vales agree qantitatively with the experimental data in both experiments, with an average relative error of 2.64% and 4.57% respectively. Case stdy Bed expansion (%) Reactor 1 Reactor 2 Calclated reslts Experimental reslts Calclated reslts Experimental reslts According to variable information available in the literatre, excellent granlar sldge can be maintained nder extreme conditions in Biobed EGSB reactors, the bed height of sch reactors often varies between 7 and 14 m. Moreover, the settling velocity of the granles in a Biobed EGSB reactor (Caldic Erport, The Netherlands) is in the range of 6~8 m/h, and the granles are not washed ot from the reactor even at a velocity of 15 m/h (Zotberg and de Been, 1997). Kato et al. (1994) pointed ot that well-formed granlar sldge 3-4 mm in diameter can be maintained in EGSB reactors nder high hydralic loading conditions, and it has also been reported that the mean diameter of the granles in the Biobed 558 Available on website

5 TABLE 3 Changes in mean granle diameter in an a fll-scale EGSB reactor Size distribtion 7/21/24 8/18/24 8/26/24 9/2/24 9/8/24 >.9 mm ~.9 mm ~.63 mm ~.45 mm Mean granle diameter/mm EGSB reactor increased from 1.5 to 3.5 mm (mean density 1 37 kg m -3 ) dring the one-year reactor operation (Gonzalez-Gil et al., 21). Hence d p is 3, 3.5, 4 mm and ρ p is 1 4 kg m -3 in or mathematical analysis, and the mean settling velocity is taken as 7 m/h, ρ 1 kg m -3. Bed expansion data calclated by Eq. (12) are shown in Fig. 3. As shown in Fig. 3, the calclated vales of the bed expansion are very low when Vp is lower than 5 m/h, while the bed behaves as a static bed; the bed begins to expand when Vp exceeds 5 m/h; the vale is approximately 2% when Vp is arond 1 m/h; the vale is merely abot 45% even when Vp is arond 15 m/h. Based on Fig. 3, even if the original bed height of the EGSB reactor at Caldic is 14 m when Vp is arond 1 m/h, the bedexpansion height is jst 16.8 m which is close to the water level of the reactor (17.3 m). That may be the reason why the reactor can operate nder sch extreme conditions (>1 m/h). Model application of the expansion model Data on the changes in mean granle diameter of a fll-scale EGSB reactor with a working volme of 275 m 3 (Yi-shi, Shandong province, China) dring different operation time are shown in Table 3. The mean density of the granles is 1 5 kg m -3. Other data for simlation analysis are as follows: ρ is 1 kg m -3, original bed height of the reactor is 5.4 m. The variations of the bed height dring different period predicted by Eq. (12) are shown in Fig. 4. As shown in Fig.4, the maximm bed height is arond 1.6 m when Vp is abot 4 m/h, which is close to the position of gas-solid-liqid (GSL) separator. Based on this figre, it can clearly be seen that the sitable Vp vale necessary to preserve the stability in the reactor is arond 4 m/h. Therefore the vale of Vp applied in the EGSB reactor was controlled at arond 4 m/h. Conclsions The model developed in this stdy is applicable to stdy the bed expansion in EGSB reactors, provided that appropriate vales sch as pward liqid velocities, physical parameters of the granles and wastewater (sch as distribtion of granle diameter, μ, ρ, ρ p ) are assigned. It can also be sed to propose a sitable Vp vale dring different operation periods of an EGSB reactor. From a practical perspective, the findings of the present research have important implications on the design and operation of sch reactors. However, it shold be noted that the impact of biogas prodction, the efficiency of the reactor, and other factors on bed expansion have not been considered. Therefore, they are not reflected in the proposed correlation. In short, frther research is needed on the possible effects of these parameters, as well as the statistical significance of the reslts. predicted bed-expansion heights(m) The following conclsions can also be drawn from this stdy: The sldge bed begins to expand when Vp exceeds 2 m/h; the bed expansion is above 2% when Vp exceeds 4 m/h (Fig. 2) In Figs. 2, 3 and 4 it is shown that the corresponding Vp is arond 3~7 m/h (d p range from 1 to 2 mm) and the corresponding Vp is 1 ~15 m/h (d p range from 3 to 4 mm) when the bed expansion in EGSB reactors is controlled in the range of 2~4%. Acknowledgements This stdy was financially spported by the China National 863 Program (22AA6119) and the Edcation Office Fondation of the Shanxi Province (China) (6JK287), and the comments of two anonymos reviewers were grateflly appreciated. References 7/21 8/18 8/26 9/2 9/ V p (m/h) Figre 4 Variations of bed height with pward liqid velocity dring different operation time FIELD J (25) Anaerobic granlar sldge bed technology [Online], (Accessed on September 2 th ) FRANKIN RJ (21) Fll scale experiences with anaerobic treatment of indstrial wastewater. Water Sci. Technol. 