An adsorptionlreaction process for the purification of biogas prior to its use as energy vector L.V.-A. Truong & N. Abatzoglou Department of Chemical Engineering, University of Sherbrooke, Canada Abstract Biogas is produced both naturally in landfills and by fermentation of organic solid, semi-solid and liquid waste. When the biogas is valorized energetically it produces CO2, which is approximately 20 times less "GHG" than CH4, its main combustible compound. Consequently, the most environmentally appropriate way for the management of biogas is its use for the production of energy. Raw biogas contains a number of undesired contaminants like H2S and NH3. The combustion of these contaminants produces SO, and NO,, known for their detrimental role in atmosphere and human health. This work presents the development at lab scale of a chemical adsorption technology for the removal of the H2S. The main phenomenon is an irreversible chemical reaction between the solid and the gas phases. The study produces data on the efficiency of the process as a function of a number of variables including: n The nature and properties of the adsorbent. The biogas flow rate and the contact time of the gas with the adsorbent. 0 The geometry of the adsorption columns and the linear velocity of the flow. D The concentration of the contaminant (H#) and humidity in the biogas. The results reported focus on a promising adsorbent available commercially and include the "breakthrough curves and a first attempt to model mathematically the phenomenon. It has been shown that the surface reaction is the limiting step and that the rate is first order with respect to the H2S concentration while zero order kinetics prevail for the solid reactant (adsorbent). The reported data constitute the basis for the scale-up of the unit at a commercial level.
44 Energv and the Environment 1 Introduction Anaerobic Digestion (AD) of organic waste produces a biogas with a high concentration of methane (CH4). The biogas formed in AD plants consists of 55-80 vol% C&, 20-45 vol%c02, 0-1.5 vol%h2s, 0-0.05 vol% NHS and it is saturated with water [l]. This biogas has a high energetic value and when it is valorized energetically, it produces CO2 which is a gas approximately 20 times less damageable (Green House effect Gas) than C& for the atmosphere; consequently, the most appropriate way for the management of biogas is its use for the production of energy. Sulphate reducing bacteria grow in the digester and use acetic or propionic acid to produce H2S. This step occurs simultaneously with the methane production [2]. The generation of H2S poses serious problems of odour, toxicity for human and animal health and corrosion. Additionally, hydrogen sulphide is known to be extremely reactive in the presence of ferrous alloys and is a hydrogen embrittlement source (i.e. steel). Whenever used for the production of energy, the biogas has to be conditioned because the combustion of hydrogen sulphide produces SO,, known for their detrimental role in the atmosphere and human health. Biogas contaminated with hydrogen sulphide can be purified by various methods [l, 31 summarized in the Table 1. Table 1: Methods available for H2S removal. 2: simplicity (+: ~ esj 4: ~nvironmental impact (+: Low) Many small scale AD plant have tried to use biogas as an energy vector but some projects ended because of the corrosion and the high maintenance cost of the process [4]. The choice of the H2S removal process is, thus, site specific. In the case of this study, the process is designed for a farm-scale unit. As the process has to be simple and the product must be easy to handle and environmentally safe, the adsorption using an H2S scavenger was chosen.
