Membrane gas absorber for H 2 S removal

Transcription

1 Membrane gas absorber for H 2 S removal B. Jefferson *#a, S. Georgaki b, A. Esquiroz c and R. Stuetz d School of Water Sciences, Cranfield University, Bedfordshire, MK43 AL, UK a Tel. 44 ()234 75; Fax 44 () ; b.jefferson@cranfield.ac.uk b Tel. 44 ()234 75; Fax 44 () ; s.georgaki@cranfield.ac.uk c Tel. 44 ()234 75; Fax 44 () ; a.esquiroz.22@cranfield.ac.uk d Tel. 44 ()234 75; Fax 44 () ; r.stuetz@cranfield.ac.uk Abstract The emission of hydrogen sulphide is a major problem associated with anaerobic treatment of sulphate and sulphite containing wastewaters. Conventional absorbing processes such as packed towers, spray towers or bubble columns are all constrained by factors such as flooding and foaming. The work presented in this paper refers to an absorption process based on a non-wetted hollow fibre membrane for the scrubbing of hydrogen sulphide from air, with water as the contact solvent. The performance of the unit was seen to be highly dependant on the gas phase velocity and the initial water. The latter is shown to relate to the dissociation reaction that occurs as H 2 S is a weak dibasic acid. Correlation of the performance data in terms of classical dimensionless correlations revealed the exponent of the hydrodynamic parameter (Gz) to increase as a function of indicating a greater importance on the gas phase velocity as the increases. Overall the process was shown to be controlled by the membrane resistance when operated at high gas phase velocities or low initial water.. Introduction The development of more efficient methods of membrane-based gas absorption has recently received considerable interest in two areas: the biological treatment of contaminated gas streams (trichloroethylene, propene, and other VOCs), and the abiotic gas absorption of waste gases (sulfur dioxide, ammonia, etc). For gas-liquid contact devices, two types of membranes can be used: hydrophobic microporous polymer material such as polypropylene and dense polymeric material such as silicone rubber (polydimethylsiloxane). The microporous material presents a higher permeability whereas dense material is more selective. Conventional absorption processes are based on contacting the gas mixture with a scrubbing solvent in packed or spray towers. Membrane based gas absorption using a microporous hydrophobic material presents many advantages over these conventional technologies, as recently reviewed by Sirkar [], and have been demonstrated for a number of duties. The principal attribute of the membrane contactor is high mass transfer expedited by (i) independent control of gas and liquid flow rates without the flooding or foaming experienced by packed and spray towers, and (ii) large interfacial areas offered by the HF membrane geometry. An early example of the technology is blood oxygenation using hydrophobic FP-PTFE (polytetrafluoroethylene) membranes, where blood flows in one side and air or pure oxygen on the other side of the membrane []. Other systems include absorption of acidic gases such as carbon dioxide, sulphur dioxide, ammonia in solvents such as water, soda etc. []. In the current study, the primary interest is the use of an abiotic hollow fibre (HF) membrane based gas-liquid absorption for contaminated gas cleanup. The test contaminant was hydrogen sulphide (H 2 S), and the absorbent a dilute caustic solution of different concentrations. *Corresponding author # Presenting author

2 2. Theory The microporous membrane based device operates as a gas absorber with a design similar to a shell and tube heat exchanger. The system under investigation in the current study operates in a non wetted mode and so the membrane pores are gas filled. The most common method of describing the mass transfer in the system is with a resistance in series approach based on transfer through the gas, membrane and liquid phases. For steady state conditions and gas liquid equilibrium at the membrane wall []: K d G i = k d g o k d m lm Hk d E l i Where K G is the overall mass transfer coefficient based on gas phase numbers, k g, k m, and k l stand respectively for the local gas phase, membrane and liquid phase mass transfer coefficients, and d o, d m and d l stands respectively for the outer, logarithmic mean, and inner fiber diameter. H is the Henry s law constant (the equilibrium concentration in the liquid divided by that in the gas) and E is the enhancement factor caused by a chemical reaction increasing the driving force in the liquid phase. The control of the process is therefore dominated by the solubility of the diffusing compound. For low solubility gases such as O 2 and CO 2, H is small and the liquid side resistance controls. However, for highly soluble or reactive gases (H 2 S), the liquid resistance becomes negligible and the overall rate is controlled by the resistance through the gas phase and the membrane. K G = k g k m 2. Materials and methods The schematic experimental set-up (Figure ) is similar to that used by Li [2] for H 2 S absorption and Kreulen [4] for CO 2 absorption. The HF cartridge used in this study (Table ) was supplied by Intersep Ltd (Wokingham, Berks, UK). Membrane specifications not supplied by the manufacturer have been estimated. Contact areas are based on internal and external fibre diameters, the packing fraction on the ratio of the total outer diameter fiber volume to the shell volume, and the surface to volume fraction on the ratio of the active surface area (area of transfer) to the shell volume Table : Membrane module specification Parameter Module Parameter Module Fibres material polypropylene Active module length (cm) 2 Inner diameter (µm) 33 Shell inner diameter (mm) 35 Outer diameter (µm) 36 Inner contact area (m 2 ).4 Pore size (µm).6 Outer contact area (m 2 ).4363 Number of fibres 93 Surface to volume fraction (m 2 /m 3 ) 24

