We investigate the morphology and dynamics of the pattern of immiscible invasion by injecting a high-viscosity liquid into a granular suspension consisting of movable solid grains in a low-viscosity liquid. Laboratory experiments conducted in a Hele–Shaw cell shed light on how the frictional forces of the grains and the viscous forces of the liquids affect the instability of the liquid–liquid interface and the formation of viscous fingers. The frictional force increases with an increase in either the volume fraction or the size of the grains, leading to higher resistance to the invading pattern. Upon changing the grain shape from spherical to irregular, both the frictional force and the rotational energy of the grains increase, resulting in more numerous but narrower fingers. Increasing either the injection rate or the viscosity of the injected liquid increases the viscous pressure within the fingers, promoting the splitting of the pattern. Although the defending liquid always has a lower viscosity than the invading liquid in this study, the former's viscous force becomes non-negligible as the viscosity ratio of the invading liquid to the defending liquid decreases to near unity, which destabilizes the fluid–fluid interface and causes a transition to an asymmetric pattern.

Pipeline blockage is a critical reliability constraint in cemented paste backfill (CPB) transport. Existing machine learning studies mainly predict rheology, pressure drop, deposition velocity, or blockage diagnosis from relatively consistent input sets, but CPB clogging risk mapping often has to use sparse records, heterogeneous labels, and missing operating variables. This study proposes a knowledge guided transfer learning framework for screening transport failure related to clogging. The framework converts available slump and yield stress measurements into a unified flowability index (UFI), combines UFI with solids mass concentration to build an interpretable chart prior, and represents unreported inlet velocity and pipe diameter through latent process priors and Monte Carlo marginalization. Bootstrap samples with jitter provide a pseudo source dataset for pretraining, followed by fine tuning and evaluation on a compiled dataset of 52 records with six positives under the main label definition. The evidence is evaluated through repeated splits containing both classes, LOPO case level analysis, label sensitivity definitions, prior learning ablations, paired baseline comparisons, UFI uncertainty propagation, and process prior sensitivity tests. The resulting risk map integrates sparse field records with explicit prior information and shows plausible increases in screening score at higher solids concentration and poorer flowability, which provides a transparent tool for design screening, diagnostic scenario comparison, and selection of cases requiring closer hydraulic assessment.

Carbonation in CO₂-storage solid-waste backfill materials couples CO₂ transport, mineral reaction, and strength evolution, making carbonation-front prediction and transport–reaction inference important for evaluating sequestration performance. This study proposes an evidence-ranked, physics-guided inverse-learning framework for carbonation-front prediction, auxiliary strength reconstruction, PDE-residual assessment, and practical-identifiability analysis. The framework represents carbonation using group-conditioned latent fields of effective CO₂ concentration and remaining reactive capacity, maps latent carbonation degree to measured depth through a differentiable front operator, and reconstructs unconfined compressive strength through a supervised auxiliary head. Empirical front laws and reaction–diffusion physics-informed neural-network variants were evaluated using held-out ranking, repeated stratified splits, residual-weight sweeps, front-operator threshold and smoothing-coefficient sensitivity checks, profile-likelihood and Fisher-information diagnostics, and controlled synthetic tests. Results show that the grouped Weibull front law achieved the best short-range carbonation-depth interpolation, while the retained constant-diffusion PINN was used as a diagnostic formulation within the physics-guided family to improve auxiliary strength reconstruction and to evaluate residual consistency, front-threshold selection, parameter sharing, and inverse-parameter behavior rather than to replace the empirical depth regressor. Increasing the PDE-residual weight substantially reduced residual magnitudes, but profile-likelihood and Fisher-information diagnostics indicated strong parameter trade-offs; the fitted diffusion, reaction, depletion, and diffusion–decay quantities are therefore interpreted as effective, observation-conditional parameters rather than unique material constants. The proposed framework provides a prediction-first and attribution-aware approach for analyzing carbonation evolution in solid-waste backfill materials and supports coordinated assessment of front advancement, strength response, and transport–reaction behavior, while explicitly delimiting the generalization and physical interpretation that can be supported by sparse literature-derived observations.