Sorption-enhanced steam gasification (SESG) represents a promising route for hydrogen-rich syngas production. However, the rapid deactivation of conventional CaO-based sorbents, and the efficiency loss associated with the high temperature sorbent regeneration step, remain as critical challenges. In this study, a redox-activated SrMnO₃ sorbent was used for isothermal SESG of biomass in a fluidized-bed to investigate fuel flexibility, torrefaction effects, and long-term stability. Four biomass feedstocks as well as their torrefied counterparts were evaluated. All untreated feedstocks produced hydrogen-rich syngas with H₂ concentration >60% and H₂/CO ratios >4 under isothermal operation at 850 ℃. Torrefaction decreased H₂ purity and syngas yield due to reduced volatile matter content. Long-term experiments in both a fluidized-bed reactor and TGA demonstrated excellent cyclic stability of SrMnO₃ and strong ash resistance. The structural and compositional properties of the sorbents were characterized in detail, confirming the chemical and structural stability of the SrMnO₃ sorbent during prolonged cyclic operation. Process simulation further showed that SESG of biomass significantly enhanced cold gas efficiency while maintaining a comparable heat demand compared to state-of-the-art indirect biomass gasification. Overall, SrMnO₃ exhibits strong potential as a robust sorbent for efficient and flexible biomass-to-hydrogen conversion.

Hydrogel desorption shows promise for thermal management and passive cooling engineering applications, but challenges persist in understanding particle desorption and shrinkage. A Lattice Boltzmann-based phase-field model was developed by combining experiments and mesoscopic simulations to simulate conjugate heat transfer, desorption kinetics, and shrinkage of single hydrogel particles, achieving <5% deviation from experiments. Results show that raising temperature from 343 to 353 K increases 1-h desorption mass by ~12% of initial mass (vs. 1.3% for 313-323 K), driven by rising saturation vapor pressure and falling evaporation enthalpy. Shrinkage is more sensitive at high temperature and low humidity, with surface area reductions of ~17% (343-353 K) and~13% (20-30% RH), compared to ~5% under mild conditions. Reducing particle initial mass from 5.75 g to 0.16 g shortens the time to release 50% of the initial water from 516 min to 117 min, indicating greater efficiency for smaller particles. This work bridges the macro-pore scale gap and offers a numerical tool with design insights for passive cooling.

To improve understanding of proppant transport mechanisms in complex fracture networks, a field-scale model based on the Multiphase Particle-in-Cell (MP-PIC) method was developed using fracture geometries derived from real shale outcrops and validated against experimental data. Results indicate that: (1) intense vortex formation during early-to-mid injection stages or at high flow rates exacerbates the longitudinal heterogeneity of proppant distribution. (2) In the near-wellbore zone, the synergistic effects of fracture width variations, high flow velocities, and natural weak planes drive proppants to migrate preferentially along paths aligned with the maximum and minimum principal stress directions, forming a dual-channel transport pattern. (3) The volume of proppant entering secondary fractures decreases with distance from the injection point, and the proppant dune height within dominant channels exhibits stepwise attenuation. Larger intersection angles between secondary and main fractures hinder proppant migration. (4) Smaller proppant size and lower density improve the planar sweep and distribution uniformity coefficients, while increased fracturing fluid viscosity extends the proppant sweep range and further improves uniformity; high injection rates promote long-distance proppant transport and broader coverage but may reduce uniformity, leading to sparse proppant distribution and necking at fracture mouths. These findings provide quantitative guidance for optimizing hydraulic fracturing designs.