3 3 = H 2 S gas cylinder (BOC gases, UK). 2 = Flow regulator (Platon, range.-.2 ml/min). 3 = Pressure gauge (Platon, range.2-5 bar), 4= Membrane module (Intersep Ltd), 5 = Pressure regulator (liquid side) (Platon, needle valve), 6= PVC Tank with absorbant (5 l, PVC), 7= Centrifugal motor-pump (Lafert,.25 KW, 36 rpm), 8= Flow regulator (Platon, needle valve), 9= Pressure gauge (Platon, range.2-5 bar), = Pressure regulator (gas side) (Platon, needle valve) and = Gas meter (Odalog, Australia) Fig. : Experimental setup The feed gas mixture containing H 2 S was taken from ppmv bulk cylinder. The gas mixture was passed through the fibre shell at the desired flow rate whilst a dilute caustic solution was pumped through the fibre lumens. The pressure on the gas side was maintained below the bubble point of the system to avoid dispersive flow. A fresh caustic solution was used during each trial with s between 7 and 3. Hydrogen sulphide was measured in the gas phase with a Jerome 63-X and liquid phase measurement were conducted why spectrophotometry after the appropriate wet chemical preparation. The mass transfer coefficient is calculated from the measurements by conducting a mass balance, in the case of gas limited mass transfer the equation simplifies to [2] [G=gas flow rate (m 3.s - ), A = transfer area (m 2 )]: G C K = G ln A C IN OUT 3. Results and Discussion 3. Removal Gas phase data revealed equilibration to be rapidly established (within five minutes). The removal efficiency of H 2 S from the gas stream upon a single pass through the membrane module revealed a strong dependency on the initial water chemistry. To illustrate, at an 2 ml.min - gas flow and an 3 ml.min - liquid flow, the removal efficiency at 7, 8, 9 and was 52, 66, 96 and 99.9 % respectively and thereafter remained at 99.9% for all higher s tested. At each individual, the removal efficiency decreased as a function of gas to liquid ratio (Figure 2). To illustrate, at a of 7 the removal efficiency decreased from 92 % to 5 % as the G:L ratios increased from 8 to 2 respectively. For comparison, at 2 the removal efficiency remained above 9 % up to G:L ratios of 4 and then decreased such that at a G:L ratio of 2 the removal efficiency was c7 %. The removal process was also seen to be independent of liquid velocity except at low water where a slight increase in removal was observed as the liquid velocity increased.

4 4 The implication of the observed results is that as expected the process is controlled by the gas phase and membrane resistances. Further, as the increases the process becomes less gas phase controlled ultimately switching to membrane resistance controlled transfer. This is in agreement with Li s studies working with very acidic water solutions [2] and operating over a velocity range of -6 m.s -. The difference in observed performance as a function of is a consequence of the fact that H 2 S is a weak dibasic acid which dissociates in water according to: H a = 7.4 pk a 2 = S H HS 2 H S 2 At 7 approximately 5% of the H 2 S is in its molecular form that can volatilise from solution and cause odour problems. Above 7 the majority of the sulphide is in the ionic form either as HS - op S 2-. At 3 a second cross over occurs where approximately 5% of the sulphide is in each ionic form. Sulphide can also be oxidised to form either elemental sulphur or sulphate depending on. At lower s it is believed that the reactions will only tend to elemental sulphur and that the reaction is very slow with oxygen unless a catalyst is present. Whereas at higher s the reaction continues to form sulphate, sulphur or sulphur dioxide. The consequence is to enhance the mass transfer as a result of the driven reaction in the liquid phase removing the build up of sulphide species in the liquid phase which would otherwise retard further transfer. Removal efficiency (%) G/L ratio Figure 2: Removal efficiency against G:L ratio hydraulic gas phase loading rate to aciveie 3. ppmv effluent gas concentration (m 3.m -2.h - ) Figure 3: Maximum loading rate to achieve 3 ppmv in the outlet gas phase The consequence of the dissociation reaction is to increase the hydraulic loading rate that the plant can be operated at whilst maintaining the same effluent gas concentration. For instance, to maintain an outlet gas concentration of 3 ppmv the maximum hydraulic loading rate increased from to 96 m 3.m -2.h - as the increased from 7 to 3, an increase of around 8 times (Figure 3). 3.2 Mass transfer analysis The overall mass transfer coefficient (K G ) was calculated as a function of both gas and liquid flow rates and under all liquid phase s. At any given, K G varied only as a function of gas flow rate and showed no variation with liquid flow rate in agreement with previous research [3] and reflects that the reaction in the liquid phase removes any hydrodynamic impacts. The average K G varied from.47x -4 to 6.45x -4 m.s - (Figure 4) with the variation in measurements increasing as the rose and reflects that as the increases the hydrodynamics become more important. The membrane resistance was evaluated experimentally by means of a Wilson plot (Figure 5) to increase from.6x -3 to 3.35x -3 m.s - as the increased from 7 to 3. Similar experimental numbers have been