Energy and the Environmmt 45 2 Materials and methods 2.1 Materials A commercially available adsorbent was chosen for this study. This adsorbent is a combination of iron oxides (Fe203, Fe304) and an activator oxide attached to a calcined montmorillonite carrier matrix and which is enhancing catalytically the reactive adsorption phenomenon. The amount of activator is 0.125-5%w/w of the adsorbent. The activator is one or more oxides of a group consisting of platinum, gold, silver, copper, cadmium, nickel, palladium, lead, mercury, tin and cobalt oxides [5]. The material has a relatively high internal porosity (&=0.75). The particle diameter varies between 4.0 to 6.5 mm. The apparent density of the dry adsorbent is 1000 kg/m3. The experimental data have proven that 1 g of adsorbent can adsorbed up to 0.1 1 g of H2S. The specific surface is 5.4 cm21g. The adsorbent is environmentally safe in both non-reacted and ready to disposal forms. It can also be used as fertilizer. 3-WAY VALVE FOR SAMPLING FLOWMETER FIXED BED ADSORPTION COLUMN SOLUTION H,S GAS CYLINDER TO PURGE NaOH SOLUTION Figure 1: Schematic of the experimental set-up. 2.2 Apparatus and procedure At the first phase of this work, the H2S adsorption from a compressed gas mixture (H2S: 3000 ppmv or 10000 ppmv, CO2 29% and balance CH4) is being studied as function of contact time, column geometry, linear velocity, initial H2S concentration and water content in gas. The adsorption of hydrogen sulphide is studied under dynamic conditions in columns of 1.5 and 2.5 in (3.81 et 6.35 cm) diameter at constant gas flow rate of 201 of gas per hour, at atmospheric pressure and ambient temperature. The contact time between the gas and the media is changed by varying the height of the adsorbent bed. Two contact times were tested (30 and 60 S) so far. Two different H2S content gas mixtures, both water saturated and dry have been tested. The saturation, wherever required, was done
46 Energv and the Environment by bubbling the initial dry gas through a hydrogen sulphide saturated aqueous solution, prepared in the laboratory. The solubility of H2S in water at 20 C is 3980 mg/l of water. The specific gravity of H2S is 1.2. A 100% H2S gas stream was bubbled through the water at a known flow rate and for a fixed period of time to ensure that the right amount of H2S was adsorbed in the water and saturate it. The first target of the experimentation was to explore which is the limiting step of the phenomenon (diffusion or reactive adsorption at the surface of the adsorbent) and to evaluate the breakthrough curve. A list of the experiments is presented in Table 2. Table 2: List of experiments.... H2S... conc.......... Contact........... time. l..... Column..... diameter........... 1... Water.... saturation I................... The experimental protocol is as follows: A chosen gas flow rate of 201 per hour was fed through the fixed bed of the adsorbent media. The linear velocity and the composition of the gas (H2S and water content) were controlled. The linear velocity was controlled by the choice of the diameter of the column, (0.18 cm/s for the 2.5 in diameter column and 0.49 c ds for the 1.5 in diameter column) the H2S concentration depends on the mixture of gas chosen, the water content depends on if the bubbling through the hydrogen sulphide was used. The hydrogen sulphide concentration at the inlet and at the outlet of the fixed bed column was analysed. The analyses were done either with a GC PE Photovac: Voyager portable digital gas chromatograph or precision detector tubes (colorimetric tubes) from Matheson-Kitagawa. The two methods of analysis gave the same results and the same range of precision and sensitivity; thus, a decision was taken to perform all analyses by the colorimetric method for brevity reasons. This decision was also taken on the grounds that there are no possible interferences from other gas contaminants because the gas used was a simulated biogas with precisely known qualitative and quantitative composition. The analysis was performed at regular time intervals. The gas at the exit of the
Energy and the Environment 47 column is bubbled through a NaOH solution to remove the hydrogen sulphide still remaining in the mixture and the so cleaned gas is purged. Since the reaction is irreversible and media could not be regenerated, at the end of its experimental run, the media was disposed of. 3 Mathematical description 3.1 Mass transfer kinetics The mass transfer kinetics are calculated from the following equations: Where Re was calculated with the linear velocity of gas and the average diameter of the media particles. Diffusivity and viscosity are calculated at P=l atm and T=293 K for gas mixtures using the equations proposed in [6]. DH2S-gas : 0.1413 cm2/s p,.,: 1.267 X 10.'~ g 1cm.s p,,,: 0.001 glcm3 dp: 0.5 cm &p: 0.75 ap: 9 cm-' By comparing these results on diffusion time with the time measured for the global adsorption process to take place we can conclude that the mass transfer occurs almost instantaneously in the column and that it should not be the limiting step of the phenomenon. Table 3: Results. Linear velocity of gas u2.5in: 0.18 C ~ S ~1.5 in:0.49 cds k mass transfer k~: 0.98 C ~ S kd: 1.25 cmls 0 mass transfer 0=0.1135 S O= 0.0891 S
