We investigate the numerical approximation of parabolic interface optimal control problems using the stable generalized finite element method (SGFEM). For spatial discretization of the state variable, we employ the SGFEM, while the backward Euler method is utilized for temporal discretization. The control variable is discretized using the variational discretization method. We derive error estimates for the optimal control, state, and adjoint state. Finally, several numerical examples are provided to verify the analysis results.
Defect opening and fracture in high-pressure CO2 pipelines under near-critical and dense-phase conditions involve rapid depressurization, phase-change-related cooling and complex pipe-wall deformation. In this study, three fracture tests were conducted on DN150 X65 pipe sections subjected to pure CO2 release. The initial pressure ranged from 8.2 to 8.7 MPa, and the initial temperature ranged from 18 to 50 degrees C. External axial defects with a depth of approximately 90% of the wall thickness were pre-machined. Pressure, temperature, fracture timing-wire signals and high-speed images were recorded, and the fracture behavior was analyzed using macroscopic morphology, SEM observations, image-based apparent CTOA measurement and decompressionwave prediction. The three tests showed distinct responses: limited axial crack propagation followed by circumferential tearing, defect opening only, and local large deformation without stable crack propagation. Rapid pressure drop, pressure rebound and decompression plateaus were observed during CO2 release, which may be associated with transient phase-change behavior, pressure-wave reflection, rupture-disc opening and release-area evolution. In Tests 1 and 3, the minimum measured temperature in the upstream pipe reached 49.56 degrees C, with the maximum cooling occurring at the pipe bottom, indicating strong vertical thermal stratification. High-speed imaging in Test 1 showed that the image-based apparent CTOA remained within a relatively stable range during the visible crack-opening stage. SEM observations in Test 1 confirmed ductile microvoid coalescence without typical brittle fracture features. The decompression model captured the plateau trend but overestimated the initial wave speed and underestimated the plateau duration. These results provide experimental support for fracture response and depressurization assessment of small- and medium-diameter CO2 pipe sections within the present pressure and temperature range.
Investigating the relationship between the distribution of light and heavy fractions in thermal dissolution soluble portions (SPs) and the composition and structural characteristics of raw coal is crucial for elucidating the depolymerization mechanisms of soluble organic matter in coal. In this study, fourteen medium and low-rank coals from China were thermally dissolved at 320 degrees C using an isometric toluene/methanol mixture as the solvent. The correlations between the relative contents of three fractions (light oil, asphaltenes, and preasphaltenes) in the SPs and the composition and structural characteristics of the raw coals were systematically examined. The results show that the SP yield is positively correlated with the volatile matter content of raw coal. Higher H/C and O/C atomic ratios in the coal samples are associated with increased yields of both SPs and light oil; however, as these ratios increase, the proportion of light oil in the SPs gradually decreases. Fourier transform infrared spectroscopy (FT-IR) analysis further reveals that the proportion of light oil in the SPs increases with the aromaticity of raw coal, but decreases with longer aliphatic side chains, suggesting that extended aliphatic side chains hinder the formation of light oil during thermal dissolution. Additionally, as the oxygen enrichment coefficient increases, the proportion of light oil first decreases and then rises. This study provides important theoretical insight into the depolymerization mechanisms of organic species in coal.
Deep learning has been widely applied to vascular segmentation tasks because of its powerful capabilities in multi-scale feature extraction and representation learning. However, renal arteries exhibit large variations in vessel diameter, tree-like branching structures, and complex spatial topology, which pose significant challenges for existing models in preserving vessel integrity and continuity. To address these challenges, we propose a detail-preserving network based on an encoder-decoder architecture. Specifically, a multi-scale feature fusion module captures vessels with varying diameters by integrating multi-scale contextual information, while a detail-preserving downsampling module is designed to alleviate the loss of fine vessel information during downsampling. In addition, a residual convolutional block is incorporated to enhance feature representation and promote the continuity of vascular structures. Experiments on a limited retrospective clinical renal artery dataset showed that the proposed network achieved favorable performance compared with several representative segmentation methods in terms of the harmonic mean of precision and sensitivity, intersection over union, and the centerline Dice coefficient. These results suggest the potential value of the proposed network for preserving fine vessel continuity and maintaining the topological integrity of the vascular tree.
A two-stage centrifugal turning flue with ash hoppers is proposed to address the severe sticky ash deposition on the high-temperature economizer in a circulating fluidized bed (CFB) hot-water boiler firing Zhundong high-alkali coal, which has drastically shortened the soot-blowing cycle. The design exploits the centrifugal inertial effect of flue gas turning to remove coarse fly ash particles, thereby reducing the ash load in flue gas at the economizer inlet. Based on measured ash composition data from on-site coal samples and the actual boiler geometry, a sequentially coupled "flue gas–solid separation–tube bundle deposition calculation" approach is employed to compare the ash deposition characteristics between the baseline and retrofitted configurations. The results show that the two-stage centrifugal turning flue achieves a fly ash mass removal efficiency of 48.1%, with nearly all coarse particles captured, while the system pressure drop increases by only 92 Pa. After the retrofit, the total ash deposition on the economizer decreases by 34.8%, and the transverse deposition non-uniformity coefficient is reduced by 29.1%. The proposed scheme requires only local modification of the existing flue with low engineering cost and can serve as a reference for ash deposition control in similar boilers.