5 5 reported for polypropylene and polyethersulfone membranes [2]. In all cases, including the current investigation, the experimental values were considerably lower than theoretically predicted and this is thought to be due to a combination of partial wetting of the pore and the membrane substrate offering a resistance to mass transfer [2]. No dimensionless correlations for H 2 S transfer through a membrane have been proposed previously but appropriate expressions have been published for air and carbon dioxide (Table 2). Comparison of the proposed correlations with the existing data revealed the Sirkar correlation to under predict and the Kreulen correlation to over predict the experiment data. However, this is not unexpected as the current data does not comply with the hydrodynamic range over which the correlations were intended. Correction of the original correlation to fit the current data has been achieved by adjusting the exponents a and b in the Graetz expression originally proposed by Kreulen. Appropriate fits were achieved when the exponent a increased from.6 to.46 and the exponent b increased from -.3 to.43 as the increased from 8 to 3 (Figure 6). The trend with exponent b demonstrates a linear fit of the form: b = K G x -4 (m.s - ) 2 8 /KG (s.m - ) Figure 4: Mass transfer coefficient as a function of at a gas flow rate of 5 ml.min - Figure 5: Wilson plot based on gas velocity for 3.6 b a /vg (s.m - ) v g (m.s - ) exponent b exponent a r m /R G ([/k m ]/[/K G ]) Figure 6: Variation of dimensionless number correlation exponents as a function of initial water Figure 7: Influence of membrane resistance on the total resistance to mass transfer The value of the exponent describes the gas phase hydrodynamic dependency on the transfer process. At low the exponent is small demonstrating no real impact in changing gas velocity as solubility of H 2 S in the liquid phase retards mass transfer. Whereas at high s the exponent is large and the gas velocity has a large impact on the mass transfer indicating no barrier to transfer in the liquid side. The ratio of the transfer resistance caused by the membrane compared to the total decreases for all gas velocities as the increases (Figure 7). To illustrate, at a gas velocity

6 6 of.5 m.s - the membrane transfer coefficient contributes 5 % of the total at which decreases to 23 % at 3. However, the contribution of the membrane resistance increases as a function of gas velocity such that as expected the membrane resistance contributes the major component of the total at higher velocities. Overall, the system shifts from gas phase to membrane control at high gas phase velocities or low water s. This is in agreement with Li work [2], which was conducted at much higher gas velocities, where the membrane resistance was seen to control the transfer process. Table 2: Dimensionless correlation Conditions Author Correlation range Air on the shell side, no chemical reaction Semmens [4] Sh =.22 Re Sc. Sh =.62Gz Gz > 4 CO 2 on the shell side, chemical reaction with Kreulen [5] 33 OH - This study 4. Conclusions b Sh = agz Performance of a single pass hollow fibre membrane for the absorption of hydrogen sulphide has been shown to be dependant on the gas phase velocity and the initial water. The importance of water is a consequence of the fact that hydrogen sulphide is a weak dibasic acid that dissociates in water and hence generates a reaction in the liquid phase. The impact of this is to increase the maximum hydraulic throughput that is possible whilst maintaining a constant outlet gas phase concentration. Analysis of the data reveals that the system fits a classical Sherwood-Graetz correlation with adjusted constants to reflect the impact of water. Overall the process shifts from gas phase to membrane controlled as the gas velocity increases or the water decreases. References [] Sirkar, K. K. (992) Other membrane processes, IN: W.S.W. Ho and K. K. Sirkar (Eds.)Membrane Handbook, Chapman and Hall, New York, [2] Li, K., Wang, D., Koe, C. C. and Teo, W. K. (998) Use of asymmetric hollow fibre modules for elimination of H2S from gas streams via a membrane absorption method. Chem. Engng. Sci., 53, -9. [3] Jefferson, B., Nazareno, C., Stuetz, R., Parsons, S. A. and Judd, S. (22) Membrane gas absorbers for H 2 S removal design, operation and technology integration into existing odour treatment strategies. In: Proc. Odors and Toxic Air Emissions, 28 th April st May, Albiquerque, USA, 22., pp24 [CD-ROM] [4] Semmens, M. J., Qin, R. and Zander, A. (989) Using a microporous hollow fibre membrane to separate VOCs from water. Journal AWWA, April [5] Kreulen, H., Smolders, C. A., Versteeg, G. F. and van Swaaij, W. P. M. (993) Determination of mass transfer rates in wetted and non wetted microporous membranes. Chem. Engng. Sci., 48,