Search

The traditional concept of the Thiele or Weisz modulus can fail to assess the relevance of pellet-scale transport limitations in the case of heterogeneously catalyzed gas-phase reactions such as oxidative coupling of methane (OCM). Irreducible transport limitations occur when surface-produced gas-phase intermediates (e.g., radicals) react fast in the pores of catalyst pellets, causing strong pellet-scale concentration profiles that directly affect the product distribution. To assess the effect of the pellet diameter on pellet-scale gradients of reactants and intermediates during OCM, a pellet-scale model is used in combination with detailed kinetic models for the gas (57 species, 317 reactions) and surface chemistry (11 species, 26 reactions). On the basis of a preliminary analysis using a skeleton reaction mechanism a so-called Boudart number, Phi(Bou), is proposed for assessment of the potential relevance of transport limitations during this type of heterogeneously catalyzed gas-phase reactions. The effect of pellet-scale gradients on product yields and selectivities is evaluated using a rigorous 1D heterogeneous reactor-scale model in combination with the above-mentioned detailed kinetic models for the gas and surface chemistry. Pellet-scale diffusion limitations are found to have a significantly negative effect on C-2 selectivities during OCM with the applied kinetics. When the Weisz-Prater criterion is satisfied for reactants and products, pellet-scale gradients of the intermediates are responsible for this drop in selectivity. For larger pellet diameters, the drop in selectivity is due to diffusion limitations of the C-2 products and the corresponding increased importance of secondary total oxidation reactions.

A lateral flow reactor (LFR) is a low pressure drop fixed bed reactor especially suitable for end-of-pipe treatment, for example, the Shell DeNO(x) system for selective catalytic reduction of NOx. Inherently, an LFR is prone to flow maldistribution, which in turn may negatively affect conversion. A detailed study of the hydrodynamics and overall reactor performance of an LFR is presented. The characterization of the flow patterns is carried out in detail with a 2D computational fluid dynamics (CFD) model. In addition, a "double 1D" hydrodynamic model is developed and validated against the CFD results. This computationally much more efficient model is extended with catalytic reaction kinetics, accounting for intraparticle mass transfer limitations. LFR efficiency can be defined as the ratio of the amount of catalyst needed in a fixed bed with uniform distribution over the amount of catalyst needed in the LFR to reach the same overall conversion. Despite the inherently limited flow uniformity, this efficiency of the LFR remains higher than 95% as long as the flow uniformity is above 50%, that is, for nearly all practical cases.

Understanding ignition and extinction behavior is of crucial importance for oxidative coupling of methane (OCM). Therefore, for the first time the bifurcation behavior of OCM has been investigated while considering both homogeneous gas phase reactions and heterogeneous reactions using a detailed microkinetic model. Three different adiabatic reactor models are considered: a plug flow reactor (PFR), a continuously stirred tank reactor (CSTR) and a lumped thermal reactor (LTR) model. The latter represents the limiting case with zero backmixing (cf. PFR behavior) for species and perfect thermal backmixing (cf. CSTR behavior). For homogeneous processes this reactor type could for example be realized by adding a high thermal conductivity inert to the reactor tubes, for catalytic processes a high thermal conductivity catalyst could be used. The bifurcation behavior in these reactor types is compared with a focus on methane conversion, C 2 yields and their dependence on operating conditions such as inlet composition, inlet temperature and space time. Steady state multiplicity is observed for adiabatic CSTR and LTR models. This multiplicity of steady states is not observed for isothermal reactor models, indicating that it is caused solely by thermal backmixing and is not related to chemical feedback features such as autocatalysis. The start-up procedures or initial conditions determine the actual steady state that is obtained. Among the three investigated reactor types, a LTR shows the highest product yields and the lowest extinction temperatures, which allows autothermal operation at a much lower inlet temperature compared to a PFR and CSTR. For OCM without catalyst, autothermal operation on the ignited branch at ambient inlet temperatures and reasonable space times is only possible by using methane-to-oxygen ratios below 3 leading to low selectivities. For catalytic OCM compared to OCM without catalyst, the range for autothermal operation is much broader and it is much easier to find feasible operating conditions allowing autothermal operation at ambient inlet temperatures. By operating a LTR on the ignited branch at ambient inlet temperature of 300 K, methane-to-oxygen ratio CH 4 :O 2 = 6, space time V/F CH4,0 = 0.02 s, bulk density of Sn-Li/MgO = 1000 kg cat /m 3 and pressure P = 1 bar, overall C 2 selectivities (i.e. sum of ethane, ethylene and acetylene selectivity) of 80% can be obtained at methane conversions as high as 30%.