48 Energv and the Environment 3.2 Mass balance for reaction The proposed model is a semi-batch stirred reactor where A is the H2S and B, the available site for reaction in the media. The reaction is considered irreversible A+B + C. The mass balance is : FA IN = FA + consumption of A in reactor FA,IN - FA = consumption of A in reactor FA,IN V - FA = -TA V= k caa cbp According to this model the B is placed inside the semi-batch reactor and it is consumed with time while the A is fed into the reactor continuously. FA,IN is a constant for each experiment but FA changes with time. In this work we have considered that the semi-batch reactor behaves as a completely stirred reactor (STR) while in reality the column behaves as a semi-batch Plug Flow Reactor (PFR). Thus, we expect to predict efficiencies lower than in real adsorption experiments. To respect the completely stirred reactor hypothesis, CA ought to be equal to the CA at the exit of the reactor. Since this is no sufficiently close to the reality of the experimental runs and as there are no data so far to analyze the system as a PFR we decided to use CA=(C~~~-C~~~~) instead of CA=CAout, which is a more realistic approach. Two more options are under development: 1) use an average CA value equal to (CAin+CAout)/:! and 2) sample the gas stream along the column and develop the kinetics based on a PFR behaviour. -TA= reaction rate (mg H2S reacted/cm3 adsorbent*hr) k = reaction constant (hi') CB = number of reactive sites in adsorbent (sites per cm3 adsorbent) a = order of reaction of CA p = order of reaction of CB V = volume of adsorbent (cm3 of adsorbent) 3.3 Estimation of the kinetic parameters The results of these calculations allow us to determine the limiting step of the overall reaction rate of adsorption and to propose a preliminary modeling tool. Our hypothesis is that the reaction rate is a function of the H2S concentration and the sites available for the adsorption. The data used to calculate the available adsorption sites follow: based on the available media saturation data, provided by the adsorbent seller, it has been calculated that lg of media contains 1982 X 10" available adsorption sites. The basic assumption in these calculations is that each reacted molecule of H2S consumed one site. The quantity if media (167 g) was saturated COUT/CIN =l and by calculating the mass of H2S adsorbed (18.7 g) the number of active site was found. A kinetic equation is proposed for describing
Energy and the Environmmt 49 the reaction rate - ra= k CAu cbp, A non-linear regression algorithm was used to evaluate the best-fit kinetic parameters of this equation. The regression gave the following results: -TA = 0.0243 c ~ cb0.03 ~ ' ~ ~ ~ The model proposed is suggesting that the reactive adsorption rate depends only on reaction kinetics and that the reaction is close to 1" order with respect to the H2S concentration and zero order with respect to the media. I Experimental versus Model Figure 2: Experimental data versus model value. 4 Results and discussion Experimental runs were performed to examine the effect of the process parameters on the reactive adsorption process. The parameters tested are: contact time, geometry of the column and linear velocity of the flow, initial concentration of hydrogen sulphide and the water content in the stream. The runs performed led to the following results: 4.1 Contact time It has been proven that the chemical reaction is the limiting step. Contact time proved to be the most important parameter. Nevertheless, even with a contact time of 60 S, the breakthrough curve resembles the ones characterizing the mass transfer controlled adsorption processes (no-pulse like adsorption front) and a part of the H2S is not retained by the bed and exits the column since the very first hours of the process.
50 Energv and the Environment l Breakthrough curves c 0.25-0 2 0 6 0.15 0.10 0.05 D00 O:00:00 WW:W 48:OO:W 72:OO:OO 96:OO:W 120:W.OO Time lhrsl Figure 3: Breakthrough curves for the two contact time. Data from Figure 3 were obtained by varying the height of the bed. It can be observed that the exit concentration is two times lower when the contact time is higher. 4.2 Column geometry and linear velocity of the gas The following results are obtained by maintaining the same flow rate (201/h), contact time (30 S) and initial concentration (10000 ppmv) and varying the bed diameter. Thus for the same contact time the gas stream has a lower linear velocity. l Breakthrough curves Figure 4: Breakthrough curves for the two linear velocities. Figure 4 shows the data obtained after 48 hours of operation. It is clearly shown that, with all other parameters kept constant, the column geometry and the linear velocity of the gas have a high impact on the efficiency of the reactive adsorption. Although this seems contradictory to the fact that the process is proven to be under kinetic control, the Reynolds number for the same flow rate is inversely proportional to the diameter and consequently the flow is more laminar leading to a more significant deviation from PFD calculations. Thus, for the same theoretical contact times the larger column allows a higher average contact time and higher adsorption efficiency.