Hard-sphere/pseudo-particle modeling (HS-PPM) provides a parallel and more versatile implementation of molecular dynamics simulation based on hard-sphere model at large scales with remarkable performance. For more practical applications, however, more sophisticated and realistic boundary conditions have to be incorporated. In this work, the artifact of inlet-outlet coupling due to periodic boundary is basically eliminated by adding collisions at both ends and proportional-integral-differential (PID) control of density at the outlet based on implicit boundary conditions, maintaining reasonable flow field there for both high (up to 13.0) and low (down to 0.0) Mach numbers. Moreover, geometrically complex boundaries are approximated by a large number of polygonal (in particular, triangular) elements, instead of exact analytical descriptions, to balance the accuracy and efficiency. Given the challenges of nanoscale experimentation, the drag coefficients from simulation conducted at matching dimensionless numbers were compared with macroscopic results. With these improvements, the flow around a conical particle and the Inflatable Re-entry Vehicle Experiment (IRVE) was successfully simulated and demonstrated consistent to experiment and numerical results from direct simulation Monte Carlo (DSMC) towards the continuum flow limit (with the Knudsen number ranging from 0.1 to 1.7). Furthermore, the drag coefficient of nano-particles in fluid flow, which is challenging for experimental measurement, was obtained from simulation results, and the potential of HS-PPM in simulating particles and systems with complex structures is exemplified by the gas flow and diffusion in porous micro/nano-particles.

To enhance the calibration efficiency and accuracy of Discrete Element Method (DEM) parameters for cohesive bulk materials, a collaborative method integrating Particle Swarm Optimization (PSO) and Backpropagation (BP) neural networks is proposed. Key macroscopic indicators (steady-state shear stress, angle of repose) are obtained via Jenike shear and funnel tests across a 0-50% moisture range. Orthogonal experiments determine micro-parameters (e.g., static/rolling friction, surface energy) to build a macro-micro mapping database. The core of the PSO-BP dual-model lies in its collaborative mechanism: the forward BP model predicts macroscopic responses to replace time-consuming DEM simulations, while the PSO algorithm optimizes the inverse BP model to accurately infer optimal micro-parameters from experimental macro-indicators (steady-state shear stress, angle of repose). Validation shows low errors (1.14% for angle of repose, 1.63% for steady-state shear stress) and good chute flow velocity agreement. This method overcomes traditional limitations of arbitrariness and ignored parameter coupling, providing reliable support for DEM simulation and equipment design for cohesive bulk materials.

Sinkholes, a common geotechnical failure, have been reported more frequently in recent years due to complex climate change and urbanisation. This study investigates mechanisms of sinkhole development, considering the influence of triggering depth and compaction degree through experiments integrated with Particle Image Velocimetry (PIV) technique and Discrete Element Method (DEM). Laboratory tests simulating the formation of sinkhole in subgrade soil are carried out under varying triggering depths and void ratios, where the soil displacement is recorded over time. DEM simulations are performed and validated against the experimental results analysed by PIV technique. The results show good agreement between DEM and experimental investigations, especially in capturing time-dependent stages and displacement fields of sinkhole development. Furthermore, the numerical investigation indicates that increasing the triggering depth results in later development of sinkhole with greater vertical displacement around the triggering point to promote its propagation towards the ground surface. Interestingly, dense soil exhibits highly localised mobilisation confined within well-defined displacement boundaries, whereas loose soil displays a wider displacement zone characterised by a funnel-shaped pattern. The interparticle contact behaviour further releases the insightful mechanism of sinkhole progression, giving considerable value to our understanding and prediction of this catastrophic failure.

This study performs a comparative assessment of three Machine Learning (ML) models to determine their robustness in data-sparse industrial environments for predicting and optimising the performance and energy consumption of an immersion mill. By employing active learning strategies that target high-variance regions in process space, a limited experimental dataset (15 batches) was found sufficient to accurately predict mill performance, provided the correct model architecture is chosen. Three models were used: Random Forest Regression (RFR), Gradient Boosted Trees (GBT) and Symbolic Regression (SR). Based on performance when used on completely unseen data, as well as interpretability and usability, the SR models were found to be the most effective for this application. The simple algebraic form of the SR models allowed for direct use in exploring 'what-if' scenarios, and equally allowed either model to serve as a constraint for the other to minimise energy use to achieve a target particle size. An unexpected finding during this work was that the relationship between final particle size and impeller speed and grinding media content is weaker than for classic vertical stirred mill designs owing to transport and/or mixing mechanisms novel to this particular mill design. The result of this study is a set of predictive models that can be used in optimising the immersion milling process, and effectively responding to changes in feed Particle Size Distribution (PSD) whilst minimising energy use.

