Field-tests at Kiviõli Keemiatööstus OÜ oil-shale processing plant in Kiviõli, Estonia

Transcription

1 Field-tests at Kiviõli Keemiatööstus OÜ oil-shale processing plant in Kiviõli, Estonia Introduction The field-tests for plasma-treatment of exhaust gases were carried out in several different sites of Kiviõli Keemiatööstus OÜ oil-shale processing plant. The objective of the tests carried out in the first two sites was to decrease the concentration of VOC-s in the exhaust air of the buildings. In these buildings, the air contamination was caused due to the breathing of shale-oil processing units and the VOC concentrations were relatively small but resulted in considerable odour. Third and fourth sites were tanks for storage of oil-shale products and in these sites the expected concentration of VOC-s was higher. In these four sites, the composition was of mixture of various aliphatic and aromatic VOC-s. The final site was a de-phenolization unit where the treated gas had specifically high Butyl-Acetate concentration. Experimental details General scheme for the exhaust gas treatment is shown in figure 1. In most of the experiments, the gas was flowing through the plasma-treatment system where different parts were connected by 100 mm duct pipes. In the experiments where the contaminated air had to be diluted, it was mixed with outdoor air and a valve was used to regulate the amount of contaminated gas. A filter was used to clean the exhaust from mist and particles which could otherwise contaminate the plasma reactor. Subsequently, the gas was flowing through the plasma reactor which was switched on and off to see the effect of the plasma. A calibrated flow meter was used for the measurement of gas flow rate and a ventilator was used to adjust the airflow through the system. A separate pipeline was attached to the outlet of the plasma reactor for the gas composition analysis with FTIR spectrometer. Small amount of the gas was pumped through the FTIR spectrometer with pipes of 0.4 mm inner diameter) and an air filter was used to protect the spectrometer from particles and aerosols while a carbon filter was used to protect the pump.

2 Figure 1.Schematic picture of the experimental setup. Plasma reactor The plasma unit was a stack type dielectric-barrier discharge reactor designed specifically for the purpose of such field tests. It consists of mesh electrodes made of stainless steel, separated by dielectric barriers made of phlogopite. A scheme of this reactor design is given in the right part of figure 2. The electrodes are alternating connected to sinusoidal high voltage and ground. The stack is housed in a solid frame made of stainless steel also containing the high voltage feed-through. The Swagelok connectors shown in figure 2 (white arrows) were removed to connect the 100 mm pipes. The input power of the reactor was changeable up to the value of approximately 165 W.As the flow rates were usually in the range of 1-10 m 3 /h, the corresponding SIE values were in the range of 50 to 250 J/L. Figure 2: Left: Field test plasma unit with connectors for laboratory use (white arrows), Right: Scheme of stack reactor design

3 Gas composition analysis Two methods were used for the gas-composition analysis. In-situ measurements from the outlet of the plasma device were carried out with FTIR spectrometer Bruker Alpha having multi-pass cell with the length of 4.8 m. The FTIR method was not usable for distinguishing between various species of VOC-s but characteristic bands in the spectra around 3000 cm -1 could be used as a measure of the total VOC concentration (see FTIR spectra below). Only exception was the dephenolization unit where the gas contained dominantly Butyl-Acetate and the FTIR spectra also clearly corresponded to Butyl-Acetate. Due to the presence of large number of different species with overlapping spectra, the FTIR method didn t allow to obtain quantitative values of the VOC concentrations. However, the decrease of the characteristic bands of VOC could be used as a measure of relative change of the VOC concentration after the plasma treatment. In addition to VOC-s, the FTIR spectra contained also bands characteristic to ozone, NOx and CO and there was calibration data available for these species. As a result, the concentration of these species could be measured quantitatively by FTIR in some experimental conditions. One problem with the gas composition analysis was the saturation effect (tanks for example). Practically all IR radiation absorbs at certain wavelength regions at very high VOC concentrations, and the spectrum is not usable. In such case, the concentration of inorganic compounds was also not reliably measured. At very small concentrations, there was the problem with measurement accuracy (condensation unit). In the case of fast and large variation of VOC concentrations (dephenolization units and tanks), the FTIR analysis was also complicated. Mass-Spectrometry/Gas-Chromatography (MS/GC) was also used for ex-situ analysis of species. The gas samples for MS/GC analysis was taken simultaneously before and after the working plasma device for 5 minutes and the samples were treated later in the laboratory. The MS/GC analysis method was calibrated for some species (aromatic) and it was possible to quantitatively analyze these species. The MS/GC method also allowed to distinguish between certain specimens of VOC-s or at least to suggest the general class of a particular specimen (alkane, alkene, alcohol, cyclic, aromatic, ketone etc.). The relative change of a particular specimen due to plasma treatment could also be obtained. Description of the test sites The first tests were carried out in two buildings where the air had strong odour. First one was condensation room of a generator where there were a number of condensation tanks breathing into the room (4 for medium fraction and 2 for heavy fraction of oils). In addition, there were 2 tanks with mixture of water and phenol. The average VOC concentration in this building was too small to give measurable FTIR spectrometer signal. To get reasonable signal we treated the air directly above the openings of tanks (1 medium and 1 heavy fraction) breathing into the room. The tanks were emptied and filled during the measurements and building was opened to moving air which occasionally changed the concentration of VOC-s. The MS/GC signal was also very small and could only be used as an indication for some species. Second building where the air was treated was for purification of heavy oils. In this building, there was also rather small concentration of VOC-s but still in the

4 detection limits of FTIR. The MS/GC data was not usable for quantitative analysis. Relatively small applied power was used in the first two sites (usually W). The flow rates were about 9-10 m 3 /h which gives energy density of about J/L ( kwh/m 3 ). Several tests were carried out for the cleaning of the exhaust originating directly from the tanks of oil-storage. This exhaust had large concentration of VOC-s. The tanks were outside of buildings and the concentration varied greatly due to the changing wind conditions. It was not possible to get reliable data from these sites. Part of the exhaust of a scrubber of a de-phenolation unit containing Butyl-Acetate was directed through the plasma device. The applied power was 150 W in all these measurements while the flow rates were either 2.4 or 4.8 m 3 /h. This results in Specific Input Energy (SIE) of 55 and 110 J/L (0.015 and kwh/m 3 ). Results Cleaning of building air (odour) In the case of the exhaust from the tank with medium fraction of oil, there was considerable decrease of the VOC species. The FTIR spectrum was typical for the gasoline and had characteristic bands of hydrogen-carbon bonds between 2850 and 3000 cm -1 (higher absorption intensity marked as blue in the figure in left). The decrease of the intensity of absorption bands due to the plasma-treatment (lower absorption marked as red in the figure) depended on the specific band and decrease was larger for the band at 2940 cm -1 compared to the one at 2870 and 2970 cm -1. This result suggests different treatment efficiencies for different compounds. As a note, the partially oxidized VOC-s result also considerable absorption at the wavelengths around 2970 cm -1. At relatively small inlet concentrations the decrease was approximately %. At about three time s higher concentrations, the decrease of the band seemed to be somewhat higher, % but measurement error was also at least 10 %. We also applied larger power in that case which means that both the energy density and temperature was higher and can explain somewhat larger removal. When the air-samples were obtained above the tanks with heavy fraction of oil, the intensities of characteristic VOC bands were generally smaller and the shape was also somewhat different. Similarly to the case of medium fraction, the application of plasma resulted in different decrease of the intensities for the registered FTIR bands. Some bands (most notably at 2940 cm -1 ) decreased

5 faster than the band with maximum and 2970 cm -1. In average, the application of plasma treatment resulted in the decrease of about 30 % for band at 2940 cm -1 while it was only about 10 % for other species. Unfortunately, with present FTIR setup, it is not possible to assign specific bands to specific species. The ozone concentration depended also on the VOC inlet concentration and was higher when VOC inlet concentration was smaller. Highest obtained ozone concentrations were 100 ppm suggesting inefficient use of plasma produced oxygen radicals. The removal efficiency generally improved with treatment time indicating the positive effect of temperature increase. As a final note concerning the results obtained in the first building, the air contamination comes mostly from tanks with medium fraction of oils and thus the removal of about 60 % of VOC from the room air can be expected at used conditions. According to FTIR data, even smaller concentrations of VOC-s were present in the second building where the heavy oils were purified. The plasma treatment allowed to remove up to 80 % of the VOC-s according to FTIR (specific input energy of 55 J/L). The ozone concentration increased up to 180 ppm indicating again somewhat inefficient removal of the species. There was also some production of CO and the concentrations reached about 10 ppm according to calibrated data of FTIR absorption. For this building it was also possible to obtain the relative removal of VOC-s at various specific input powers (figure 3). The inlet concentration somewhat varied but there was a clear trend that at higher energies, the removal efficiency was also highest. As a note, the current probe necessary for power measurement was broken during these measurements and the power necessary for the determination of SIE was obtained from the power source. This may result in considerable error in the SIE values at present case as shown also in the figure.

6 Figure 3.The percentage of outlet concentration of VOC-s in plasma reactor according to absorption by FTIR. The MS/GC data indicated about % decrease of aromatic compounds above the tank with medium fraction of heavy oil at 50 W. At higher power values, the removal efficiency was unexpectedly somewhat smaller (35 %). In the case of heavy fraction, the concentrations of aromatic species were too small to give detectable MS/GC signal (< 0.5 mg/m 3 ). Table 1.The MS/GC concerning the removal of several aromatic species by plasma-treatment at two different power values. VOC 50 W 100 W Before After Before After Benzene Ethyl-benzene 1.1 < <0.5 Toluene The MS/GC chromatograms of gas samples where plasma treatment at either 100 W or 50 W was applied are shown in figure 4 and 5. Before plasma treatment, there were several peaks corresponding to unsaturated plasma species and other heavier unspecified species. The plasma treatment decreased the concentration of most of these species to the noise level while there appeared new species which were mostly identified as partially oxidized organic species formaldehyde, acetic acid and various ketones.

7 Figure 4. Chromatograms from gas samples obtained above tanks with medium fraction of oil before and after plasma treatment at 100 W (50 J/L). The diamonds indicate peaks which are peculiarities of the present MS/GC measurement and present regardless of gas composition. The concentrations of organic species were somewhat larger during the treatment at 50 W (figure 5), but otherwise the trends were similar. Thus, the measurements show that for most of the species, the increase of power from 50 to 100 W did not improve the removal efficiency. One explanation for the observed results could be the removal of unsaturated species already at smaller applied power values while the removal of aromatic species was inefficient even at higher power values. In addition, the initial oxidation of unsaturated species likely results in the production of new active radicals (O, OH, HO 2 ) which enhance also the removal efficiency of aromatic species. In the case of heavy fraction of oils, there were practically no other species detectable and practically no aromatic species were removed.

8 Figure 5. Chromatograms from gas samples obtained above tanks with medium fraction of oil before and after plasma treatment at 50 W (21 J/L). The diamonds indicate peaks which are peculiarities of the present MS/GC measurement and present regardless of gas composition. Figure 6. Chromatograms from gas samples obtained in oil cleaning room before and after plasma treatment at 125 W (55 J/L). The diamonds indicate peaks which are peculiarities of the present MS/GC measurement and present regardless of gas composition.

9 Intensity, a.u. The chromatogram for gas sample obtained in the oil-cleaning room before and after plasma treatment is shown in figure 6. The specific input energy was about 55 J/L. The peaks of most of the observed species were rather low and most of the peaks decreased to noise level after the plasma treatment. The peaks corresponding to aromatics also decreased about twice. There was one peak at 5 min corresponding to a partially oxidized organic specimen which did not decrease at used power value. Cleaning the exhaust from the tanks In these sites, the variation of VOC composition was too high and the concentrations were also too high for the measurements by FTIR. The dilution did not work out properly due to the changes in wind which altered the dilution ratio. The MS/GC analysis was not done for these sites. Butyl-acetate removal in de-phenolization unit Figure 7 depicts two full spectra obtained from the outlet of the plasma device without plasma and with plasma switched on. Without plasma, the FTIR spectra were dominated by H 2 O, CO 2 and butyl-acetate (BA) bands. After switching on the plasma, the bands corresponding to BA decrease while there appear new bands corresponding to ozone (O 3 ), carbon monoxide (CO) and various NO x species. The bands corresponding to methane were also sometimes visible. Bands corresponding to other VOC species were not clarified which suggests that CO, CO 2 and H 2 O were the main reaction products. 0,9 0,8 0,7 0,6 O 3 plasma OFF Plasma ON 0,5 0,4 BA 0,3 0,2 0,1 0-0,1 BA N 2 O CO CO Wavenumber, cm -1 Figure 7.FTIR spectra of the exhaust gas containing butyl-acetate with and without plasma treatment. BA

10 Figure 8 shows the time-series (4 measurements during 1 minute) of several FTIR bands. The violet shaded areas show times when plasma was switched on. There is large variation on butyl acetate concentration which made the analysis of plasma effect complicated. The variation of the concentration was much larger than the effect of plasma. In addition, during very high butyl-acetate spikes, the analysis of other species (e.g. ozone and NO x species) was also not possible. In one occasion (shown as best data at ), the concentration of BA was relatively stable (still slightly decreasing) and it was possible to see the effect of plasma. The data obtained during last three short plasma treatments between 200 and 215 were also somewhat usable but had much larger uncertainty. Figure 8.Time dependence of the butyl-acetate peak at 2950 cm -1, ozone peak at 1000 cm -1 and N 2 O peak at 2200 cm -1. Four measurements were made during 1 minute. Plasma ON times are shown as shaded areas. The outlet concentration of butyl-acetate was between 60 to 80 % of the inlet concentration at various plasma treatment conditions (SIE values at 55 and 110 J/L and varying BA inlet concentration). During the treatments with numbers , the SIE value was 55 J/L and the outlet concentration was approximately 65-70%. The exact concentrations of butyl-acetate cannot be determined with present setup without proper calibration. A study concerning the removal of butyl-acetate by plasma and catalyst [1], showed that in the case of 120 ppm butyl-acetate, the removal by plasma did not depend on temperature ( C) and depended almost linearly on input energy. There was about 25 % of removal of butyl-acetate at highest SIE value of 330 J/L (10 % of removal at 120 J/L). The main oxidation products detected were CO and CO 2. The use of catalyst resulted in about two-fold increase of the removal of butyl-acetate and could be also used to improve the removal in the present case.

11 Discussion of the results The efficiency of plasma treatment depends on the type of organic compounds and on the concentration range of the compounds. For certain compounds (unsaturated hydrocarbons for example) which have fast reactions with plasma produced reactive oxygen species the removal efficiency is relatively high. The energy input has to be increased for higher concentration of such species. For other compounds (saturated hydrocarbons and aromatics) the radical reactions proceed more slowly and efficiency remains smaller. There exists a concentration range where the removal percentage of these species does not depend on the VOC concentration and same amount of energy is needed regardless of concentration. Considerable amount of ozone formed in the second case is also an indication of inefficiency. In the present tests, there was a mixed composition of the gases. In most of the cases, the removal percentage did not depend considerably on the concentration of VOC-s while relatively large amount of ozone was formed. The use of suitable catalyst after the plasma zone can both decrease the concentration of ozone in the outlet and further oxidize the unreacted VOC-s in the exhaust. Conclusions and outlook Best results were expectedly obtained during the odour removal where the initial concentrations of VOC-s were relatively small. The largest removal of 80 % was obtained in the heavy oil purification building at moderate energy input of 55 J/L. In other cases, the removal efficiency remained below 65 % at similar input energies. The presence of ozone in the outlet of plasma unit indicates inefficient use of plasma produced radicals and hints at possible enhancement options. The use of catalyst may further improve the removal efficiency of VOC-s while also decreasing the outlet ozone levels. The concentration of butyl-acetate in the de-phenolization unit varied in a very large range which made the analysis of plasma treatment complicated. At relatively small inlet concentrations of butylacetate, about % could be removed by plasma at input energies of 55 J/L. The use of plasmacatalytic systems may further improve the removal of butyl-acetate as seen by other studies. References [1] Demidiouk V, Moon S I, Chae J O 2003 Toluene and butyl acetate removal from air by plasmacatalytic system, Catalysis Communications