Knudsen force generated by thermally driven gas flow in a microscale structure has been used for gas detection and has shown immeasurable potential in the field of microelectromechanical system (MEMS) gas sensors due to its novel sensing characteristics. In this article, the performances of three kinds of Knudsen force gas sensors with improved isosceles triangular shuttle arm structures were studied. In the first design, the top side and right side lengths were equal; in the second, the top side and bottom side lengths were equal; and for the third, the bottom side and right side lengths were equal. A detailed investigation including gas flow, thermal characteristics, Knudsen force, and coupling effects between the shuttle-heater pairs was conducted using the direct simulation Monte Carlo (DSMC) method and the main mechanisms for gas flow presented were almost the same in this work. However, the second design returned the highest Knudsen force performance. The value increased by 42.9% (P = 387 Pa) compared to the Knudsen force of the original square shuttle arm. The results also demonstrate that the coupling effects become weak toward the right with an increase in the number of shuttle-heater pairs.

We adopted the transient numerical method for the simulation of an ELECTRE vehicle with an opposing jet at an altitude of 53.3 km and 13 Ma to explore the jet characteristics as well as the performance in heat and drag reductions of the opposing jet in hypersonic nonequilibrium flows. The time-accurate, nonequilibrium N-S equations coupled with the five-species Park chemical kinetic model and vibrational energy excitation were applied, and an open source solver Hy2FOAM based on the OpenFOAM platform was adopted. Three opposing jets with different jet radii (R7 jet, R14 jet, and R21 jet) were investigated. The results show that with the increasing jet flow rate, the jet mode of the opposing jet with a small jet radius varies from the overflow mode to the long penetration mode (LPM) and finally to the short penetration mode (SPM), while that with a large jet radius directly changes from the overflow mode to the SPM. The state of the jet in the overflow mode is stable, whereas in SPM and LPM, it is unstable. The investigation of the heat and drag reductions for the R7, R14, and R21 jets shows that except for the jet in LPM, the jet in SPM and overflow mode can provide effective thermal protection, and the thermal protection is enhanced with the increasing jet flow rate. Moreover, the jet in both LPM and SPM can effectively reduce the aerodynamic drag, but the jet in overflow mode cannot provide effective drag reduction. Moreover, the jet with a large radius and in the overflow mode has a better thermal protection effect, and a small jet radius contributes to the drag reduction.

A 2D axi-symmetric theoretical model of dielectric porous media in intermittent microwave (IMW) thermal process was developed, and the electromagnetic energy, multiphase transport, phase change, large deformation, and glass transition were taken into consideration. From the simulation results, the mass was mainly carried by the liquid water, and the heat was mainly carried by liquid water and solid. The diffusion was the dominant mechanism of the mass transport during the whole process, whereas for the heat transport, the convection dominated the heat transport near the surface areas during the heating stage. The von Mises stress reached local maxima at different locations at different stages, and all were lower than the fracture stress. A material treated by a longer intermittent cycle length with the same pulse ratio (PR) tended to trigger the phenomena of overheat and fracture due to the more intense fluctuation of moisture content, temperature, deformation, and von Mises stress. The model can be extended to simulate the intermittent radio frequency (IRF) process on the basis of which one can select a suitable energy source for a specific process.

Microchannel heat exchanger (MCHE) exhibit fast thermal transients and strong regime coupling, making intrusive sensing impractical. We develop an end-to-end framework; discrete wavelet transform (DWT)-Markov transition field (MTF)-2D convolutional neural network (2DCNN)-modified whale optimization algorithm (MWOA)-bidirectional long short-term memory (Bi-LSTM); for outlet-temperature prediction: DWT yields multi-scale sub-bands; each sub-sequence is embedded as a MTF image for 2DCNN based regime classification; for the predicted regime, a Bi-LSTM forecaster is trained on multivariate series with hyperparameters globally tuned by MWOA. Across four scenarios (ramp-up–plateau, exp-rise, sinusoidal, smoothed-random-pulse), the method surpasses LSTM/Bi-LSTM and CNN-Bi-LSTM baselines. Illustratively, performance under exp-rise is moderate (RMSE ≈ 1.69, MAE ≈ 1.325, MAPE ≈ 2.53%, r ≈ 0.984), whereas under smoothed-random-pulse it is highly accurate (RMSE ≈ 0.081, MAE ≈ 0.048, MAPE ≈ 0.21%, r ≈ 0.997). These results endorse a ‘classify-first, optimize per condition’ strategy for stable, condition-aware MCHE forecasting.