1.      Introduction

The Oxidative Coupling of Methane (OCM) represents a direct valorization route for natural/shale/bio gas resources which has captivated both academic and industrial stakeholders for almost 40 years¹. It is widely recognized that the main challenges of OCM, situated in difficult CH₄ activation combined with an inverse relationship between methane conversion and selectivity towards the desired C₂ (C₂H₆, C₂H₄) products and strong exothermicity, can be overcome only via simultaneous optimization of process layout, reactor design and catalyst composition². In fact, the only claimed commercial implementation so far, the GEMINI process by Siluria and Linde³, has introduced significant novelties in all these three fields.

Nevertheless, most of the OCM process studies currently available in literature are based on a single catalyst and a single reactor configuration. The performance indicators at the outlet of the OCM reactor are ‘imposed’⁴ or simulated with simplified kinetics⁵. Variations in CH₄ conversion and C₂ selectivity with the process conditions or catalyst composition would have a dramatic impact on the separation and recycle sections downstream and cannot be easily extrapolated to other catalysts.

In order to overcome these limitations, in the present work a multi-scale simulation framework is proposed which enables process modelling for OCM by incorporating a detailed reactor model and a fundamental kinetic model specifically accounting for the catalyst properties.

2.      Simulation framework

A schematic representation of the simulation framework developed is shown in Figure 1. Three different modelling challenges were tackled and are displayed in the figure:

A.     Process model

Process simulations were performed via the commercial software Aspen Plus. An example of a potential process flowsheet⁶ is shown in Figure 1/A. The major challenge of the present work lies in the detailed modelling of the reaction section. This was tackled by coupling Aspen Plus to an external, in-house code. The implementation occurred via a customized user unit operation model⁷ (Figure 1/B1), namely a Fortran interface which is capable of: reading input from Aspen streams and block variables, write input to the external code, execute the external code, read output from the external code, write output to Aspen streams.

B.     Reactor model

The external code solves mass, energy and momentum balances for a catalytic fixed-bed reactor (Figure 1/B1), which can be operated both isothermally and adiabatically⁸. The reactor model is 1-dimensional, i.e. no radial gradients are considered on the reactor scale, and heterogeneous, i.e. irreducible concentration and temperature gradients on the particle scale are taken into account. The two phases considered on the particle scale, and indicated in Figure 1/B2, are: the intraparticle phase, consisting of the catalyst particles and the gas contained in their pores, and the interstitial phase, accounting for the gas around the particles. Textural properties of the catalyst particles (surface area, porosity, tortuosity, diameter) are included among the model parameters.

Figure 1. Multi-scale simulation framework deployed in the present work. A. Example of process flowsheet for OCM⁶; B. Customized user unit operation model, enabling the simulation of an OCM reactor (B.1), including particle-scale phenomena (B.2)⁸; C. Simplified representation of the reaction scheme⁹.

C.     Microkinetic model

A microkinetic model for the OCM reaction, previously developed in our research group and validated for five different catalysts⁹, was embedded in the reactor model. The model accounts for 13 molecules, 11 radicals and 11 surface species, and comprises 39 reversible gas-phase reactions and 26 reversible reactions on the catalyst surface. A simplified representation of the reaction scheme is reported in Figure 1/C. The possibility of reproducing the performances of different catalysts arises from the presence in the model of catalyst descriptors, i.e. kinetically significant catalyst features - such as chemisorption enthalpies, sticking coefficients, density of active sites - that specifically account for the impact of the catalytic material on the reaction kinetics.

3.      Case study: distributed oxygen feed and intermediate cooling

In this case study, the reaction section of a potential OCM plant was simulated considering a Sn-Li/MgO catalyst⁹. The base case, corresponding to scenario A.I in Figure 2, was characterized by a single fixed-bed reactor operated adiabatically, at atmospheric pressure, with a methane-to-oxygen molar ratio equal to 10. The space time was set to 17 kg s/mol_CH4,0 via a design specification to achieve complete oxygen conversion at the reactor outlet. The inlet temperature was chosen equal to 840K via a design specification to operate the reactor with maximum temperature equal to 1273K⁸. The results obtained in terms of CH₄ conversion, C₂ selectivity and C₂ yield are reported in the first column of Table 1.