44 (8) 1-6. GONZALEZ-GIL G, LENS PNL, VAN AELST A, VAN AS H, VER- SPRILLE AI, and LETTINGA G (21) Clster strctre of anaerobic aggregates of an expanded granlar sldge bed reactor. Appl. Environ. Microbiol. 67 (8) HU JC (23) Theory and Technology of Anaerobic Biological Treatment. China Architectre Indstrial Pblishing Company. Beijing, China. HULSHOFF POL LW (1989) The Phenomenon of Granlar Anaerobic Sldge. Ph.D. Thesis, University of Wageningen, The Netherlands. HULSHOFF POL LW; DE CASTRO LOPES SI, LETTINGA G and LENS PNL (24) Anaerobic sldge granlation. Water Res. 38 (6) HWU CS, VAN LIER JB, LETTINGA GATZE (1998) Physicochemical and biological performance of expanded granlar sldge bed Available on website 559

6 reactors treating long-chain fatty acids. Process. Biochem. 33 (1) JEISON D and CHAMY R (1999) Comparison of the behavior of expanded granlar sldge bed (EGSB) and pflow anaerobic sldge blanket (UASB) reactors in dilte and concentrated wastewater treatment. Water Sci. Technol. 4 (8) KATO MT, FIELD JA, VERSTEEG P and LETTINGA G (1994) Feasibility of expanded granlar sldge bed reactors for the anaerobic treatment of low-strength solble wastewaters. Biotechnol. Bioeng LEITAO RC (24) Robstness of UASB Reactors Treating Sewage nder Tropical Conditions. Ph.D. Thesis, University of Wageningen, The Netherlands. LENS PNL, BOSCHVANDEN ML, HULSHOFF POL LW and LET- TINGA G (1998) Effect of staging on volatile fatty acid degradation in slphidogenic granlar sldge reactor. Water Res. 32 (4) LIU Y, XU HL, SHOW KY and TAY JH (22) Anaerobic granlation technology for wastewater treatment. World J. Microbiol. Biotechnol LIU YH, HE YL, LI YZ and YANG SC (25) The settling characteristics and terminal velocity of granlar sldge in UASB reactors. Acta Scientiae Circmstantiae. 25 (2) LIU YH, HE YL, LIU ZK (25b) Measrement and stdy of settling velocity of granlar sldge in anaerobic reactors. Indstrial Water & Wastewater (China) 36 (5) MARIN P, ALKALAY D, GUERRERO L, CHAMY R and SCHIAP- PACASSE MC (1999) Design and startp of anaerobic flidized bed reactor. Water Sci. Technol. 4 (8) MAHMOUD NIDAL, ZEEMAN G, GIJZEN H and LETTINGA G (23) Solids removal in pflow anaerobic reactors, a review. Bioresor. Technol. 9 (1) 1-9. NICOLELLA C, VAN LOOSDRECHT MCM, DI FELICE R and ROVATTI M (1999) Terminal settling velocity and bed expansion characteristics of biofilm-coated particles. Biotechnol. Bioeng. 62 (1) PERRY RH and GREEN DW (1997) Chemical Engineers Handbook (7 th edn.) McGraw-Hill, New York. PIRES EC, HANISCH WS and ANDRADE MAN (2) An original procedre for physical simlation of pflow anaerobic sldge blanket reactors. Bioprocess. Eng. 23 (4) RICHARDSON JF and ZAKI WN (1954) Sedimentation and flidization. Trans. Inst. Chem. Eng pp. SEGHEZZO L, ZEEMAN G, VAN LIER JB, HAMELERS HVM and LETTINGA G (1998) A review: the anaerobic treatment of sewage in UASB and EGSB reactors. Bioresor. Technol. 65 (3) SETIADI T (1995) Predicting the bed expansion of an anaerobic flidised-bed bioreactor. Water Sci. Technol. 31 (9) SARAVANAN V, SREEKRISHNAN T R (25) Hydrodynamic stdy of biogranles obtained from an anaerobic hybrid reactor. Biotechnol. Bioeng. 91 (6) WANG YC, ZUO JE and XIAO JH (22) A stdy on characteristics of anaerobic granlar sldge in an EGSB Reactor. China Biogas 2 (4) 3-7. ZHANG XH (24) Engineering of Water Polltion Control (2 nd edn.). China Metallrgy Indstrial Pblishing Company. Beijing, China. ZOUTBERG GR and DE BEEN PETER (1997) The Biobed EGSB (expanded granlar sldge bed) system covers shortcomings of the pflow anaerobic sldge blanket reactor in the chemical indstry. Water Sci. Technol. 35 (1) Available on website