Читать книгу Engineering Solutions for CO2 Conversion - Группа авторов - Страница 37

Utilization of carbon dioxide with chemical conversion entails its use as a feedstock in order to transform it into other valuable products such as polymers, fuels, methanol, pharmaceuticals, urea, etc., which otherwise would need to be manufactured by using petrochemicals.

A reaction of particular importance in carbon dioxide utilization with chemical conversion is the Sabatier reaction, whereby CO_x is converted to methane by hydrogenation and subsequently introduced into the gas grid:

(2.3)

CFD simulations of the Sabatier reaction are single phase, which is an advantage from the perspective of the computational resources needed, although they require a multispecies approach and a careful selection of the turbulence model in those cases where the Reynolds number is high. CFD can play an important role in research oriented towards the implementation of different catalysts to accelerate the production of methane. The general approach followed in that case is to obtain experimental data on the reaction kinetics first and introduce them subsequently into the simulation set‐up.

A recent example of the modeling of the Sabatier reaction combined with the water gas shift reaction in a microchannel reactor was presented by Engelbrecht et al. [59], who used the commercial software COMSOL Multiphysics to perform their simulations. They assumed equally distributed flow within the microreactor, and therefore, only one microchannel was considered. To successfully model the process, the computational domain was divided in two zones: a free flow region, where the reaction kinetics without the catalyst was defined, and a porous layer, where the modified reaction kinetics was introduced to accommodate the effect of the catalyst. Figure 2.5a–c shows three parity plots between their numerical and experimental results, with fair agreement between them. Figure 2.5d shows a velocity contour plot and indicates the two areas mentioned above (the porous layer marked as a catalyst layer in the plot and the free flow zone at the center), whereas Figure 2.5e displays the computational mesh used. The reaction kinetics published earlier by Ohya et al. [60] were introduced in the software. In their experiments, the catalyst layer was formed using 8.5 wt% Ru–Cs/Al₂O₃. The model was successful in predicting the CO₂ conversion percentage as a function of working temperature and pressure, which were compared to the data obtained experimentally.

Figure 2.5 (a) Parity plot showing the match between the numerical and the experimental results at atmospheric pressure, (b) at 5 bar, and (c) at 10 bar, (d) contours of axial velocity (legend in m s⁻¹) at 250 °C and atmospheric pressure, and (e) image of the computational mesh used to obtain the numerical data.

Source: Engelbrecht et al. [59]. © Elsevier.

Engelbrecht et al. [59] thus used already published data for the reaction kinetics and introduced them on their numerical model, which was subsequently validated by direct comparison with their own experiments in terms of carbon dioxide conversion percentage. The authors not only confirmed their numerical setup as a useful tool, but also confirmed the data of Ohya et al. [60] as an adequate way of describing the reaction kinetics of the methanation reaction in the presence of 8.5 wt% Ru–Cs/Al₂O₃ catalyst. The added value of their simulation was therefore to spare the experimental load needed otherwise to obtain the reaction kinetics and on the other hand the insight provided in terms of variable contours and quick parametric sweep possibilities.

A similar methodology was developed by Alarcón et al. [61], who used their own experimental setup to obtain the reaction kinetics and introduced it into the simulation. They developed their model using the commercial software ANSYS Fluent and tested the effect of a catalyst formed with 15 wt% Ni as the active component and ceria as the promoter. Similarly, they only simulated one channel and defined both a free flow and a porous zone where the reaction kinetics were modified in order to account for the presence of the catalyst.

In both instances, the effect of the catalyst was thus indirectly considered by measuring the reaction kinetics and introducing their defining parameters in the simulation set‐up. The approximation of introducing the modified reaction kinetics in a porous bed is thus a well‐established trend in the literature, which can yield valuable insight into how a particular catalyst affects the development of a carbon conversion reaction.

Other numerical techniques might be useful for the numerical study of the methanation reaction in other types of reactors. In those cases where the methanation reaction occurs in a fluidized bed of catalytic powder, the Euler–Euler method and the more computationally expensive CFD‐DEM approach can be applied. Liu et al. for instance implemented the methanation reaction of carbon monoxide plus the water shift reaction in an Euler–Euler model of a fluidized bed using the solver twoPhaseEulerFoam, within the open‐source CFD library OpenFOAM. The main code was modified to accommodate the kinetic model proposed by Kopyscinski et al. [62]. The effect of feed composition and the catalyst inventory on the concentration profiles across the fluidized bed were obtained and compared to the experimental data of Kopyscinski et al. [63] showing fair agreement. Another instance of the application of the Euler–Euler method to the study of the methanation reaction was presented by Sun et al. [64], who used the specific purpose software MFIX to study the methanation reaction of carbon monoxide in a fluidized bed using the Euler–Euler approach. The effect of operating parameters and catalyst inventory are investigated and the reactor will get eventually optimized. An example of the use of the coupled CFD‐DEM model to analyze the hydrodynamics within a methanation fluidized bed reactor was presented by Wu and Tian [65]. DEM provides further insight as a result of the assessment of the behavior of single solid particles within the bed at the expense of a substantially increased computational time.