Performance and Modeling of an Up-flow Anaerobic Sludge Blanket (UASB) Reactor for Treating High Salinity Wastewater from Heavy Oil Production

Transcription

1 Environment Protection China Petroleum Processing and Petrochemical Technology 2012,Vol. 14, No. 3, pp September 30, 2012 Performance and Modeling of an Up-flow Anaerobic Sludge Blanket (UASB) Reactor for Treating High Salinity Wastewater from Heavy Oil Production Liu Chunshuang; Zhao Dongfeng; Guo Yadong; Zhao Chaocheng (College of Chemical Engineering, China University of Petroleum, Qingdao Shandong ) Abstract: In this study, an up-flow anaerobic sludge blanket (UASB) reactor was applied to treat the high salinity wastewater from heavy oil production process. At a HRT of 24 h, the COD removal reached as high as 65.08% at an influent COD ranging from 350 to 640. An average of 74.33% oil reduction was also achieved in the UASB reactor at an initial oil concentration between 112 and 205. These results indicated that this heavy oil production related wastewater could be degraded efficiently in the UASB reactor. Granular sludge was formed in this reactor. In addition, two models, built on the back propagation neural network (BPNN) theory and linear regression techniques were developed for the simulation of the UASB system performance in the oily wastewater biodegradation. The average error of COD and oil removal was -0.65% and 0.84%, respectively. The results indicated that the models built on the BPNN theory were wellfitted to the detected data, and were able to simulate and predict the removal of COD and oil by the UASB reactor. Key words: up-flow anaerobic sludge blanket (UASB); high salinity; heavy oil produced wastewater; granule sludge; BPNN 1 Introduction A large amount of saline wastewater is commonly produced during crude oil dehydration process after oil extraction [1]. This wastewater usually has high salt and oil concentration, but is deficient in nutrients and contains dissolved refractory organic compounds such as polymers, surfactants, and suspended solids which lead to the same characteristics as water surface substances coexisting with oil [2]. It is difficult to treat such wastewater effectively by applying aerobic biological treatment only [3]. An approach towards appropriate technology for the treatment of high salinity wastewater from heavy oil production process has become an imperative task. The up-flow anaerobic sludge blanket (UASB) reactor has been widely used due to its simple design, easy maintenance, low operating cost, and flexibility to changing environmental conditions [4-6]. It has been applied in various industrial wastewater treatment processes, such as paper-pulp liquors [7], food processing wastewater [8], fiberboard manufacturing wastewater [9] and pharmaceutical wastewater [10]. However, no study was reported on the application of the UASB reactor for treating heavy oil production related wastewater. The objective of this study was to investigate the feasibility of using a UASB system to treat the high salinity wastewater originated from heavy oil production process. The influence of HRT on the performance of the UASB system was investigated. Furthermore, two models based on the BPNN and linear regression techniques were used to simulate the performance of the UASB reactor for treating the high salinity heavy oil production related wastewater. 2 Materials and Analytical Methods 2.1 Wastewater characteristics The heavy oil production related wastewater in this study was obtained from the Shengli Oilfield located in Shandong province, China. According to analysis, the heavy oil in this wastewater had a density of 1.01 g/cm 3, a wax Corresponding Author: Zhao Dongfeng, the director of Environment and Safety Technology Center of China University of Petroleum; zhaodf@vip.sina.com; Telephone:

