A Comparison Between Monolithic and Wire Gauze Structured Catalytic Reactors for CH4 and CO Removal from Biogas-Fuelled Engine Exhaust

The application of the wire gauzes as the catalytic supports can provide a number of advantages in biogas exhaust abatement. In this paper, a model of wire gauze structured reactor for biogas exhaust removal is proposed and model based calculations are performed to compare the wire gauze catalytic reactor with the classic monolith. The modelling bases on kinetic data experimentally obtained in a small-scale tubular reactor for cobalt and palladium (as reference) oxide catalysts doped with promoters (Ce, Pd). The heat and mass transfer characteristics of the wire gauze reactor are taken from the former studies by the authors. The simulations show that for assumed reactor parameters, a combination of the promoted cobalt oxide catalyst and the wire gauze support can give high conversion of methane and carbon monoxide.


Introduction
Nowadays, biomass utilization is becoming more and more problematic, due to its complex composition. Although the idea of the gasification of renewable sources such as forestry residue and municipal wastes is not novel, in practice waste storage rather than its utilization prevails. One possible application of biomass waste is gasification to obtain the flammable gas containing H 2 , CO and CH 4 . The biogas may then be used as a fuel in either a biogas engine or turbine to produce energy at a local level [1]. During the energy production via biogas combustion, biogas engine emits a range of pollutants such as NO x , CO, CH 4 and also some volatile organic compounds (VOCs), which must be cleaned to a level regulated by the European standards for waste incineration plants (Waste Incineration Directive, 2000/76/EC). Although there are a few possible ways of gaseous pollutants abatement, catalytic removal seems to be the most efficient in many applications (e.g. automotive or energetics). Ceramic or metallic monoliths with noble metals as an active material are the most ubiquitous catalytic reactors due to their low flow resistance and usually sufficient mass transport properties. However, the so-called ''short channel structures'' working in a developing laminar flow regime show much more enhanced mass and heat transport accompanied with a relatively low flow resistance, which makes them good candidates for the biogas engines cleaning installations [2,3]. Among them, wire gauzes have been demonstrated to be able to shorten reactor length by several times when comparing with the ceramic monoliths [4]. In spite of the fact that the idea of wire gauze-based structured reactor is not new and has been developed for ammonia oxidation to nitric acid [5], the literature data describing heat [6,7] and mass [3] transfer phenomena of wire gauzes are scarce.
In this study, a reactor composed of wire gauzes with a catalyst deposited on them, for the abatement of trace contaminants gases from the biogas-fuelled engines, is modelled and compared with a classic monolithic reactor. The sketch of the reactor arrangements is presented in Fig. 1 together with the wire gauze internal assumed for the modelling and experimentally studied before [8,9]. A series of metal oxides-based structured catalysts were prepared and tested in carbon monoxide and methane catalytic combustion (CC). The kinetic data of CO and CH 4 CC were experimentally derived using a small-scale test tubular reactor.
(a) Steel support pre-calcination in order to remove the superficial impurities the stainless steel sheets were first cleaned ultrasonically, cleaned in alkaline solution and then rinsed in distilled water. To form an alumina layer on the steel surface the sheets were calcined at 1,000°C for 10 h. It was proved that this kind of treatment causes alumina whiskers growing on the stainless sheet support containing aluminium [10]. (b) Primer deposition to improve adherence of a washcoat layer, the primer layer was deposited. The stainless steel sheets were dipped in the boehmite solution obtained using Yoldas method [11], withdrawn with controlled speed 3 cm min -1 and dried at room temperature. (c) Coating deposition to obtain the c-Al 2 O 3 , the Al(OH) 3 powder (Sigma-Aldrich 23,918-6) was calcined at 700°C for 6 h. Thus obtained alumina was dispersed in HNO 3 aqueous solution in the following proportions [10]: The solution was then vigorously stirred at 18°C in a closed vessel for 16 h. The precoated supports were dipped in the alumina solution and then withdrawn with control speed 3 cm min -1 . The steel supports pretreated in this way were then dried in a ventilated oven at 500°C for 3 h with the temperature ramp of 5°C min -1 . (d) Catalyst precursor deposition a series of composite oxide catalysts were prepared using an impregnation method. The alumina coated metal sheets were immersed into metal-nitrate(V) solution of various concentrations for 1 h. After impregnation the catalysts were dried in ambient conditions and then calcined in 500°C for 3 h in a ventilated oven. The catalyst (metal oxide) percent in the deposited alumina layer, determined from XRF (Thermo QUANT'X) quantitative analyses, is presented in Table 1.