Oxidative coupling of methane (OCM) is considered one of the most promising routes to directly convert methane into more valuable hydrocarbons. The uncertain economics related to the tradeoff between conversion and C2 selectivities is the primary reason why OCM is currently not industrially applied. In the last decades, numerous studies have focused on developing a viable catalyst that has the potential to improve the low C2 yields. But is the primary issue of OCM truly a catalyst problem? Because of the high exothermicity of the OCM process, thermal effects and path dependence are dominating in all OCM reactors of practical importance. Furthermore, irreducible diffusion limitations exist on the pellet scale. Understanding how to exploit these mass and heat transfer effects by reactor engineering is a prerequisite for the breakthrough of OCM. In this work an overview is given of criteria used to assess mass and heat transfer resistances, parametric sensitivity and runaway in catalytic packed bed reactors. The importance of mass transport limitations, runaway and/or ignition for OCM, either on the pellet scale or on the reactor scale, is shown in several examples. Both simple and advanced reactor concepts are discussed with a focus on their heat and mass transfer characteristics. Clear progress has been made in the past lustrum on all these fronts and it seems that research is on the verge of a real breakthrough to make OCM happen on an industrial scale.

An adequate use of gas pulsation can create an ordered, dynamically structured bubble flow in a bed of Geldart B particles. A structured bed is more homogeneous, responds to external control and is scalable. While earlier studies have focused on describing the self-organization of the gas bubbles, the solid mixing and gas–solid contact patterns have remained unclear. In this work, the solids circulation and mixing behavior in structured and unstructured beds at various pulsation frequencies are compared with a traditional fluidized bed operation. The degree of lateral mixing is hereby quantified through an effective lateral dispersion coefficient extracted from CFD-DEM (discrete element modelling) simulations in a thin fluidized bed system. Mixing shows major quantitative and qualitative differences amongst the investigated cases. The coordinated motion of the gas bubbles wraps the solid flow into a series of compartments with minimal interaction, whereby effective lateral dispersion coefficients are an order of magnitude lower than in an unstructured operation. More importantly, unlike a traditional bed, dispersion in a structured bed is driven by advection and is no longer a diffusive process. Compartmentalization decouples the time scales of micro- and macromixing. Every pulse, the compartments rearrange dynamically, causing a level of local axial mixing that is scale-independent. While further work is necessary to fully understand the compartmentalization at a larger scale, the circulation described here indicates that a dynamically structured bed can provide a tight control of mixing at low gas velocities and a narrower distribution of stresses in the solid phase compared to traditional devices.

Vortex reactors take advantage of the synergy between enhanced heat and mass transfer rates and multifunctional phenomena at different temporal and spatial scales. Proof-of-concept experiments with our novel and innovative STAtor-Rotor VOrtex Chamber (STARVOC) confirm its advantageous features for the sustainable production of chemicals and fuels. STARVOC is a high-g contactor that uses carrier flow (gas or liquid) tangential injection to drive a rotor attached to low-friction bearings. The vortex chamber inside the rotor contains a secondary phase or phases, such as a solids bed, a liquid layer, or a suspension. Carrier fluid passes through the perforated rotor wall and contacts a densely and uniformly distributed secondary phase with enhanced slip velocities. Experiments focused on pressure profiles, rotor angular velocity, and solids azimuthal velocity. With air as the carrier fluid and different solid particle beds as the secondary phase, STARVOC reached bed azimuthal velocities up to four-fold compared to those reached in Gas-Solid Vortex Units with fully static geometry. These results show its potential to improve interfacial heat and mass transfer rates and take advantage of flow energy and angular momentum. Due to its process intensification capabilities, STARVOC is a promising alternative for the state-of-the-art chemical industry.

Recently computational fluid dynamics (CFD) has been successfully applied for the evaluation of the start-of-run performance of three-dimensional (3D) coil geometries in steam cracking reactors. However, determining the full economic potential of a coil involves tracking its performance throughout the run and not only at start-of-run. Therefore in this work a novel method has been developed that allows to assess the most debated characteristic of these 3D coil geometries, i.e. the potential extension of the run length in combination with the evolution of the product yields during the time on stream. An algorithm based on dynamic mesh generation is presented for simulating coke formation in 3D steam cracking reactor geometries, tracking the apparent geometry deformation caused by the growing coke layer. As a proof-of-concept, a Millisecond propane cracker is simulated over the first days of its run length, and this for three different coil designs: a bare tube, a finned tube and a continuously ribbed reactor design. Our simulations show that the ribbed reactors overall outperform the others although in these enhanced tubular geometries the growth of the coke layer is far from uniform. Because of this, the reactor geometry will change over time, which will in turn influence the fluid dynamics, product yields and successive coke formation substantially.

The process intensification abilities of gas–solid vortex units (GSVU) are very promising for gas–solid processes. By working in a centrifugal force field, much higher gas–solid slip velocities can be obtained compared to gravitational fluidized beds, resulting in a significant increase in heat and mass transfer rates. In this work, local azimuthal and radial particle velocities for an experimental GSVU are simulated using the Euler–Euler framework in OpenFOAM and compared with particle image velocimetry measurements. With the validated model, the effect of the particle diameter, number of inlet slots and reactor length on the bed hydrodynamics is assessed. Starting from 1g-Geldart-B type particles, increasing the particle diameter or density, increasing the number of inlet slots or increasing the gas injection velocity leads to an increased bed stability and uniformity. However, a trade-off has to be made since increased bed stability and uniformity lead to higher shear stresses and attrition.