2 Liu Chunshuang, et al. Performance and Modeling of UASB Reactor for Treating High Salinity Wastewater from Heavy Oil Production content of 2.1%, and a viscosity of mpa s. Furthermore, this wastewater composed of oil-in-water emulsion was characteristic of a high salt concentration with a BOD/ COD ratio of 0.18 and a salt concentration of 1.15% 1.46% after primary oil separation and floatation (Table 1). Table 1 Characteristics of heavy oil production related wastewater COD, Salinity, g/l Oil concentration, ph TN, TP, Reactor An expanded-bed reactor with a three-phase separator was used as the continuous-flow reactor in this study (Figure 1). The reactor with a total volume of 2.25 L and an effective volume of 1.4 L was equipped with a thermostat to maintain a temperature of 30±1 throughout the experiment for creating favorable anaerobic conditions. The influent stream was continuously provided via a peristaltic pump, and the effluent was discharged by a flooding weir. The seed sludge was obtained from the wastewater treatment facilities (secondary settling tank) of a heavy oil production related water treatment plant and a distant noodle factory in Qingdao, China. The initial concentration of seed sludge was previously determined to contain approximately mg of SS/L. Figure 1 Sketch of the UASB reactor used in this study 1 Medium tank; 2 Peristaltic pump; 3 Expanded-bed reactor; 4 Temperature controller; 5 Recycle pump; 6 Water trap; 7 Wet gas meter; 8 Effluent reservoir; A Water pipe 2.3 Analytical methods Raw wastewater and UASB effluent were collected at fixed sampling sites between 10:00 am and 11:00 am daily for the analysis of COD. Measurement of COD was performed according to the standard methods. The ph value of water samples was monitored daily by a portable ph meter (type sension1, made by HACH). During the experiment, the sludge samples were taken at the middle section of the reactor for the MLSS and MLVSS analyses. MLSS and MLVSS were also measured according to the standard methods. The oil content was measured according to the infrared spectrophtotometric (IR) method (by the national standard GB/T ) using an infrared photometric oil content analyzer (Model 510, made by the China Invent Instruments). The granular sludge samples were taken from the UASB reactor at the end of operation. The mineral composition of the granules was analyzed by an ICP instrument (type Dmax/2400). The appearance and shape of the granules were observed with a scanning electron microscope (type HITACHI-570). 3 Results and Discussion 3.1 Acclimation to the heavy oil production related wastewater After having been inoculated with the sludge, the reactor was initially fed with sewage wastewater. Then the sewage wastewater was replaced by a combination of sewage and wastewater originated from heavy oil production with the proportion of heavy oil production related wastewater increasing gradually in series, i. e.: 20% (w/v, i.e. wastewater/volume of influent), 40% (w/v), and 80% (w/v). Finally the reactor was loaded with 100% (w/v) heavy oil production related wastewater (Figure 2). It can be seen from Figure 2 that it took 75 days before the completion of the acclimation period. The UASB reactor performed well in the first 15 days when a 93.21% COD removal was achieved. Introducing 20% (w/v), 40% (w/v) and 80% (w/v) of the heavy oil production related wastewater resulted in a transient oscillation of COD removal rate at each transition phase. At the end of the feeding period after having been loaded with 100% (w/v) of heavy oil production related wastewater in the system, the COD removal efficiency of the UASB reactor dropped to 68.47%. 91

3 China Petroleum Processing and Petrochemical Technology 2012,14(3):90-95 Figure 2 COD removal efficiency during the cultivation period 3.2 Performance of the UASB reactor The UASB reactor was subjected to three successive step changes in organic loading rate (OLR) by means of decreasing the hydraulic retention time (HRT) (Figure 3). In the first 20 days with an average OLR of about 0.23 kg COD/(m 3 d) at a HRT of 48 h (stage 1), the average total COD removal rate was 65.08%. Then the OLR was increased to 0.46 kg of COD/(m 3 d) by decreasing the HRT to 24 h (stage 2), and the COD removal showed a transient oscillation and was then stabilized at 63.02%. The effluent ph value was between 6.5 and 7.0 during this period, while the influent ph value was in the range of Finally, when the OLR increased to 0.61 kg COD/(m 3 d) at a HRT of 15 h (stage 3), the COD concentration in the effluent remained at a high level and the COD removal was only 55.21%. The effluent ph value was further decreased to It can be seen that the COD removal and the effluent ph value decreased with a decreasing HRT. At a HRT of 15 h, the COD removal and ph value reached at a minimum of 55.21% and 6.5, respectively. The main reason for this phenomenon was that some bacteria did not have enough time to degrade the organic carbon when the HRT decreased. In all, a COD removal efficiency of 55% 65% was obtained during this process, which was similar to the 65% removal rate of ABR for treating the heavy oil production related wastewater [11]. The effluent volatile fatty acids (VFAs) were increased when the HRT decreased from 25 to 30, and 45, respectively (Figure4). The main reasons for this phenomenon are that the proper structure of the UASB reactor benefits the hydrolysis of non-soluble large molecular organics into soluble organics by exoenzyme, Figure 3 COD removal achieved by the UASB reactor for treating oily wastewater at HRT of 48 h (stage 1), 24 h (stage2), and 15 h (stage 3), respectively COD in; COD out; COD removal such as VFA, to achieve the COD removal. In addition, some of these VFAs could not be utilized timely by microorganisms in the reactor when the HRT decreased. Furthermore, at a salt concentration ranging from 1.15% to 1.46% in this study, the microorganisms also could survive and maintain a suitable COD removal rate. This result was in agreement with the relevant literature information. Ji, et al. [1] and Dalmacija [12] found out that the acclimated granules showed adaptability to a salt concentration of less than 1.5% in an active sludge reactor. Figure 4 VFAs of the UASB reactor for treating oily wastewater at a HRT of 48 h (stage 1), 24 h (stage2) and 15 h (stage 3), respectively VFAs in; VFAs out The oil removal rate, influent oil concentration and effluent oil concentration are illustrated in Figure 5. The influent oil concentration fluctuated between and during this study. However, the oil removal rate was maintained at a high level of about 70% 80% regardless of sharp fluctuations in influent oil content, showing its tolerance of shock loading. The decreasing 92

