Environmental Science
Processes & Impacts View Article Online
Published on 20 January 2014. Downloaded by University of Illinois at Chicago on 30/10/2014 23:13:22.
PAPER
Cite this: Environ. Sci.: Processes Impacts, 2014, 16, 1001
View Journal | View Issue
A demonstration of biofiltration for VOC removal in petrochemical industries Lan Zhao,*ab Shaobin Huangab and Zongmin Weia A biotrickling filter demo has been set up in a petrochemical factory in Sinopec Group for about 10 months with a maximum inlet gas flow rate of 3000 m3 h1. The purpose of this project is to assess the ability of the biotrickling filter to remove hardly biodegradable VOCs such as benzene, toluene and xylene which are recalcitrant and poorly water soluble and commonly found in petrochemical factories. Light-weight hollow ceramic balls (F 5–8 cm) were used as the packing media treated with large amounts of circulating water (2.4 m3 m2 h1) added with bacterial species. The controlled empty bed retention time (EBRT) of 240 s is a key parameter for reaching a removal efficiency of 95% for benzene, toluene, xylene, and 90% for total hydrocarbons. The demo has been successfully adopted and practically applied in waste air treatments in many petrochemical industries for about two years. The net inlet concentrations of benzene, toluene and xylene were varied from 0.5 to 3 g m3. The biofiltration process is highly efficient for the removal of hydrophobic and recalcitrant VOCs with various concentrations from the
Received 11th October 2013 Accepted 9th January 2014
petrochemical factories. The SEM analysis of the bacterial community in the BTF during VOC removal showed that Pseudomonas putida and Klebsiella sp. phylum were dominant and shutdown periods could
DOI: 10.1039/c3em00524k
play a role in forming the community structural differences and leading to the changes of removal
rsc.li/process-impacts
efficiencies.
Environmental impact This work presents a biotrickling lter demo set up in a petrochemical factory in Sinopec Group for about 10 months with a maximum inlet gas ow rate of 3000 m3 h1. The purpose of this project is to assess the capability of biotrickling lters for treatment of hardly biodegradable VOCs such as benzene, toluene and xylene which are recalcitrant and poorly water soluble VOCs in petrochemical factories. Lightweight hollow ceramic balls (F 5–8 cm) were adopted as packing media, combined with large volumes of circulating water (2.4 m3 m2 h1) with bacteria and also the controlled empty bed retention time (EBRT) of 240 s a removal efficiency of 95% (for benzene, toluene and xylene) and 90% for total hydrocarbon can be achieved. The demo has been successfully adopted and practically applied in waste air treatment in petrochemical industries for about two years.
Introduction Volatile organic compounds (VOCs) are compounds that are characterised by high vapor pressure and low water solubility. Many VOCs are human-made chemicals produced and used in the manufacturing of paints, pharmaceuticals, and refrigerants. VOCs are typically industrial solvents such as benzene, toluene and trichloroethylene, fuel oxygenates such as methyl tert-butyl ether (MTBE), and by-products such as chloroform produced by chlorination in water treatment. VOCs are also components of petroleum fuels, hydraulic uids, paint thinners, and dry cleaning agents. Many of these compounds are carcinogenic or mutagenic for human beings and animals. Also they are
a
College of Environmental Science and Engineering, South China University of Technology, Guangzhou, P. R. China. E-mail: [email protected]; Fax: +86-2039380508; Tel: +86-20-39380587
b
Guangdong Provincial Key Laboratory of Atmospheric Environment and Pollution Control, South China University of Technology, Guangzhou, P. R. China
This journal is © The Royal Society of Chemistry 2014
harmful to the environment. Therefore, there is an ongoing need for the development of better treatment methods and prevention technologies to reduce or eliminate VOCs in various environments. Currently, a number of technologies have been developed to eliminate VOCs. For instance, combustion, absorption, condensation, bioltration, UV/bioltration, and UV/TiO2/bioltration technologies are widely used. UV/TiO2 photooxidation (in the form of photocatalysis or direct photolysis) is a promising technology for the elimination of VOCs from contaminated gas streams. It involves the use of ultraviolet (UV) light to partially oxidize the pollutants through the production of ozone and radicals such as hydroxyl radicals, but does not transfer the contaminants from one phase to another.1 Besides, UV/TiO2 can be used to directly dissociate the compound through direct photolysis. These processes are capable of oxidizing a wide range of contaminants in a very short time period (usually a fraction of one second to seconds).2 A bioltration process combined with photocatalytic oxidation UV/TiO2 was developed
Environ. Sci.: Processes Impacts, 2014, 16, 1001–1007 | 1001
View Article Online
Published on 20 January 2014. Downloaded by University of Illinois at Chicago on 30/10/2014 23:13:22.
Environmental Science: Processes & Impacts
to degrade ethylbenzene vapors with higher operation cost by Hinojosa-Reyes et al.3 The greatest degradation rate of EB (0.414 ng m2 min1) was obtained with an inlet loading rate of 127 g m3 h1. The hybrid process enhanced the overall EC of ethylbenzene by 40%. Unfortunately, the photooxidation of many VOCs (e.g. aromatic and chlorinated compounds) is limited in its application due to its propensity to produce byproducts that might be more toxic than their parent compounds.4,5 However, these UV oxidation by-products are oen more biodegradable than their parent compounds as they are more hydrophilic, smaller, and/or simpler in structure.6 Bioltration treatment is the most attractive technology from economic and environmental perspectives due to its low operation costs and low energy requirements.7 A typical biolter consists of a lter bed usually comprised of natural materials (soil, compost, bamboo or peat) which is kept wet to maintain a biologically active layer surrounding the biolter materials, known as the “biolm”. Bioltration utilizes microorganisms inquilinous in the biolter media to degrade the pollutants in a waste gas stream purging into water and CO2. However, bioltration treatments can hardly degrade recalcitrant insoluble VOCs. Biotrickling lters are similar to biolters except that the packing medium is rigid or semi-rigid and inorganic. Also, a nutrient containing liquid phase is continuously recirculated through the system. Microorganisms grown on the packing medium may also be suspended in the recirculating liquid phase. Continuous or periodic addition/recirculation of liquid allows for control of nutrient concentrations and reaction conditions. One disadvantage of biotrickling lters in comparison to biolters is the lower specic surface area of the packing media, which makes it difficult to treat poorly water-soluble compounds. Clogging is another potential problem due to a readily available nutrient supply. Jianjun Li et al.8 investigated the performance of a eld-scale biotrickling lter (BTF) in the removal of waste gases containing low concentrations of mixed volatile organic compounds. Results showed that acetone and methyl ethyl ketone (MEK) were more easily removed than toluene and styrene. The removal efficiency for acetone and MEK reached over 99% on days 28 and 25 of the operation, whereas those for toluene and styrene were 80 and 63% on day 38. The maximum individual elimination capacities for styrene, toluene, acetone, and MEK were 10.2, 2.7, 4.7, and 8.4 g m3 h1, respectively. These values were achieved at inlet loading rates of 13.9, 3.3, 4.8, and 8.5 g m3 h1, respectively and at an empty bed retention time of 14 s. C. Lu et al.9,10 studied the effects of pH values, moisture rates, and ow patterns on the trickle-bed biolter performance for BTEX (benzene, toluene, ethylbenzene and xylene) removal. The BTEX removal efficiency increased as the pH values of the nutrient feed increased in the range of 5–8. In the pH range 7.5–8 and nutrient feeding rate (NFR) range 6.02–8.6 L m3 h1, removal efficiencies of each compound were greater than 80% with a loading of 143 g BTEX m3 h1. The BTEX removal efficiency increased as the NFR increased in the range 3.44–8.6 L m3 h1. For a 17.2 L m3 h1 NFR, the BTEX removal efficiency was slightly lower than those at 6.02 L m3 h1 NFR and 8.6 L m3 h1 NFR, but was signicantly higher than that at 3.44 L m3 h1 NFR.
1002 | Environ. Sci.: Processes Impacts, 2014, 16, 1001–1007
Paper
The optimum NFR was in the range of 6.02–8.6 L m3 h1. In the optimum pH and moisture ranges, the BTEX removal efficiencies were greater than 80%. As a result, biotrickling ltering appeared to be an effective process for BTEX removal with high loading. A counter-current ow biotrickling lter has the advantages of achieving more homogeneous BTEX removal and biomass growth in each section than the co-current ow system, but it suffered from a higher pressure drop across the bed. Jian M. Chen et al.11 also studied microbes bound to a medium in which wheat bran, red wood powder and diatomaceous earth carriers were used as inoculants for a biotrickling lter (BTF) for treating gases contaminated with a mixture of benzene, toluene and o-xylene (BTo-X). An overall removal efficiency of more than 87.9% was achieved aer a start-up period of as low as 4 days. At BTo-X loading rates (LRs) below 60.0 g m3 h1, the BTF performance was similar to those of EBRTs at 90, 60, 45 and 30 s with an elimination capacity (EC) approaching the LR; stable REs above 91.3% for benzene and toluene and above 82.8% for o-xylene were achieved. A maximum EC of 97.7 g m3 h1 was obtained at the inlet load of 146.4 g m3 h1. The mass ratio of carbon dioxide produced to the BTo-X removed was approximately 2.62, which conrmed the complete degradation of BTo-X. The results demonstrate that microbes bound to a solid carrier can be an alternative to traditional liquid inoculums applied in BTFs and highlight their potential applicability to BTF technologies. This study developed a low cost bioltration process to treat hydrophobic waste air from the petrochemical factories without integration of UV/TiO2 photolysis. It is an effective technology to remove recalcitrant air pollutants such as benzene, toluene, xylene, aromatics, etc. The effect of shutdowns on BTF performances was comprehensively investigated.
Methodology Demo set-up and procedure The demo process involving two series biotrickling lters was carried out (Fig. 1) for a period of ten months for VOC removal. The contaminated stream was generated using waste air from the petrochemical factory in Guangzhou, China, that delivered VOCs at a uctuating rate. In the pretreatment tank (F 0.94, 3 m height), the maximum ow rate is 3000 m3 h1 (1.2 m s1) and the spray ow rate is 5 m3 h1 with a 3.0 Pa pressure drop. Prints and oils were rst removed prior to the process to eliminate any interfering chemicals. Subsequently, temperatures were set to be lower than 40 C and the humidity was set to >90%. The contaminated stream from a compressor can then enter all the reactors from the top. The biotrickling lter column was made up of a stainless steel (SS) of 3.8 m diameter and 12 m height, which is featured with 2 min of empty bed retention time (EBRT) from Nantong Langgao Equipment Ltd. The empty bed ow rate was maintained at 0.067 m s1. Waste air can be opened and closed with a valve at the bottom. Four gas chromatography (GC) sampling ports (S1–S4 from Agilent 6750) were set up to make S1 as the inlet of petrochemical waste air, S2 as the pretreatment outlet, S3 as the I biotrickling lter outlet, and S4 as the II biotrickling lter outlet. The biotrickling
This journal is © The Royal Society of Chemistry 2014
View Article Online
Published on 20 January 2014. Downloaded by University of Illinois at Chicago on 30/10/2014 23:13:22.
Paper
Fig. 1
Environmental Science: Processes & Impacts
Demo set-up for the biofiltration of petrochemical waste air.
lter was maintained under ambient temperature. A modied recycling system with a 2.5–3 mm diameter screw nozzle and 30 m3 valid volume from Nantong Langgao Equipment Ltd. was designed to save 80% water energy cost. Recycling water was controlled at 2.4 ton h1 (2.4 m3 m2 h1) with a 500 Pa pressure drop. The temperature inside the biolter column was monitored with a wet bulb thermometer (LY-09, Qingdao Jingcheng Instrument Ltd.), and the pressure drop was monitored biweekly with a pressure meter (Comark, UK).
was calculated. The identication of isolated bacteria was performed using a Biolog identication system (Microlog-1, visually interpreted system) and GN-2 (Biolog) plates. All chemicals used were of analytical grade (SD Fine Chemicals, China). The bacterial community had been inoculated from the activated sludge drawn from Zhujiang wharf for about 6 months prior to the work. To develop biolms on the surface of packing materials, 17–48 L h1 of the activated sludge plus 450 L of nutrient solution were continuously recycled for 5 days during which only fresh air was supplied to the BTF before the startup. The results showed that Pseudomonas putida and Klebsiella sp. phylum were dominant microbes in the BTF, also the initial inoculums played an important role and entailed a lengthened startup period in the enhancement of BTF removal capacity due to the inorganic nature of the packing material. Moreover, the presence of a continuous water layer over the biolm reduces the removal performance for hydrophobic pollutants due to mass transfer limitations from the gas to the aqueous phase. The biolm was completely formed aer 25 days (Fig. 2). Fig. 3 shows the morphological changes of Pseudomonas putida phylum before and aer the treatment of VOCs for about 10 months (50 000 times magnication). Crystals were also induced in the agar plate by a mixed bacterial community (Fig. 4 and 5).
Packing medium Light weight hollow ceramic media have been used extensively in recent years because they are cheap and have diversied holes which can be used to adsorb various pollutants once they become acclimatized to the substrate. The volume of packing is 200 m3 in which three layers are supported by a F 60 mm sieve plate from Nantong Langgao Equipment Ltd.
Fig. 2
SEM photograph of the biofilm.
Microbial analysis of the biotrickling lter During the start up of the unit, the reactor was inoculated with respect to the medium in a closed loop to maximize the cell adhesion to the packing media for about 42 days. The composition of the medium included 1.0 g L1 KNO3, 10.48 g L1 Na2HPO4$12H2O, 1.50 g L1 KH2PO4, 0.05 g L1 MgSO4, 8.0 g L1 C4H4Na2O, and 1.0 mL L1 trace metal solution (50 g EDTA, 2.20 g ZnSO4, 5.50 g CaCl2, 5.06 g MnCl2$4H2O, 5.00 g FeSO4$7H2O, 1.10 g (NH4)6Mo7O2$4H2O, 1.57 g CuSO4$5H2O, and 1.61 g CoCl2$6H2O per liter; pH ¼ 7.0, from GZ Chemical Reagent Factory, Analytical Reagent). Petrochemical waste air was introduced into the unit with relatively uctuating ow rates (30–450 mg m3 benzene, 50– 1000 mg m3 toluene, and 140–1200 mg m3 xylene). The number of visible colonies formed aer 7 days was counted, and the average number of cells per gram lter bed This journal is © The Royal Society of Chemistry 2014
SEM of Pseudomonas putida before and after the treatment of VOCs (50 000 times).
Fig. 3
Environ. Sci.: Processes Impacts, 2014, 16, 1001–1007 | 1003
View Article Online
Published on 20 January 2014. Downloaded by University of Illinois at Chicago on 30/10/2014 23:13:22.
Environmental Science: Processes & Impacts
Fig. 4 Crystals induced in the agar plate by a mixed bacterial community.
Paper
recycling water volume was designed as 2.4 m3 m2 h1, corresponding to an optimistic empty bed gas residence time of 120 s as veried by the designer of the factory. The waste air inlet concentration was also varied in the range of 30–450 mg m3 benzene, 50–1000 mg m3 toluene, and 140–1200 mg m3 xylene, because the concentrations are typical values from the petroleum plant and are representative of commonly low, medium and high concentrations found in similar factories. The bioltration performance is discussed in terms of the VOC inlet load (IL, g m3 h1), the removal efficiency (X) and the elimination capacity (EC, g m3 h1) are evaluated using the following equations from the Handbook on Waste Gas Treatment Engineering Technology, Version 1: IL ¼
Q Cgo ; V
X¼
Cgo Cgs ; Cgo
EC ¼
Q Cgo Cgs V
where Q is the ow rate of the inlet air steam, V is the bed volume of the lter material and Cgo is the inlet concentration of VOCs.
Fig. 5
Microscopy image of precipitated crystals.
Analytical procedure VOCs in the inlet and outlet of the petrochemical factory and the four sampling ports were analyzed through a gas chromatograph (Fisons, MD-800) equipped with a FID detector and a capillary column (silica cap column OV-1, 0.25 mm, 30 m, 0.32 mm: MEGA, Italy), and a mass spectrometer (GC/MS, Saturn 2200, Varian Inc.) was used as a detection device for the identication of organic compounds. At least three replicate samples were injected into the GC/MS. Concentrations were determined from a calibration curve that was calculated using known concentration standards of VOC compounds. The standards were prepared by introducing known amounts of the compounds into a 2.8 L sealed plastic bag equipped with a rubber septum. Accuracy tests were performed regularly and conrmed by spiking 75 ppm of each standard compound for all compounds. Percent recoveries were consistently between 95–105%, and the range of 90–110% is used as the quality assurance criterion. A testing method of GB/ T 16046-1995 was used for the detection of benzene, toluene, and xylene, and the total hydrocarbon (count by CH4) is detected using GB/T 15263-1994 standards. The pH of the medium was determined by suspending 6 g medium in 100 mL distilled water. The suspension was stirred with a magnetic bead, and the pH meter of the solution was detected with a calibrated pH meter (Schott Lab 860).
Results and discussion Bioltration of petrochemical VOC vapors was carried out over a period of 10 months under various operating conditions. Various gas ow rates were tested at 2500–3000 m3 h1, which is very typical for the emissions of VOCs in the factory, and the
1004 | Environ. Sci.: Processes Impacts, 2014, 16, 1001–1007
Gas flow rate and inlet VOC concentration effects The gas ow rate and the inlet pollutant concentration are the most important parameters in the bioltration process. Both parameters quantify the amounts of pollutants to be removed in the biotrickling lter. In the present work, the combined effect of the VOC inlet concentration and the gas ow rate on the biotrickling lter performance was investigated. Only the results obtained at the steady state are discussed. Fig. 6 presents the removal efficiency of VOCs versus the inlet concentration for the various tested gas ow rates in the bioltration system. The general trend of the variation of EC versus the VOC IL for the various gas ow rates shows increases in elimination capacity (EC) with increased VOC IL to a certain value which depends on the gas ow rate. The inlet petroleum waste air of the demo has not achieved the maximum EC conditions. In these situations, the increase in the VOC inlet concentration enhanced the transfer rate of VOCs from the gas phase to the biolm so that more microorganisms participate in the biodegradation activity. This behavior can be described as a diffusion limitation regime. At the maximum EC, the entire active microbial population is involved in the biodegradation kinetics and the diffusion limitation does not occur under these operating conditions. As IL is increased above the upper limit of the diffusion limitation regime, EC rst remains constant to its maximum value and then decreases with higher IL. Also, the increase in the VOC inlet concentration to above the maximum EC conditions does not cause a signicant decrease in both X and EC. In fact, the experimental results described in these studies were adequately represented by zero-order kinetics for the pollutant biodegradation rate, which leads to a constant quantity of pollutant degraded per unit time and lter bed volume, i.e. elimination capacity. In the present experiments, the increase in the amount of VOCs transferred to the biolm enhanced by increasing the VOC inlet concentration to above
This journal is © The Royal Society of Chemistry 2014
View Article Online
Environmental Science: Processes & Impacts
Published on 20 January 2014. Downloaded by University of Illinois at Chicago on 30/10/2014 23:13:22.
Paper
Fig. 6
Removal efficiency of VOCs at the exit of the biotrickling filter vs. VOC inlet concentration for various gas flow rates.
the maximum EC conditions seems to have an inhibition effect on the VOC biodegradation rate, which causes a decrease in the amount of VOCs biodegraded. The inability of the system to signicantly biodegrade VOCs of at very high IL concentrations (upper 80%) is attributed to the saturation of VOC concentrations which lead to the insufficient diffusion of the VOCs on the biolm. This analysis reveals that the kinetics of VOC biodegradation in the biolm is not likely to be a zero-order and probably includes a term relating to inhibition at high VOC concentrations. Even though VOCs were in very low IL values between 0.01 and 0.1 g m3, the outlet VOCs still remained at around 1–5 mg m3, which means a certain portion of the pollutant did not reach the liquid phase due to a low mass transfer rate, also the lack in the ability of the system to effectively biodegrade the VOCs at lower 20% IL concentrations is attributed to the low diffusion rates of the VOCs on the biolm, resulting in the
This journal is © The Royal Society of Chemistry 2014
decrease in the biodegradation efficiency. In these conditions, rst-order kinetics may apply and the removal efficiency will subsequently decrease. The pH values in the biolter have changed during operation. pH values remained at 7–8 during the rst two months, then dropped to the range 6–7. Finally, the pH values decreased to 5–5.9 aer 7 months. It is well understood that the organisms used in the bioltration applications typically only survive within a certain pH range 5–8.5, also the organic acids produced during the natural decomposition of the organic matter contribute to the decrease in pH values. For instance, the formation of carbonic acid due to the decomposition of the VOCs would lead to the decrease in pH values. Therefore, the treatment of a contaminant in a new biolter medium should be tested to determine the optimum pH values. At high loading rates, the biolter may acidify due to the formation of acidic intermediates. Hence, pH control is very critical in the
Environ. Sci.: Processes Impacts, 2014, 16, 1001–1007 | 1005
View Article Online
Published on 20 January 2014. Downloaded by University of Illinois at Chicago on 30/10/2014 23:13:22.
Environmental Science: Processes & Impacts
case of high loading rates. In this project, calcium carbonate and dolomite were used as a buffer to neutralize acidity. Aer 8 months, the removal efficiency decreased day by day. We speculate that the following factors may inuence the efficiency. Several reasons could be attributed to the low removal efficiency. First, the temperatures dropped dramatically to 10– 18 C aer the initial 8 months, resulting in the low bacterial activity. Second, the aging of the biolm would lead to the loss of vital cells. Third, the system could be clogged and needs to be re-cleaned. Finally, the decrease in the nutrient concentrations would cause the occurrence of the low removal efficiency. A tailored approach with high efficiency recovery, mainly by physical shock, using a high pressure sprayer, ultrasonic vibration, etc. could be developed to solve the problem within 15 days aer 8 months of operation and start another project.
Shutdown period analysis Shutdown periods may play a role in the bacterial community morphological changes and removal efficiency changes (Fig. 7 and 8). From Fig. 7 and 8, we can clearly observe that the removal efficiencies are widely different from time to time. The overall results indicate a down-and-up trend with more uctuations for benzene and xylene. The overall trend is attributed to the relationship between the concentrations of the pollutants and the
Paper
vitality of the microorganisms. It is obvious that the shutdowns have an immediate effect on the removal efficiency that is lowered initially but bounced back quickly. And these microorganisms are more efficient in degrading toluene and xylene than benzene when the vitality of the microorganisms is affected by the shutdown periods. However, more research is needed in investigating the relationship between the shutdown period and cell vitality as well as morphological changes.
Exhaust analysis The below table lists concentrations of waste gases from the biotrickling lter. Table 1 clearly indicates that benzene and total hydrocarbon cannot be eliminated well even aer 8 months, which is in accordance with the ndings in Fig. 7 and 8. At the nal stage, CO and SO2 were also generated. CO may be produced by the following pathway: RH / R / RO2 / RCHO / RCO / CO wherein R is a hydrocarbon radical, and CO may be generated by the thermal decomposition of RCO radicals or by oxidization of R radicals. Also anaerobic environment degradation in the biolm, biotrickling lter and recycling waste water residues could lead to the formation of CO.
Fig. 7 Influence of shutdown periods on the removal efficiency of VOCs (June 1–14).
Fig. 8 Influence of shutdown periods on the removal efficiency of VOCs (June 15–24).
1006 | Environ. Sci.: Processes Impacts, 2014, 16, 1001–1007
This journal is © The Royal Society of Chemistry 2014
View Article Online
Paper
Published on 20 January 2014. Downloaded by University of Illinois at Chicago on 30/10/2014 23:13:22.
Table 1
Environmental Science: Processes & Impacts I Biotrickling filter exhaust
Sampling time
Phenol (mg m3)
Benzene (mg m3)
Xylene (mg m3)
Toluene (mg m3)
Total hydrocarbon (mg m3)
SO2 (mg m3)
H2S (mg m3)
CO (mg m3)
NOx (mg m3)
2013-5-27 9:00 2013-2-21 9:00 2012-11-1 9:00 2012-8-10 9:00 2012-6-7 10:00 2012-2-29 9:00
0.361 0.252 0.278 0.303 0.175 0.05
609.5 2.8 50.3 265.8 100.8 73.1
4.6
Processes & Impacts View Article Online
Published on 20 January 2014. Downloaded by University of Illinois at Chicago on 30/10/2014 23:13:22.
PAPER
Cite this: Environ. Sci.: Processes Impacts, 2014, 16, 1001
View Journal | View Issue
A demonstration of biofiltration for VOC removal in petrochemical industries Lan Zhao,*ab Shaobin Huangab and Zongmin Weia A biotrickling filter demo has been set up in a petrochemical factory in Sinopec Group for about 10 months with a maximum inlet gas flow rate of 3000 m3 h1. The purpose of this project is to assess the ability of the biotrickling filter to remove hardly biodegradable VOCs such as benzene, toluene and xylene which are recalcitrant and poorly water soluble and commonly found in petrochemical factories. Light-weight hollow ceramic balls (F 5–8 cm) were used as the packing media treated with large amounts of circulating water (2.4 m3 m2 h1) added with bacterial species. The controlled empty bed retention time (EBRT) of 240 s is a key parameter for reaching a removal efficiency of 95% for benzene, toluene, xylene, and 90% for total hydrocarbons. The demo has been successfully adopted and practically applied in waste air treatments in many petrochemical industries for about two years. The net inlet concentrations of benzene, toluene and xylene were varied from 0.5 to 3 g m3. The biofiltration process is highly efficient for the removal of hydrophobic and recalcitrant VOCs with various concentrations from the
Received 11th October 2013 Accepted 9th January 2014
petrochemical factories. The SEM analysis of the bacterial community in the BTF during VOC removal showed that Pseudomonas putida and Klebsiella sp. phylum were dominant and shutdown periods could
DOI: 10.1039/c3em00524k
play a role in forming the community structural differences and leading to the changes of removal
rsc.li/process-impacts
efficiencies.
Environmental impact This work presents a biotrickling lter demo set up in a petrochemical factory in Sinopec Group for about 10 months with a maximum inlet gas ow rate of 3000 m3 h1. The purpose of this project is to assess the capability of biotrickling lters for treatment of hardly biodegradable VOCs such as benzene, toluene and xylene which are recalcitrant and poorly water soluble VOCs in petrochemical factories. Lightweight hollow ceramic balls (F 5–8 cm) were adopted as packing media, combined with large volumes of circulating water (2.4 m3 m2 h1) with bacteria and also the controlled empty bed retention time (EBRT) of 240 s a removal efficiency of 95% (for benzene, toluene and xylene) and 90% for total hydrocarbon can be achieved. The demo has been successfully adopted and practically applied in waste air treatment in petrochemical industries for about two years.
Introduction Volatile organic compounds (VOCs) are compounds that are characterised by high vapor pressure and low water solubility. Many VOCs are human-made chemicals produced and used in the manufacturing of paints, pharmaceuticals, and refrigerants. VOCs are typically industrial solvents such as benzene, toluene and trichloroethylene, fuel oxygenates such as methyl tert-butyl ether (MTBE), and by-products such as chloroform produced by chlorination in water treatment. VOCs are also components of petroleum fuels, hydraulic uids, paint thinners, and dry cleaning agents. Many of these compounds are carcinogenic or mutagenic for human beings and animals. Also they are
a
College of Environmental Science and Engineering, South China University of Technology, Guangzhou, P. R. China. E-mail: [email protected]; Fax: +86-2039380508; Tel: +86-20-39380587
b
Guangdong Provincial Key Laboratory of Atmospheric Environment and Pollution Control, South China University of Technology, Guangzhou, P. R. China
This journal is © The Royal Society of Chemistry 2014
harmful to the environment. Therefore, there is an ongoing need for the development of better treatment methods and prevention technologies to reduce or eliminate VOCs in various environments. Currently, a number of technologies have been developed to eliminate VOCs. For instance, combustion, absorption, condensation, bioltration, UV/bioltration, and UV/TiO2/bioltration technologies are widely used. UV/TiO2 photooxidation (in the form of photocatalysis or direct photolysis) is a promising technology for the elimination of VOCs from contaminated gas streams. It involves the use of ultraviolet (UV) light to partially oxidize the pollutants through the production of ozone and radicals such as hydroxyl radicals, but does not transfer the contaminants from one phase to another.1 Besides, UV/TiO2 can be used to directly dissociate the compound through direct photolysis. These processes are capable of oxidizing a wide range of contaminants in a very short time period (usually a fraction of one second to seconds).2 A bioltration process combined with photocatalytic oxidation UV/TiO2 was developed
Environ. Sci.: Processes Impacts, 2014, 16, 1001–1007 | 1001
View Article Online
Published on 20 January 2014. Downloaded by University of Illinois at Chicago on 30/10/2014 23:13:22.
Environmental Science: Processes & Impacts
to degrade ethylbenzene vapors with higher operation cost by Hinojosa-Reyes et al.3 The greatest degradation rate of EB (0.414 ng m2 min1) was obtained with an inlet loading rate of 127 g m3 h1. The hybrid process enhanced the overall EC of ethylbenzene by 40%. Unfortunately, the photooxidation of many VOCs (e.g. aromatic and chlorinated compounds) is limited in its application due to its propensity to produce byproducts that might be more toxic than their parent compounds.4,5 However, these UV oxidation by-products are oen more biodegradable than their parent compounds as they are more hydrophilic, smaller, and/or simpler in structure.6 Bioltration treatment is the most attractive technology from economic and environmental perspectives due to its low operation costs and low energy requirements.7 A typical biolter consists of a lter bed usually comprised of natural materials (soil, compost, bamboo or peat) which is kept wet to maintain a biologically active layer surrounding the biolter materials, known as the “biolm”. Bioltration utilizes microorganisms inquilinous in the biolter media to degrade the pollutants in a waste gas stream purging into water and CO2. However, bioltration treatments can hardly degrade recalcitrant insoluble VOCs. Biotrickling lters are similar to biolters except that the packing medium is rigid or semi-rigid and inorganic. Also, a nutrient containing liquid phase is continuously recirculated through the system. Microorganisms grown on the packing medium may also be suspended in the recirculating liquid phase. Continuous or periodic addition/recirculation of liquid allows for control of nutrient concentrations and reaction conditions. One disadvantage of biotrickling lters in comparison to biolters is the lower specic surface area of the packing media, which makes it difficult to treat poorly water-soluble compounds. Clogging is another potential problem due to a readily available nutrient supply. Jianjun Li et al.8 investigated the performance of a eld-scale biotrickling lter (BTF) in the removal of waste gases containing low concentrations of mixed volatile organic compounds. Results showed that acetone and methyl ethyl ketone (MEK) were more easily removed than toluene and styrene. The removal efficiency for acetone and MEK reached over 99% on days 28 and 25 of the operation, whereas those for toluene and styrene were 80 and 63% on day 38. The maximum individual elimination capacities for styrene, toluene, acetone, and MEK were 10.2, 2.7, 4.7, and 8.4 g m3 h1, respectively. These values were achieved at inlet loading rates of 13.9, 3.3, 4.8, and 8.5 g m3 h1, respectively and at an empty bed retention time of 14 s. C. Lu et al.9,10 studied the effects of pH values, moisture rates, and ow patterns on the trickle-bed biolter performance for BTEX (benzene, toluene, ethylbenzene and xylene) removal. The BTEX removal efficiency increased as the pH values of the nutrient feed increased in the range of 5–8. In the pH range 7.5–8 and nutrient feeding rate (NFR) range 6.02–8.6 L m3 h1, removal efficiencies of each compound were greater than 80% with a loading of 143 g BTEX m3 h1. The BTEX removal efficiency increased as the NFR increased in the range 3.44–8.6 L m3 h1. For a 17.2 L m3 h1 NFR, the BTEX removal efficiency was slightly lower than those at 6.02 L m3 h1 NFR and 8.6 L m3 h1 NFR, but was signicantly higher than that at 3.44 L m3 h1 NFR.
1002 | Environ. Sci.: Processes Impacts, 2014, 16, 1001–1007
Paper
The optimum NFR was in the range of 6.02–8.6 L m3 h1. In the optimum pH and moisture ranges, the BTEX removal efficiencies were greater than 80%. As a result, biotrickling ltering appeared to be an effective process for BTEX removal with high loading. A counter-current ow biotrickling lter has the advantages of achieving more homogeneous BTEX removal and biomass growth in each section than the co-current ow system, but it suffered from a higher pressure drop across the bed. Jian M. Chen et al.11 also studied microbes bound to a medium in which wheat bran, red wood powder and diatomaceous earth carriers were used as inoculants for a biotrickling lter (BTF) for treating gases contaminated with a mixture of benzene, toluene and o-xylene (BTo-X). An overall removal efficiency of more than 87.9% was achieved aer a start-up period of as low as 4 days. At BTo-X loading rates (LRs) below 60.0 g m3 h1, the BTF performance was similar to those of EBRTs at 90, 60, 45 and 30 s with an elimination capacity (EC) approaching the LR; stable REs above 91.3% for benzene and toluene and above 82.8% for o-xylene were achieved. A maximum EC of 97.7 g m3 h1 was obtained at the inlet load of 146.4 g m3 h1. The mass ratio of carbon dioxide produced to the BTo-X removed was approximately 2.62, which conrmed the complete degradation of BTo-X. The results demonstrate that microbes bound to a solid carrier can be an alternative to traditional liquid inoculums applied in BTFs and highlight their potential applicability to BTF technologies. This study developed a low cost bioltration process to treat hydrophobic waste air from the petrochemical factories without integration of UV/TiO2 photolysis. It is an effective technology to remove recalcitrant air pollutants such as benzene, toluene, xylene, aromatics, etc. The effect of shutdowns on BTF performances was comprehensively investigated.
Methodology Demo set-up and procedure The demo process involving two series biotrickling lters was carried out (Fig. 1) for a period of ten months for VOC removal. The contaminated stream was generated using waste air from the petrochemical factory in Guangzhou, China, that delivered VOCs at a uctuating rate. In the pretreatment tank (F 0.94, 3 m height), the maximum ow rate is 3000 m3 h1 (1.2 m s1) and the spray ow rate is 5 m3 h1 with a 3.0 Pa pressure drop. Prints and oils were rst removed prior to the process to eliminate any interfering chemicals. Subsequently, temperatures were set to be lower than 40 C and the humidity was set to >90%. The contaminated stream from a compressor can then enter all the reactors from the top. The biotrickling lter column was made up of a stainless steel (SS) of 3.8 m diameter and 12 m height, which is featured with 2 min of empty bed retention time (EBRT) from Nantong Langgao Equipment Ltd. The empty bed ow rate was maintained at 0.067 m s1. Waste air can be opened and closed with a valve at the bottom. Four gas chromatography (GC) sampling ports (S1–S4 from Agilent 6750) were set up to make S1 as the inlet of petrochemical waste air, S2 as the pretreatment outlet, S3 as the I biotrickling lter outlet, and S4 as the II biotrickling lter outlet. The biotrickling
This journal is © The Royal Society of Chemistry 2014
View Article Online
Published on 20 January 2014. Downloaded by University of Illinois at Chicago on 30/10/2014 23:13:22.
Paper
Fig. 1
Environmental Science: Processes & Impacts
Demo set-up for the biofiltration of petrochemical waste air.
lter was maintained under ambient temperature. A modied recycling system with a 2.5–3 mm diameter screw nozzle and 30 m3 valid volume from Nantong Langgao Equipment Ltd. was designed to save 80% water energy cost. Recycling water was controlled at 2.4 ton h1 (2.4 m3 m2 h1) with a 500 Pa pressure drop. The temperature inside the biolter column was monitored with a wet bulb thermometer (LY-09, Qingdao Jingcheng Instrument Ltd.), and the pressure drop was monitored biweekly with a pressure meter (Comark, UK).
was calculated. The identication of isolated bacteria was performed using a Biolog identication system (Microlog-1, visually interpreted system) and GN-2 (Biolog) plates. All chemicals used were of analytical grade (SD Fine Chemicals, China). The bacterial community had been inoculated from the activated sludge drawn from Zhujiang wharf for about 6 months prior to the work. To develop biolms on the surface of packing materials, 17–48 L h1 of the activated sludge plus 450 L of nutrient solution were continuously recycled for 5 days during which only fresh air was supplied to the BTF before the startup. The results showed that Pseudomonas putida and Klebsiella sp. phylum were dominant microbes in the BTF, also the initial inoculums played an important role and entailed a lengthened startup period in the enhancement of BTF removal capacity due to the inorganic nature of the packing material. Moreover, the presence of a continuous water layer over the biolm reduces the removal performance for hydrophobic pollutants due to mass transfer limitations from the gas to the aqueous phase. The biolm was completely formed aer 25 days (Fig. 2). Fig. 3 shows the morphological changes of Pseudomonas putida phylum before and aer the treatment of VOCs for about 10 months (50 000 times magnication). Crystals were also induced in the agar plate by a mixed bacterial community (Fig. 4 and 5).
Packing medium Light weight hollow ceramic media have been used extensively in recent years because they are cheap and have diversied holes which can be used to adsorb various pollutants once they become acclimatized to the substrate. The volume of packing is 200 m3 in which three layers are supported by a F 60 mm sieve plate from Nantong Langgao Equipment Ltd.
Fig. 2
SEM photograph of the biofilm.
Microbial analysis of the biotrickling lter During the start up of the unit, the reactor was inoculated with respect to the medium in a closed loop to maximize the cell adhesion to the packing media for about 42 days. The composition of the medium included 1.0 g L1 KNO3, 10.48 g L1 Na2HPO4$12H2O, 1.50 g L1 KH2PO4, 0.05 g L1 MgSO4, 8.0 g L1 C4H4Na2O, and 1.0 mL L1 trace metal solution (50 g EDTA, 2.20 g ZnSO4, 5.50 g CaCl2, 5.06 g MnCl2$4H2O, 5.00 g FeSO4$7H2O, 1.10 g (NH4)6Mo7O2$4H2O, 1.57 g CuSO4$5H2O, and 1.61 g CoCl2$6H2O per liter; pH ¼ 7.0, from GZ Chemical Reagent Factory, Analytical Reagent). Petrochemical waste air was introduced into the unit with relatively uctuating ow rates (30–450 mg m3 benzene, 50– 1000 mg m3 toluene, and 140–1200 mg m3 xylene). The number of visible colonies formed aer 7 days was counted, and the average number of cells per gram lter bed This journal is © The Royal Society of Chemistry 2014
SEM of Pseudomonas putida before and after the treatment of VOCs (50 000 times).
Fig. 3
Environ. Sci.: Processes Impacts, 2014, 16, 1001–1007 | 1003
View Article Online
Published on 20 January 2014. Downloaded by University of Illinois at Chicago on 30/10/2014 23:13:22.
Environmental Science: Processes & Impacts
Fig. 4 Crystals induced in the agar plate by a mixed bacterial community.
Paper
recycling water volume was designed as 2.4 m3 m2 h1, corresponding to an optimistic empty bed gas residence time of 120 s as veried by the designer of the factory. The waste air inlet concentration was also varied in the range of 30–450 mg m3 benzene, 50–1000 mg m3 toluene, and 140–1200 mg m3 xylene, because the concentrations are typical values from the petroleum plant and are representative of commonly low, medium and high concentrations found in similar factories. The bioltration performance is discussed in terms of the VOC inlet load (IL, g m3 h1), the removal efficiency (X) and the elimination capacity (EC, g m3 h1) are evaluated using the following equations from the Handbook on Waste Gas Treatment Engineering Technology, Version 1: IL ¼
Q Cgo ; V
X¼
Cgo Cgs ; Cgo
EC ¼
Q Cgo Cgs V
where Q is the ow rate of the inlet air steam, V is the bed volume of the lter material and Cgo is the inlet concentration of VOCs.
Fig. 5
Microscopy image of precipitated crystals.
Analytical procedure VOCs in the inlet and outlet of the petrochemical factory and the four sampling ports were analyzed through a gas chromatograph (Fisons, MD-800) equipped with a FID detector and a capillary column (silica cap column OV-1, 0.25 mm, 30 m, 0.32 mm: MEGA, Italy), and a mass spectrometer (GC/MS, Saturn 2200, Varian Inc.) was used as a detection device for the identication of organic compounds. At least three replicate samples were injected into the GC/MS. Concentrations were determined from a calibration curve that was calculated using known concentration standards of VOC compounds. The standards were prepared by introducing known amounts of the compounds into a 2.8 L sealed plastic bag equipped with a rubber septum. Accuracy tests were performed regularly and conrmed by spiking 75 ppm of each standard compound for all compounds. Percent recoveries were consistently between 95–105%, and the range of 90–110% is used as the quality assurance criterion. A testing method of GB/ T 16046-1995 was used for the detection of benzene, toluene, and xylene, and the total hydrocarbon (count by CH4) is detected using GB/T 15263-1994 standards. The pH of the medium was determined by suspending 6 g medium in 100 mL distilled water. The suspension was stirred with a magnetic bead, and the pH meter of the solution was detected with a calibrated pH meter (Schott Lab 860).
Results and discussion Bioltration of petrochemical VOC vapors was carried out over a period of 10 months under various operating conditions. Various gas ow rates were tested at 2500–3000 m3 h1, which is very typical for the emissions of VOCs in the factory, and the
1004 | Environ. Sci.: Processes Impacts, 2014, 16, 1001–1007
Gas flow rate and inlet VOC concentration effects The gas ow rate and the inlet pollutant concentration are the most important parameters in the bioltration process. Both parameters quantify the amounts of pollutants to be removed in the biotrickling lter. In the present work, the combined effect of the VOC inlet concentration and the gas ow rate on the biotrickling lter performance was investigated. Only the results obtained at the steady state are discussed. Fig. 6 presents the removal efficiency of VOCs versus the inlet concentration for the various tested gas ow rates in the bioltration system. The general trend of the variation of EC versus the VOC IL for the various gas ow rates shows increases in elimination capacity (EC) with increased VOC IL to a certain value which depends on the gas ow rate. The inlet petroleum waste air of the demo has not achieved the maximum EC conditions. In these situations, the increase in the VOC inlet concentration enhanced the transfer rate of VOCs from the gas phase to the biolm so that more microorganisms participate in the biodegradation activity. This behavior can be described as a diffusion limitation regime. At the maximum EC, the entire active microbial population is involved in the biodegradation kinetics and the diffusion limitation does not occur under these operating conditions. As IL is increased above the upper limit of the diffusion limitation regime, EC rst remains constant to its maximum value and then decreases with higher IL. Also, the increase in the VOC inlet concentration to above the maximum EC conditions does not cause a signicant decrease in both X and EC. In fact, the experimental results described in these studies were adequately represented by zero-order kinetics for the pollutant biodegradation rate, which leads to a constant quantity of pollutant degraded per unit time and lter bed volume, i.e. elimination capacity. In the present experiments, the increase in the amount of VOCs transferred to the biolm enhanced by increasing the VOC inlet concentration to above
This journal is © The Royal Society of Chemistry 2014
View Article Online
Environmental Science: Processes & Impacts
Published on 20 January 2014. Downloaded by University of Illinois at Chicago on 30/10/2014 23:13:22.
Paper
Fig. 6
Removal efficiency of VOCs at the exit of the biotrickling filter vs. VOC inlet concentration for various gas flow rates.
the maximum EC conditions seems to have an inhibition effect on the VOC biodegradation rate, which causes a decrease in the amount of VOCs biodegraded. The inability of the system to signicantly biodegrade VOCs of at very high IL concentrations (upper 80%) is attributed to the saturation of VOC concentrations which lead to the insufficient diffusion of the VOCs on the biolm. This analysis reveals that the kinetics of VOC biodegradation in the biolm is not likely to be a zero-order and probably includes a term relating to inhibition at high VOC concentrations. Even though VOCs were in very low IL values between 0.01 and 0.1 g m3, the outlet VOCs still remained at around 1–5 mg m3, which means a certain portion of the pollutant did not reach the liquid phase due to a low mass transfer rate, also the lack in the ability of the system to effectively biodegrade the VOCs at lower 20% IL concentrations is attributed to the low diffusion rates of the VOCs on the biolm, resulting in the
This journal is © The Royal Society of Chemistry 2014
decrease in the biodegradation efficiency. In these conditions, rst-order kinetics may apply and the removal efficiency will subsequently decrease. The pH values in the biolter have changed during operation. pH values remained at 7–8 during the rst two months, then dropped to the range 6–7. Finally, the pH values decreased to 5–5.9 aer 7 months. It is well understood that the organisms used in the bioltration applications typically only survive within a certain pH range 5–8.5, also the organic acids produced during the natural decomposition of the organic matter contribute to the decrease in pH values. For instance, the formation of carbonic acid due to the decomposition of the VOCs would lead to the decrease in pH values. Therefore, the treatment of a contaminant in a new biolter medium should be tested to determine the optimum pH values. At high loading rates, the biolter may acidify due to the formation of acidic intermediates. Hence, pH control is very critical in the
Environ. Sci.: Processes Impacts, 2014, 16, 1001–1007 | 1005
View Article Online
Published on 20 January 2014. Downloaded by University of Illinois at Chicago on 30/10/2014 23:13:22.
Environmental Science: Processes & Impacts
case of high loading rates. In this project, calcium carbonate and dolomite were used as a buffer to neutralize acidity. Aer 8 months, the removal efficiency decreased day by day. We speculate that the following factors may inuence the efficiency. Several reasons could be attributed to the low removal efficiency. First, the temperatures dropped dramatically to 10– 18 C aer the initial 8 months, resulting in the low bacterial activity. Second, the aging of the biolm would lead to the loss of vital cells. Third, the system could be clogged and needs to be re-cleaned. Finally, the decrease in the nutrient concentrations would cause the occurrence of the low removal efficiency. A tailored approach with high efficiency recovery, mainly by physical shock, using a high pressure sprayer, ultrasonic vibration, etc. could be developed to solve the problem within 15 days aer 8 months of operation and start another project.
Shutdown period analysis Shutdown periods may play a role in the bacterial community morphological changes and removal efficiency changes (Fig. 7 and 8). From Fig. 7 and 8, we can clearly observe that the removal efficiencies are widely different from time to time. The overall results indicate a down-and-up trend with more uctuations for benzene and xylene. The overall trend is attributed to the relationship between the concentrations of the pollutants and the
Paper
vitality of the microorganisms. It is obvious that the shutdowns have an immediate effect on the removal efficiency that is lowered initially but bounced back quickly. And these microorganisms are more efficient in degrading toluene and xylene than benzene when the vitality of the microorganisms is affected by the shutdown periods. However, more research is needed in investigating the relationship between the shutdown period and cell vitality as well as morphological changes.
Exhaust analysis The below table lists concentrations of waste gases from the biotrickling lter. Table 1 clearly indicates that benzene and total hydrocarbon cannot be eliminated well even aer 8 months, which is in accordance with the ndings in Fig. 7 and 8. At the nal stage, CO and SO2 were also generated. CO may be produced by the following pathway: RH / R / RO2 / RCHO / RCO / CO wherein R is a hydrocarbon radical, and CO may be generated by the thermal decomposition of RCO radicals or by oxidization of R radicals. Also anaerobic environment degradation in the biolm, biotrickling lter and recycling waste water residues could lead to the formation of CO.
Fig. 7 Influence of shutdown periods on the removal efficiency of VOCs (June 1–14).
Fig. 8 Influence of shutdown periods on the removal efficiency of VOCs (June 15–24).
1006 | Environ. Sci.: Processes Impacts, 2014, 16, 1001–1007
This journal is © The Royal Society of Chemistry 2014
View Article Online
Paper
Published on 20 January 2014. Downloaded by University of Illinois at Chicago on 30/10/2014 23:13:22.
Table 1
Environmental Science: Processes & Impacts I Biotrickling filter exhaust
Sampling time
Phenol (mg m3)
Benzene (mg m3)
Xylene (mg m3)
Toluene (mg m3)
Total hydrocarbon (mg m3)
SO2 (mg m3)
H2S (mg m3)
CO (mg m3)
NOx (mg m3)
2013-5-27 9:00 2013-2-21 9:00 2012-11-1 9:00 2012-8-10 9:00 2012-6-7 10:00 2012-2-29 9:00
0.361 0.252 0.278 0.303 0.175 0.05
609.5 2.8 50.3 265.8 100.8 73.1
4.6