Kinetic Tests
Catalytic oxidation of CH 4 and CO was carried out in 4 mm I.D. tubular quartz reactor CATLAB (Hiden Analytical; cf. the manufacturer's web page http://www.hidena nalytical.com). The outlet gases were analyzed with the quadruple mass spectrometer and Thermo FT-IR spectrometer equipped with the Gasera PA101 photoacoustic gas analysis module. Atom mass to charge values, m/z, used to detect the oxidation products and substrates were as follows: methane (16), water (18), oxygen (32), carbon monoxide (28), carbon dioxide (44). Prior to the catalytic tests catalyst samples were oxidized in synthetic air flow (Airproducts) at 500°C for 1 h. The reaction mixture contained 4,000 ppm CH 4 or 4,000 ppm CO in the synthetic air (calibration gas, Airproducts). The total flow rates of reaction mixtures were: 25 and 80 cm 3 min -1 during CH 4 and CO oxidation, respectively. The kinetic parameters used for reactor modelling for both CC reactions considered are provided in Table 1. They were calculated based on the approximation of a tubular reactor model assuming first order kinetics for both reactions, which is discussed below. All the kinetic experiments were repeated 5 times and the average value was taken for the Arrhenius plot. The error of kinetic constant k estimation from a single kinetic experiment never exceeded 9 % referred to the average value at a given temperature.

Reactor Modelling
The plug-flow model of the wire gauze reactor, which was derived and experimentally confirmed in [8], neglected heat evolved during the reaction, thus also the energy balance, due to extremely low concentration of VOCs. In this paper the model has been improved by introducing the energy balance. The more general discussion on the reactor modelling can be found for example in [12].

Material Balance Equations
The steady state material balance of reactant A in a heterogeneous catalytic reactor with the boundary conditions (B. C.) is as follows: In this model, homogeneous reactions are ignored. Mass transfer is balanced by the reaction at the catalyst surface: The term g, represents the effectiveness factor of catalyst. The effectiveness factor, g, and the Thiele modulus, /, are described by Eqs. (3) and (4), respectively: The effectiveness factor and the Thiele modulus were calculated for each considered catalyst ( Table 1). The characteristic dimension, l, i.e. the thickness of the porous catalyst layer, was determined from SEM photographs of the wire gauze samples and in each case l & 20 lm.

Energy Balance Equations
When considering the chemical reactions on the catalyst surface, especially those of high exothermic effect, the energy balance has to be introduced to reactor equations. The energy balance for the bulk and solid phases can be written, respectively:

Reaction Kinetics
Palladium is the most commonly used catalyst for the methane oxidation [13]. It was proved, that the methane oxidation over Pd/Al 2 O 3 strongly depends on oxygen content. However, under oxygen-rich conditions, the catalyst surface is fully covered with oxygen, thus the rate expression with respect to O 2 can be neglected [13]. When considering the reaction order with respect to CH -4 , the order varies between 0.45 and 1.2 depending on the catalyst system. According to Lee and Trimm [13], the reaction order over Pd/Al 2 O 3 in excess of oxygen equals unity. Oxidation of carbon monoxide over the precious metals was a subject of surveys of many groups, and provided numerous kinetic behaviour patterns. The literature provides many mechanisms of CO oxidation including the Langmuir-Hinshelwood-Hougen-Watson (LHHW) mechanism for CO oxidation over alumina-supported platinum catalysts [14] as well as the Eley-Rideal mechanism over aluminasupported palladium catalyst [15], with a CO desorption as the rate limiting step. According to the literature data, CO desorption may become very slow below 450 K [15], which is not the case in our modelling. Within the temperature range of 200-400°C the rate equation simplifies to the first order with respect to CO and zero order to O 2 [15].

Modelling Conditions
Two different types of catalyst support were modelled and compared: (a) wire gauzes, (b) classic monolithic support. The heat and mass transfer correlations applied during reactor modelling are presented in Table 2 together with the geometric parameters of the wire gauze and monolith.
The heat and mass transfer characteristics (Sh, Nu) were described using the so-called heat or mass dimensionless channel length L ÃH (L ÃM ): Sh The physical and chemical parameters of the gas mixture used during the modelling (e.g. density, viscosity, heat conductivity and heat capacity) were calculated at the local temperature of the gas phase.

Reactor Performance
In summary, the following modelling assumptions were applied: • For the reaction: CH 4 ? 2O 2 ? CO 2 ? 2H 2 O the heat of reaction at the catalyst surface was calculated to be, DH°R = -803 kJ mol -1 , • For the reaction: CO ? 0.5O 2 ? CO 2 the heat of reaction at the catalyst surface was calculated to be, DH°R = -283 kJ mol -1 , • Properties along the channel varied as a function of the local temperature, • The inlet gas temperature was assumed to 673 K, • The gas superficial velocity was assumed to 1 m s -1 ; the corresponding Reynolds numbers are given in Table 2.
The modelling results for both methane and carbon monoxide combustion are presented in Fig. 2.
When analysing the kinetic results for both reactions, the external and internal mass transfer resistances are evidently significant. This results in low value of activation energy (Ea) (cf. Table 1), which is not surprising as the kinetic tests were performed in small tubular reactor where the steel sheets with deposited catalyst were randomly distributed. In this way, the kinetic parameters obtained should be treated as apparent and they will be used for exclusively simulations in order to compare different reactor internals and catalysts. Similar effects of mass transport limitations have already been reported for methane oxidation by a number of authors [19][20][21]. The results of the simulations performed for methane and carbon monoxide combustion over the cobalt and palladium catalysts applied for the wire gauze and monolithic reactors are presented in Fig. 2a-d. In order to assess the performance of the reactors, maximum achieved conversions are compared at the maximum reactor length used for modelling. The differences between the considered reactor internals are substantial. It can be noted that the wire gauze internals enables to achieve almost 100 % conversion in much shorter reactor than the monolith, which is especially evident for the Pd0.001 catalyst (8 wt%) during methane combustion. The most important factor influencing this is higher catalyst amount arising directly from higher specific surface area of wire gauze. The impact of better mass transfer intensity is also significant, especially for very fast catalytic reactions (i.e. for very active catalysts).
A comparison of the gas hourly space velocity (GHSV) necessary to achieve final conversion of 50 % is presented in Table 3. As it can be noted, for all the catalysts considered the GHSV of the gauze reactor is 6 till 8 times higher comparing with monolithic reactor with the same catalyst. Upon referring to the ratio of specific surface areas amounting to 6.1, the impact of better mass transfer (of the wire gauze reactor) is distinct.
It should also be inferred from Fig. 2 that the catalysts are very different in their performances. The outstanding activity of the Pd0.001 reference sample is confirmed. However, high surface area and high mass transport of wire gauzes allows for exploitation of low-loaded Co1Ce1 catalyst in methane combustion. This catalyst exhibits similar activity to highly loaded cobalt sample Co1 (Fig. 2a). For the Co1Ce1 catalyst, 80 % conversion can be achieved in 1 m long reactor filled with wire gauzes, while for the monolith, the same conversion would require 4 m long  reactor (cf. Fig. 2a, b). During CO combustion, similar performances can be noted for the Pd0.001 catalyst (8 wt% PdO) and Co1Pd0.001 catalyst (1 wt% PdO). Again, Co1 catalyst performance in wire gauze reactor is worth emphasising. The temperature distribution is not shown in this study. However, temperature increases inside the reactor due to heat evolved during the reaction (assuming negligible heat losses). The gas-solid temperature difference was significantly lower for the wire gauze reactor than for the monolith (on average, 7 vs. 83 K, respectively). This results in less hazardous overheating of the catalyst when using wire gauze support. The problem is of high importance for the methane CC due to high reaction heat.
A question arises about accuracy of the modelling performed. The model together with the transfer coefficients and the kinetic equation used (although for another catalyst composition and for combustion of n-hexane) was successfully tested in a large laboratory-scale reactor (up to 10 Nm 3 h -1 ). The experimental results were in excellent agreement with the modelling; the details are provided in [8,22].
The reactor evaluations and comparisons presented here should be treated as rather preliminary. The kinetic data were derived using artificial air thus the influence of other components appearing in the exhaust (H 2 O, N 2 , O 2 , CO 2 ) may influence the results to a certain degree. The mass transfer limitations which occur during experiments were already mentioned. The modelling is, in fact, strictly valid only for the assumed reactor parameters such as temperature, gas velocity, etc. (see model assumptions above and Table 2). However, the results indicate on a very promising way for the CC intensification: wire gauzes display large specific surface area and intense mass transfer (see Table 2) which is coupled with layered catalyst of high activity and high efficiency factor (due to thin layers, see Table 1).

Conclusive Remarks
The simulations presented here proved decisive for the evaluation of the catalytic reactor inner structure. It has been demonstrated that both catalyst activity and reactor structured internals play important roles in tuning the overall reactor performance. Indeed, it has been demonstrated that the wire gauze internals even with metal oxide (non-nobel) catalysts offer undeniable benefits, which can be exploited in the CC applications in biogas engines. The wire gauze internals enable significant reactor shortening compared with ceramic monolith, by 2 till 10 times depending on the final conversion required and the catalyst used.
Although low loaded cobalt oxide catalysts did not show any methane conversion, small addition of cerium (0.01 wt%) gave a profound enhancement of cobalt catalyst activity. It is also worth noting that during CO combustion low loaded cobalt oxide catalyst (\10 wt%) enabled to reduce the amount of Pd by 8 times achieving the same conversion.
Remarkable results predicted for the wire gauze reactor by the reliable modelling show this reactor design as promising for many applications, especially environmental ones.