Char particles from coal pyrolysis exhibit unique settling dynamics, such as terminal velocity and drag coefficient. Accurate quantification of drag characteristics is essential for pneumatic conveying. Current simulations adopting spherical drag models suffer large deviations for irregular char particles. To fill this gap, we built a visual platform to investigate free settling of coal and char particles (<200 μm). Terminal velocities were measured for individual particles and aggregates. Based on statistical analysis, we revised Stokes and Schiller drag formulations. In the Stokes form, the fitted coefficients for individual coal particles and coal aggregates are 26.04 ± 3.5 and 52.15 ± 4.8, respectively; for individual char particles and char aggregates, they are 41.05 ± 3.2 and 112.4 ± 4.3, respectively. In the Schiller form, the fitted coefficients for individual coal particles and coal aggregates are 26.19 ± 3.4 and 53.88 ± 4.2, respectively; for individual char particles and char aggregates, they are 40.94 ± 3.2 and 111.1 ± 4.2, respectively. These correlations can be directly employed in drag model for numerical simulation of char pneumatic conveying. Validation shows significantly improved accuracy. This work presents the first experimentally derived drag coefficient correlations for char particles, addressing the lack of quantitative data and offering a more precise tool for char transport design.

This paper develops a computational framework coupling a moisture-induced cohesive force model with the Coarse-Grained Discrete Element Method (CG-DEM) to efficiently and accurately simulate the mixed discharge process of wet coal in a laboratory-scale three-dimensional silo. The moisture-induced cohesive force model quantifies the effect of moisture on coal flowability by correlating moisture content with macroscopic mechanical parameters of the particles (cohesion, internal friction angle). Simulation results indicate that particle flow exhibits a significant central-channeling "funnel flow" characteristic, and the flow structure is dominated by the flowability of the lower coal layer. The moisture-induced cohesive force is a key factor controlling particle motion. The study further reveals the decisive influence of stacking sequence on the evolution of outlet coal quality: when the lower layer consists of high-moisture, poor-flowability Indonesian coal, the outlet calorific value shows a slow rising trend; conversely, when the lower layer consists of good-flowability Longwanggou coal, the outlet calorific value decreases rapidly. This research provides a reliable, low-cost numerical tool for understanding and predicting the discharge and mixing behavior of wet coal in industrial silos, offering guidance for optimizing coal blending and stabilizing the quality of feed coal.

Chemical recycling of polymethyl methacrylate (PMMA) to its monomer, methyl methacrylate (MMA), requires balancing primary depolymerization with the suppression of secondary gas-phase reactions. This study investigates non-oxidative MMA decomposition in a fluidized bed reactor across a temperature range of 623 to 1073 K using online FTIR spectroscopy. Experimental results reveal a significant shift in product selectivity: low temperatures favor a low-energy decarboxylation pathway (yielding CO₂ and methanol), while high temperatures promote radical cracking (yielding CO and light hydrocarbons). To describe this, a two-competing-reactions model (CRM) is used, outperforming the traditional single first-order approaches. The CRM identifies two distinct activation energies: E_a,1 = 76.5 kJ mol⁻¹ for decarboxylation and E_a,2 = 269.9 kJ mol⁻¹ for cracking. The research further demonstrates that the classical sequential decomposition model (PMMA → MMA→ light gases) overpredicts monomer yields at low temperatures. By integrating a direct solid-to-gas pathway to account for side-chain break-off and incorporating multi-volume reactor hydrodynamics, the model's predictive accuracy significantly improved. This integrated framework identifies an optimal recovery window near 723 K, achieving MMA yields over 95 %.