Helical inserts or baffles in tubular membranes can provide flux increases without additional energy usage. However, the optimal shape of these turbulence-enhancing structures remains debated, with many experimental studies being performed, though few full-scale applications reported. A model-based approach can assist experimental work in getting detailed insights into the hydrodynamic phenomena near the membrane surface. This creates the opportunity for model-based design of such membranes or an optimization of operating conditions. Both the fluid flow and particle behavior were simulated in a single Pentair X-Flow Compact Helix membrane with helical ridge, and results were compared to those of a plain tubular membrane. Higher mean wall shear stresses were found for the membrane with helical ridge as well as a wider distribution of wall shear stresses, which promote fouling mitigation. Within a shadow zone behind the helical ridge, wall shear stress was, however, slightly lower compared to the plain memb...

Using detailed kinetic models in computational fluid dynamics (CFD) simulations is extremely challenging because of the large number of species that need to be considered and the stiffness of the associated set of differential equations. The high computational cost associated with using a detailed kinetic network in CFD simulations is why one-dimensional simulations are still used, although this leads to substantial differences compared to reference three-dimensional simulations. Therefore, a methodology was developed that allows one to use detailed single-event microkinetic models in CFD simulations by on the fly application of the pseudo-steady-state assumption to the radical reaction intermediates. Depending on the reaction model size, a speedup factor of more than 50 was obtained compared to the standard ANSYS Fluent routines without losing accuracy. As proof of concept, propane steam cracking in a conventional bare reactor and a helicoidally finned reactor was simulated using a reaction model contain...

Turbulent reactive flow simulation often requires accounting for turbulence-chemistry interactions and the sub-grid phenomena. Their complexity leads to a trade-off between computational efficiency on one hand and computational accuracy on the other. We attempt to bridge this gap by coupling the power of machine learning with the turbulent reactive flow simulation, specifically in the form of a deep artificial neural network. The Lagrangian Monte Carlo method is chosen as a demonstration case as it is one of the most accurate models for turbulent reactive flow simulation, but also one of the most time-consuming. The workflow consists of training data generation, deep neural network construction, and implementation in ANSYS-Fluent, followed by an evaluation of model accuracy and efficiency, which results in an order of magnitude faster simulation without loss of accuracy thanks to our data-driven deep neural network. This approach can be of universal relevance in speeding up time-consuming models in the field of reactive flow simulation.

The stable operation of low-temperature combustion processes is an open challenge, due to the presence of undesired deviations from steady-state conditions: among them, oscillatory behaviors have been raising significant interest. In this work, the establishment of limit cycles during the combustion of hydrocarbons in a well-stirred reactor was analyzed to investigate the role of chemistry in such phenomena. An experimental investigation of methane oxidation in dilute conditions was carried out, thus creating quasi-isothermal conditions and decoupling kinetic effects from thermal ones. The transient evolution of the mole fractions of the major species was obtained for different dilution levels ( 0.0025 ⩽ X CH 4 ≤ 0.025 ), inlet temperatures ( 1080 K ⩽ T ≤ 1190 K ) and equivalence ratios ( 0.75 ⩽ Φ ≤ 1 ). Rate of production analysis and sensitivity analysis on a fundamental kinetic model allowed to identify the role of the dominating recombination reactions, first driving ignition, then causing extinction. A bifurcation analysis provided further insight in the major role of these reactions for the reactor stability. One-parameter continuation allowed to identify a temperature range where a single, unstable solution exists, and where oscillations were actually observed. Multiple unstable states were identified below the upper branch, where the stable (cold) solution is preferred. The role of recombination reactions in determining the width of the unstable region could be captured, and bifurcation analysis showed that, by decreasing their strength, the unstable range was progressively reduced, up to the full disappearance of oscillations. This affected also the oxidation of heavier hydrocarbons, like ethylene. Finally, less dilute conditions were analyzed using propane as fuel: the coupling with heat exchange resulted in multiple Hopf Bifurcations, with the consequent formation of intermediate, stable regions within the instability range in agreement with the experimental observations.

Commercially viable OCM will depend largely on process intensification, i.e., innovative reactor and process design. Such an OCM reactor should have two characteristics: limited species backmixing and sufficient thermal backmixing. Short residence times with a narrow distribution have indicated that a GSVR exhibits at least the first of these two desired characteristics. Whether it also exhibits the second is less straightforward to verify. Our reactive CFD framework catchyFOAM is used to perform adiabatic simulations of a GSVR for OCM for a wide range of inlet temperatures (648 K – 1198 K), while fixing the inlet composition (CH4:O2 = 4), mass flow rate (3.6 10 −3kg s −1), catalyst mass (12 g) and total pressure (1, 2, 5 bar). This allowed to construct CFD-based bifurcation diagrams of the outlet temperature, conversions and selectivities versus inlet temperature. A wide window of steady-state multiplicity was simulated at 5 bar, with on the ignited branch a CH 4conversion of 37% and C 2selectivity of 72%, for an inlet temperature of only 698 K. This indicates that also the second criterion for an intensified OCM reactor, i.e., to allow steady-state multiplicity, can be met in a GSVR.