Energy and the Environmmt 5 1 4.3 Initial concentration of H2S There are two ways to compare the results in this section. The comparison can be based on the concentration at the exit of the column after a period of time or the ratio COUT/CIN for the same period of time. When the contact time is 60 s and the linear velocity is maintained at 0.18 cm/s, for the first 75 hours of operation, the ratio COUT/CIN was similar for both concentrations. l!::p Breakthrough cuwes 030, 0 I0 0 05-1W00 ppmv -m ppmv 000 00000 12W00 210000 36WW 48WOO SOWW 720000 840000 Figure 5: Breakthrough curves for the two concentrations. Figure 5 shows the results of the breakthrough curves. It means that the removal efficiency of the media is the same. This is in agreement with the developed kinetics, which predict that the rate of the adsorption is first order with respect to the sulphide concentration. The percentage of hydrogen sulphide removed is. constant. If we compare the exit concentration, the one with the looooppmv initial concentration is 3-4 times higher than the one with 3000 ppmv. If the comparison criterion is an outlet hydrogen sulphide concentration as low as possible, the reactive adsorption process is more efficient with a lower H2S concentration at the entrance. For a contact time of 30 S, the results are different. The ratio COUT/CIN for H2S 10000 ppmv is twice as high as those for the ratio for H2S 3000 ppmv. At 125 hours of operation, the media that treated the CIN=lOOOO ppmv was completely saturated and the media with CIN=3000 ppmv was only half consumed. This is another proof that the process is controlled by the surface reaction kinetics. 4.4 Water content in the stream The presence of water enhances the reaction between H2S and the media. Data from the experiments prove that under the same conditions (3000 ppmv, 60 s and 0.49 cmh) the exit concentration of the treated gas was 10 times higher when the gas entered dry versus when it was saturated with water.
52 Energv and the Environment Table 4: Experimental results. Time (h) Oh00 25h00 46h00 7 1 h00 Conc.H2S exit (saturated) (PP~V) 0 0 180 275 Conc.H2S exit (dry) (ppmv> 0 45 1400 2200 This proves that the water participates in the chemical reaction with the media. The same profile was observed under other conditions (10000 ppmv, 30 s and 0.18 crnls). In the next steps of our work it will be examined whether the water acts as catalyst or as reactant; in the latter case kinetics must take into account the presence of the water which is now hidden in the reaction constant (k). 5 Conclusions The ideal conditions for the process studied are: a contact time in the column as long as possible; a linear velocity as small as possible; a water saturated gas. Since the calculated diffusion rate is much faster that reaction, the physicochemical phenomenon taking place is controlled by the surface reaction kinetics. A preliminary kinetic model has been developed predicting that the rate of the process is first order with respect to the H2S concentration. Additional work is underway to refine the model by measuring concentration profiles along the reaction column and include laminar flow calculations to the ideal PFR system considered. References Schomaker A, Boerboom H & Visser A. Anaerobic digestion of agroindustrial wastes: information networks. Technical summary on gas treatment, August 2000. Vavillin V A, Vasiliev V B, Ritov S V & Ponomarev A V. Simulation model methane as a tool for effective biogas production during anaerobic conversion of complex organic matter. Bioresource Technology, 48(3), pp 1-8, 1994. Nag1 G. Controlling H2S emissions. Chemical Engineering, 104 (3), pp 125-131,1997. Lusk P. Methane recovery from animal manures: the current opportunities casebook, NRELISR-25 145, Golden Colorado, 1998. Scranton, Jr., Delbert,C. Process and composition for increasing the reactivity of sulfur scavenging iron oxide, US Patent 5,792,438 accepted August 1998. Bird,R.B, Stewart, W.E & Lighfoot, E.N. Transport Phenomena, John Wiley & Sons: New York, 1960.