A second scenario, indicated with A.II, was simulated by considering the same process conditions as A.I, but reducing CH₄/O₂ to 5. As highlighted in the second column of Table 1, the resulting outlet temperature was way beyond the limit of 1273K, due to the higher adiabatic temperature rise associated to the higher oxygen content. This scenario exemplifies that not all sets of process conditions are possible for the operation of an adiabatic OCM reactor⁸, and, hence, was discarded.

Figure 2. Process layout of the OCM reaction section in the present case study, based on Sn-Li/MgO catalyst. The values of CH₄/O₂ reported are referred to the global methane and oxygen molar flowrates.

Scenario B.I was characterized by two Sn-Li/MgO catalyst beds in series, with distributed oxygen and intermediate cooling. The amounts of CH₄, O₂ and catalyst were kept constant in comparison with scenario A.I, but catalyst and oxygen were split in two, with 50% of the oxygen being fed to the first bed, containing half of the catalyst, while the other 50% was fed to the second bed. The inlet temperature to each bed was adapted via a design specification, again to achieve complete oxygen conversion at the reactor outlet. As it can be observed in the corresponding column in Table 1, this scenario resulted in higher inlet temperatures. This is due the fact that catalytic light-off is more difficult to achieve when methane is present in high excess⁸. The corresponding outlet temperatures were lower for both beds, due to the limited methane conversion that was obtained with such a low oxygen content. This resulted in a lower C₂ yield as compared to the base case.

Scenario B.II was analogous to B.I, this time considering CH₄/O₂= 5. In Table 1 can be observed that, compared to the base case, this process configuration resulted in an increase of +22% in the yield. The price to pay was +40K in the required inlet temperature and -19% decrease in the selectivity.

As the computational time for each simulation was in the order of 30s, the present case study could easily serve as basis for further techno-economic analyses aimed at evaluating the profitability of different configurations, such as scenarios A.I (higher selectivity) and B.II (higher yield), embedded in a full process scheme, including also the separation and the purification sections and the recycle of material streams.

Table 1. Simulation results for the present case study. For all the scenarios: p= 1bar, W_cat/F_CH4,0= 17 kg_cat s/mol_CH4. The variables of major interest are indicated in bold.

The few variations investigated in this case study were mainly aimed at demonstrating the potential of the developed simulation framework and more scenarios are currently being pursued. For instance, ongoing simulations evaluate how activities and selectivities of alternative catalysts, such as Sr/La₂O₃ and NaMnW/SiO₂, impact the optimal process configuration identified from yield and/or selectivity maximization.

4.      Conclusions

The simulation framework developed in the present work enables multi-scale modelling of the OCM process, without compromising on transport phenomena, both on reactor and particle scale, and kinetics, which is herein described in detail via a microkinetic model. The enablement of process simulations based on microkinetic models represents a step forward in the design and assessment of realistic process layouts for natural gas valorization. In fact, the possibility of evaluating the impact of process conditions, feed composition and catalyst properties on the outlet stream from the reaction section paves the way for more reliable techno-economic assessments aimed at designing an industrially-viable OCM process.

References

1.           Keller, G.E. and M.M. Bhasin, Journal of Catalysis, 1982

2.           Kondratenko, E.V., et al., Catalysis Science & Technology, 2017

3.           https://www.linde-engineering.com/en/process_plants/petrochemical-plants...

4.           Ortiz-Espinoza, A.P., et al., Computers & Chemical Engineering, 2017

5.           Godini, H.R., et al., Fuel Processing Technology, 2013

6.           Arickx, M., et al., Internal Report, 2018

7.           Tripodi, A., et al., Catalysts, 2017

8.           Pirro, L., et al., Industrial & Engineering Chemistry Research, 2018

9.           Alexiadis, V.I., et al., Applied Catalysis B, 2016