In this work, a new approach towards converting high lipid content biomass into renewable hydrogen is investigated. Using canola oil as the representative biomass, hydrogen was obtained by employing supercritical water as a reaction medium in a continuous gasification system over nickel catalyst. The experiments were conducted at fixed temperature (400°C), pressure (25 MPa), reaction time (residence time 45 s, or weight hourly space velocity 25.5 h⁻¹ for catalytic reaction) and varying feed concentrations (2–5 wt%). The results of the study revealed that increasing the feed concentration had a negative effect on hydrogen yield due to less water molecules present in the reactant. Another interesting observation is the appearance of a solid-like emulsion, composed of products from polymerization reaction. The appearance of this emulsion is strongly correlated to the decline of catalyst activity. Further characterization of the nickel catalyst used after the reaction found a significant decrease of its surface area. Moreover, deposition of carbonaceous compounds on its surface was detected, which could have inhibited the catalytic activity.

The effect of K, V, Mo, and La addition to Ru/Al₂O₃ on its performance in CO methanation removal from simulated hydrocarbon reformed gas was investigated for polymer electrolyte fuel cell (PEFC) applications. The addition of V or Mo enhanced the CO methanation removal activity, while that of K and La did not. CO removal from the simulated reformed gas to levels below 20 ppm was achieved using Ru/V/Al₂O₃ and Ru/Mo/Al₂O₃, while with the use of Ru/Al₂O_3, the lowest achievable level was 100 ppm. However, the addition of V or Mo to Ru/Al₂O₃ increased the methanation activity of not only CO but also CO₂. Therefore, the reaction temperature window, wherein CO removal can be performed with high selectivity for the methanation of CO vs. CO₂, was narrow.

A recirculation microreactor system for gas–liquid reactions has been developed, in which gas is continuously introduced into liquid, in order to effectively perform reactions with longer reaction times under slug flow at the desired volume ratio between gas and liquid. The slug flow was produced in a T-shaped mixer with an inner diameter (ID) of 0.5 mm and the obtained solution collected into a 20-mL (2.0×10⁻⁵ m³) vessel through a tube with an ID of 1 mm was introduced into the mixer again. This system was evaluated by the absorption of carbon dioxide into water. At first, this absorption with the conventional batch method was performed. It was clarified that pH decreased with time by introducing carbon dioxide. Then, the recirculation microreactor system was applied to this absorption. The solubility of carbon dioxide into water reached above the solubility equilibrium of 1.67×10⁻⁵ mol/L (1.67×10⁻⁵ kmol/m³) with both this system and the conventional batch method. However, the solubility with this system increased faster than that with the conventional batch method. The absorption rate with this system was improved by over three times compared to that with the conventional batch method. This tendency with this system and the conventional batch method was in good agreement with the reported tendencies. Furthermore, a mass transfer capacity coefficient, K_La had a peak at the flow rate of the aqueous solution of around 2.5 mL/min (2.5×10⁻⁶ m³/min), which was closer to that of carbon dioxide of 1.8 mL/min (1.8×10⁻⁶ m³/min). It would be better to perform this absorption under slug flow at a smaller volume ratio regardless of the total volume ratio between carbon dioxide and water. It was concluded that our recirculation microreactor system could effectively perform gas–liquid reactions with appropriate reaction times under slug flow at the desired volume ratio between gas and liquid.

Inexpensive glass capillary microfluidic devices with co-flow geometry were prepared to generate highly monodispersed microbubbles with a controlled size of less than 10 µm for medical applications. In the analysis, the dimensions of the injection and collection capillaries were decreased, and the gas pressure and liquid phase flow rate were increased to their respective limits. The diameter of the monodisperse microbubbles (having a coefficient of variation of less than 1.5%) could be controlled within the range of 2.4–9.2 µm, while maintaining a generation rate on the order of 10⁵ bubbles/s by adjusting the inner tip diameter of the injection capillary and the liquid phase flow rate. The inner diameter of the collection capillary was 30 µm, and the gas pressure was 0.50 MPa.

Metallic tungsten nanoparticles have attracted considerable attention because of their unique structures and excellent performances. This paper describes the synthesis of a mixture of tungsten metal and oxide nanoparticles via a fuel-rich tubular flame process, followed by mild-reduction-based post-heating treatment. The high combustion efficiency of the tubular flame process enables complete consumption of O₂ in the premixture, along with the emission of CO and H₂. The experimental results revealed that the tubular flames produced mixtures of amorphous tungsten oxide and crystalline tungsten nanoparticles with primary particle sizes of 5–10 nm, at a production rate of 4.68 mg/h and yield of 8.64%. These nanoparticles exhibited a sinter-necked structure. The oxygen content of the tubular-flame-synthesized mixtures of tungsten metal and oxide nanoparticles (16.80%) to 5.56% after the additional post-heating treatment, which also retained the particle size.

With the increasingly strict environmental protection policies, restrictions on NO_x emissions are becoming increasingly stringent. This paper focuses on modeling and optimizing NO_x emission for a coal-fired boiler with advanced deep learning approaches. Three types of deep recurrent neural network models, including recurrent neural network (RNN), long short-term memory (LSTM), and gate recurrent unit (GRU), are developed to model the relationship between operational parameters and NO_x emission of a 660 MW boiler. The hyperparameters of the models are selected by grid search and the effects of the hyperparameters on the prediction results are analyzed. Compared with the traditional back propagation neural network (BP), support vector machine (SVM) models and deep belief network (DBN), the deep recurrent neural network models have higher prediction accuracy. The experimental results show that the GRU-based NO_x prediction model has the best prediction performance among the proposed models. Then, the predicted NO_x emission is used as the objective of searching the optimal parameters for the boiler combustion through the grey wolf optimization (GWO) algorithm. The searching process of GWO is convergent. According to the simulation results, the declines in the NO_x emissions in the two selected cases were 19.49% and 17.96%, which are reasonable achievements for the boiler combustion process.