4 Liu Chunshuang, et al. Performance and Modeling of UASB Reactor for Treating High Salinity Wastewater from Heavy Oil Production HRT had little impact on the oil removal rate. Furthermore, Figure 5 also demonstrates that on the 88 th day, the 106 th day and the 128 th day the effluent oil concentrations kept at a stable level while the influent oil concentration was considerably changed and the oil removal rate showed peaks, indicating that the UASB reactor not only survived a strong impact load capacity for the COD concentration but also had a potential resistance to a shockingly high loading of some special form of organic matters, such as oil in this study. During this research, the average oil removal rate was 74.33%. Figure 6 The SEM micrographs of granular sludge in the UASB reactor Figure 5 Oil removal rate in the UASB reactor for treating oily wastewater at HRT of 48 h (stage 1), 24 h (stage2), and 15 h (stage 3), respectively Oil in; Oil out; Oil removal 3.3 Granular sludge Granular sludge could be observed in the UASB reactor during the experiments. After 135 days of operation, the granules showed a compact structure and naturally occurring cavities on the surface, as illustrated in Figure 6. Different bacterial morphologies were observed including bacilli, cocci and filaments. MLSS increased from its initial value of to at the end of run (Figure 7). The majority of granular sludge had the color of wine and black with a slightly gray color in the reactor, which agreed well with the results of Shen [13]. The possible reasons are that the granules containing a large amount of photosynthetic bacteria composed of mainly red Rhodopseudomonas with a high ability to degrade hydrocarbons (HCs) and achieve obligate halotolerance [14]. Based on the ICP analysis, the main metals in the granular sludge are shown in Figure 8. It can be seen that Al, Ca, Fe and Mg contents were high. During the granular Figure 7 Biomass in the UASB reactor during the operation process Figure 8 Analysis of metal elements in the sludge formation process, cations such as Ca 2+ and Mg 2+ in the oil production related wastewater could neutralize the negative charges on the surface of bacteria and promote the cohesion of bacteria. Furthermore, Ca 2+ on the surface 93

5 China Petroleum Processing and Petrochemical Technology 2012,14(3):90-95 of bacteria could be converted into CaCO 3 crystals, which increased the proportion and mechanical strength of granule sludge, resulting in compactness of sludge with higher ability to resist destruction [15]. 3.3 BPNN modeling Two arithmetic models were established by the software program MATLAB to simulate the performance of the UASB reactor in terms of effluent COD and effluent oil concentration by taking into account several operating parameters, including the ph value, the COD loading rate, the HRT, the influent COD, and the oil concentration. Models were based on the BPNN theory and linear regression techniques. The topological architecture of BPNN models are illustrated in Figure 9, which shows two three-level networks. Each network consists of four nodes in input layers, undetermined nodes in hidden layers and one node in the output. Each node is a BP neuron. The influent COD (oil), COD loading rate, HRT and ph value of the UASB reactor are the initial variables of input for each network. The effluent COD concentration and oil concentration are the variables. Functions and parameters used in models are described as follows: training function, trainscg arithmetic; stimulative functions, tansig (hidden layer) and purelin (output layer); study rate (lr), 0.8; and other parameters were defaults. The model program for COD and oil removal selected the data on day 76 day105 of the UASB system as training value and was subsequently simulated by using independent data sets between day 106 and day 135. Experiments were debugged and perfected through adjusting node numbers of the hidden layers, which represented unobserved-state variables. The results of interactive programming showed that when the nodes of hidden layer were 9, it was able to well-predict the performance of COD removal in the UASB reactor. And the nodes of hidden layer were 7 for oil removal (Figure 10). Figure 10 Comparison of experimental values and BPNN values of (a) COD removal and (b) oil removal experimental values; BPNN values; deviation Figure 9 Topological architecture of the BPNN model The model fitted the experimental data in the training period well and predicted the data in the subsequent period with an average error equating to -0.65% for COD removal and 0.84% for oil removal, respectively, which indicated that the simulation model built on the BPNN theory is a practical and feasible means to simulate and predict the pollutants removal by the UASB system. In a previous study, Hanbay, et al. [16] based on the wavelet packet decomposition, entropy and BPNN theory, developed a model to predict the WWTP performance. The suitable architecture of the BPNN model was determined after several trial-and-error steps. According to test re- 94

