Hydrogen production via hydrocarbon steam reforming and water gas shift reactions was investigated over a monolith-supported Pt-based diesel oxidation catalyst. The evaluation included comparison between constantly rich gas composition conditions and cycling between rich gas conditions and an inert stream. Analysis was performed along the catalyst length at temperatures ranging from 200 to 500 °C. During the constant inlet composition experiments, C3H6 steam reforming started at 375 °C, while dodecane steam reforming began at 450 °C, and resulted in less hydrogen produced. With a mixture of C3H6 and dodecane, hydrogen production originated solely from C3H6 steam reforming and, under otherwise identical conditions, was less than that observed with only C3H6, but higher than that with only dodecane. Hydrogen production from the water gas shift reaction was higher than that observed with hydrocarbon steam reforming and started at 225 °C. During cycling experiments, hydrogen production via hydrocarbon steam reforming was higher than that observed during the constant inlet composition experiments. This improvement was observed at all temperatures. Temperature programmed oxidation experiments performed after steam reforming indicate coke formed on the catalyst surface during steam reforming, and that the coke deposits were primarily toward the upstream portion of the catalyst. The data also show that the reason for better performance during cyclic operation is that less coke was deposited compared to that during noncyclic experiments.
Abstract The synthetic Diesel fuel oxymethylene ether (OME) is sulfur-free by nature, and due to the low soot formation, no active filter regeneration events are required, limiting the maximum temperatures seen by the exhaust catalysts to ~ 450 °C. These OME-specific ageing requirements will enable the application of new types of catalysts that cannot be used in conventional Diesel vehicles. Such new catalytic solutions will allow ultra-low emissions at a much-reduced cost and will hence contribute to the overall efficiency of the OME approach. In this contribution, we focus on CO abatement from OME exhaust. To enable an efficient evaluation of new catalysts under practically relevant conditions, a test bench was set up that can reproduce the transient temperature-, mass flow- and concentration profiles measured during real driving tests. In a first step, the transient test bench was used to compare CO oxidation over a commercial Diesel oxidation catalyst for OME- and conventional Diesel conditions. The same low-load cold-start drive cycle run with OME showed slightly lower raw emissions, but the CO emissions downstream of the catalyst increased by a factor of ~ 2. The main reason for the lower CO conversion is the lower temperature of the OME exhaust. In a second step, we investigated short-pulse reductive activation of Pt/ceria as a promising new technology that benefits from the OME-specific low ageing requirements. A Pt/ceria catalyst activated by a short 5–10 s reductive pulse achieved virtually 100% conversion even at exhaust temperatures below 80 °C. With one 5 s reductive activation pulse per 30-minute drive cycle, a CO conversion of > 99.9% is demonstrated over the low-load cold-start OME drive cycle, compared to 59% obtained with a standard commercial Diesel oxidation catalyst. To our knowledge, this is the first published demonstration of short pulse reductive activation of Pt/ceria for CO oxidation using realistic transient drive cycles.
Abstract The collisional deactivation of vibrationally highly excited azulene was studied in equimolar supercritical mixtures of xenon and ethane at 385 K from gas to liquid phase densities. Azulene with an energy of 18000 cm −1 was generated by laser excitation into the S t ‐ and subsequent internal conversion to the S* σ ground state. The loss of vibrational energy was monitored by transient absorption at the red edge of the S 3 ←S 0 absorption band at 290 nm. Transient signals were converted into energy‐time profiles using hot band absorption coefficients from shock wave experiments for calibration and accounting for solvent shifts of the spectra. Under all conditions, the decays were monoexponential. At densities below 10 −3 mol/cm 3 , the observed collisional deactivation rate constants k c of the mixture were equal to the sum of the individual contributions of ethane and xenon collisions as expected from simple gas kinetics. At mixture densities above 2×10 −3 mol/cm 3 , k c is smaller than the deactivation rate constant found in neat ethane at half the density. This behavior can be rationalized by a model employing an effective collision frequency which is related to the finite lifetime of collision complexes; the required parameters follow from experiments in neat xenon and ethane.
We demonstrate in this paper that the effective diffusion coefficient in the wall of a particulate filter can be determined by measuring the diffusion of NO between two adjacent channels of a filter segment in a simple apparatus based on a modified experimental method of Beeckman. The effective diffusion coefficient in the walls of an uncoated SiC particulate filter is determined as 2.8 × 10−6 m2/s at room temperature. Coating the filter with 30 or 65 g/L of wall-integrated washcoat leads to an insignificant increase in the diffusion coefficient, whereas coating the same filter with a high washcoat loading of 140 g/L leads to a decrease of the diffusion coefficient to 2.0 × 10−6 m2/s. All of the determined diffusion coefficients increase with temperature proportional to T1.5. This indicates that the diffusion in the wall is mainly molecular diffusion and that Knudsen diffusion plays a minor role. Fitting the parallel pore model to the experimental diffusion coefficient of the uncoated filter results in a tortuosity factor of 3.5. The random pore model overpredicts the effective diffusion coefficient by almost 50%. Neither of the two models reproduces the threshold-type dependency of the effective diffusion coefficient on washcoat loading. However, all experimental results are predicted by both pore models with an accuracy of better than 50%, so that an estimation of the effective diffusion coefficient might be a feasible solution for many practical simulation tasks.
The digitalization of chemical research and industry generates large amounts of data, offering new opportunities for developing and parameterizing kinetic models. Leveraging this data requires machine learning techniques capable of autonomously extracting kinetics from reactor datasets. Recently, neural ordinary differential equations (neural ODEs) were coupled with reactor models to learn kinetics from ideal reaction systems, such as plug-flow reactors. However, real reactor set-ups commonly feature non-idealities such as heat- and mass transfer limitations, described by partial differential equations (PDEs) or differential-algebraic equations (DAEs). Discretizing such non-ideal PDE or DAE reactor models by finite volumes yields algebraic balance equations, which are solved by numerical schemes. In this work, we propose to learn kinetics from non-ideal reactor data using these balance equations and implicit neural networks, avoiding expensive backpropagation through the numerical solution by utilizing the implicit function theorem. The approach is demonstrated for the example of a mass transfer limited flat plate reactor for the preferential catalytic oxidation of CO. We show that Global Reaction Neural Networks, embedding thermodynamic and stoichiometric prior knowledge, coupled with a discretized two-phase CSTR cascade reactor model extract intrinsic kinetics from 50 integral reactor experiments. These kinetics generalize to new reactor geometries, where kinetics obtained by neural ODEs fail. Further, the approach is robust, recovering accurate kinetics even when training data is perturbed by 10% Gaussian noise. We expect that combining neural network-based kinetics with non-ideal reactor models will broaden the scope of kinetic model discovery and improve access to accurate kinetic models.
Abstract Multi‐scale modeling allows the description of real reactive systems under industrially relevant conditions. However, its application to rational catalyst and reactor design is hindered by the prohibitively high computational cost associated with the chemical kinetics on the catalyst scale. Here, the computational cost is drastically reduced by introducing goal‐oriented kernel models that serve as surrogates for the chemical kinetics. This special model type allows for automated training set design and reliable results, even outside the training region. Therefore, it can be envisioned as a plug‐and‐play solution for accelerating reactive flow simulations with guaranteed accuracy.