6 Liu Chunshuang, et al. Performance and Modeling of UASB Reactor for Treating High Salinity Wastewater from Heavy Oil Production sults, the performance of the developed model was at the desirable level [16]. 3 Conclusions (1) A UASB system was demonstrated for the treatment of oil production related wastewater. The system operated at an influent COD fluctuating between 350 and 640, with the COD removal reaching as high as 65.08% at a HRT of 24 h. An average of 74.33% oil reduction was also achieved in the UASB reactor at an initial oil concentration of These results indicated that this heavy oil production related wastewater could be degraded efficiently in the UASB reactor. (2) Granular sludge was formed in this reactor and cations such as Ca 2+ and Mg 2+ played an important role in the sludge formation process. (3) Two models, built on the BPNN theory and linear regression techniques, were developed for the simulation of the UASB system performance in the biodegradation of oil production related wastewater. The models could well fit the detected data, and were able to simulate the removal of COD and oil. It proved that the simulation model built on the BPNN theory is a feasible and practical means to simulate and predict the removal of organic substances by the UASB system. Acknowledgments: The authors gratefully acknowledge the support provided by the Research & Technology Development Project of China National Petroleum Corporation (06A0302), Postdoctor Innovation Funds in Shandong Province ( ) and the Fundamental Research Funds for the Central Universities (27R A). References [1] Ji G D, Sun T H, Ni J R. Surface flow constructed wetland for heavy oil production related water treatment [J]. Bioresource Technology, 2007, 98 (2): [2] Ji G D, Sun T H, Zhou Q X, et al. Constructed subsurface flow wetland for treating heavy oil production related water of the Liaohe Oilfield in China [J]. Ecological Engineering, 2002, 18 (4): (in Chinese) [3] Lai S G. Study on biological treatment test of sewage in the condensed oil [J]. Petrochemical Industry Technology, 2003, 10 (2): [4] Vadlani P V, Ramachandran K B. Evaluation of UASB reactor performance during start-up operation using synthetic mixed-acid waste[j]. Bioresource Technology, 2008, 99: [5] Lew B, Tarre S, Belavski M, et al. UASB reactor for domestic wastewater treatment at low temperatures: A comparison between a classical UASB and hybrid UASB-filter reactor[j]. Water Science & Technology, 2004, 49: [6] Lettinga G, Hulshoff Pol L. Advanced reactor design, operation and economy[j]. Water Science & Technology, 1986,18: [7] Ahn J H, Forster C F. A comparison of mesophilic and thermophilic anaerobic upflow filters treating paper-pulpliquors [J]. Process Biochemistry, 2002, 38: [8] Oliv L, Zaiat M, Foresti E. Anaerobic reactors for food processing wastewater treatment: Established technology and new developments [J]. Water Science & Technology, 1995, 32: [9] Fernández J M, Omil F, Méndez R, et al. Anaerobic treatment of fibreboard manufacturing wastewaters in a pilot scale hybrid USBF reactor[j]. Water Research, 2001, 35: [10] Martínez J, Borzacconi L, Mallo M, et al. Treatment of slaughterhouse wastewater [J]. Water Science & Technology, 1995, 32(12), [11] Ji G D, Sun T H, Ni J R, et al. Anaerobic baffled reactor (ABR) for treating heavy oil produced water with high concentrations of salt and poor nutrient[j]. Bioresource Technology, 2009, 100: [12] Dalmacija, B. Purification of high-salinity wastewater by activated sludge process [J]. Water Research, 1996, 30 (2): [13] Shen Y L. Micro-ecological characteristics of granule sludge in anaerobic baffled reactor (ABR) [J]. China Biogas, 2005, 23 (1): [14] Li A M, Tian S Y. Research on processing organic waste water with fixed PSB in anaerobic baffle reactor [J]. Journal of Tianjin Normal University, 1994, 14 (3): (In Chinese) [15] Heijnen J J. Development and scale-up of an aerobic biofilm air-lift suspension reactor[j]. Water Science & Technology, 1993, 27: [16] Hanbay D, Turkoglu I, Demir Y. Prediction of wastewater treatment plant performance based on wavelet packet decomposition and neural networks[j]. Expert Systems with Applications, 2008, 34: