Search for: All records

This paper presents a second-order, implicit numerical model for one-dimensional, large strain thaw consolidation of ice-rich, fine-grained permafrost. The phase composition of permafrost at sub-freezing temperatures is determined using an unfrozen water content model that accounts for both capillary and adsorptive unfrozen water. The model incorporates secondary compression to improve the accuracy of long-term thaw consolidation simulations. The algorithm incorporates conduction, advection, and phase change in heat transfer and simultaneous occurrence of primary consolidation and secondary compression. Benchmarking and verification of the model show good agreement with existing numerical models. The proposed model is validated against experimental observations. The model indicates that adsorbed unfrozen water dominates over a wide range of sub-freezing temperatures, while capillary unfrozen water freezes at temperatures just below the freezing point. Numerical simulations suggest that ignoring secondary compression can lead to underestimation of excess pore pressure and settlement during both thaw and post-thaw consolidation. Void ratio and average degree of consolidation are overestimated when secondary compression is not considered. The effect of secondary compression on excess pore pressure and void ratio during thawing becomes more pronounced in thicker, field-scale permafrost layers. Results from this study highlight the importance of considering adsorptive and capillary unfrozen water to determine permafrost composition and incorporating secondary compression in thaw consolidation modeling and thaw settlement estimation for long-term civil infrastructure planning in cold regions. The proposed model provides a comprehensive framework for simulating thaw consolidation processes in permafrost regions. 

Large language models (LLMs) exhibit logically inconsistent hallucinations that appear coherent yet violate reasoning principles, with recent research suggesting an inverse relationship between causal reasoning capabilities and such hallucinations. However, existing reasoning approaches in LLMs, such as Chain-of-Thought (CoT) and its graph-based variants, operate at the linguistic token level rather than modeling the underlying causal relationships between variables, lacking the ability to represent conditional independencies or satisfy causal identification assumptions. To bridge this gap, we introduce causal-DAG construction and reasoning (CDCR-SFT), a supervised fine-tuning framework that trains LLMs to explicitly construct variable-level directed acyclic graph (DAG) and then perform reasoning over it. Moreover, we present a dataset comprising 25,368 samples (CausalDR), where each sample includes an input question, explicit causal DAG, graph-based reasoning trace, and validated answer. Experiments on four LLMs across eight tasks show that CDCR-SFT improves the causal reasoning capability with the state-of-the-art 95.33% accuracy on CLADDER (surpassing human performance of 94.8% for the first time) and reduces the hallucination on HaluEval with 10% improvements. It demonstrates that explicit causal structure modeling in LLMs can effectively mitigate logical inconsistencies in LLM outputs. 

The degradation of permafrost alters deformation and long-term strength, posing challenges to existing and future civil infrastructure in Northern Alaska. Long-term strength is a critical parameter in the design of civil projects; yet, to our best knowledge, data on the creep deformation and long-term strength of undisturbed permafrost in Northern Alaska remain limited. Soil particle fraction, unfrozen water content, temperature, and salinity may interactively affect creep deformation and long-term strength of permafrost; however, their interactive effects are not well understood. In this study, field samples of relatively undisturbed permafrost from the upper 1.5 m of the Arctic Coastal Plain near Utqiaġvik, Alaska, were first retrieved and analyzed. The permafrost was characterized as saline ice-rich silty sand and nonuniformly distributed ice. We conducted constant stress creep tests, unconfined compression strength tests, and unfrozen water content tests to assess the mechanical and physical properties of the permafrost cores. The results indicated that the long-term strength of the permafrost decreased by nearly 90% from −10°C to −2°C. At −10°C, the long-term strength increased by approximately 120% as the soil particle fraction rose from 0.14 to 0.26. The strengthening effect of soil particles diminished at higher temperatures and higher salinity due to the influence of unfrozen water. A quantitative tool has been developed to predict the long-term strength of ice-rich permafrost, incorporating the effects of soil particle fraction and temperature. The findings of this study can potentially support infrastructure design and planning in Northern Alaska in the context of a warming climate. 

This paper presents an implicit numerical model for one-dimensional thaw consolidation of saturated permafrost using finite volume approach. The model couples heat transfer with consolidation deformation and accounts for conduction, advection, phase change in heat transfer, and large strain in consolidation. The Crank–Nicolson method is used to obtain transient solutions. The overall approximation of the numerical scheme is of second-order accuracy. Numerical simulations are conducted to analyze the thaw consolidation behaviors in a finite soil layer. Numerical results indicate that, in a finite soil layer, thaw penetration and settlement have nonlinear relationships with the square root of time with decreasing rate. The excess pore water pressure and void ratio at the thaw front decrease with time. Thaw consolidation behaviors can be strongly influenced by the thermal conductivity of soil grains. The advection heat-transfer mechanism has a negligible effect on thaw consolidation in low-permeability soil. 

Advances in robotics represent a potential shift in the construction industry. Construction planning is planned based on craft work; it is necessary to emphasize external factors such as construction robotics. Improving constructability can enhance design-phase construction opportunities, thereby expanding the potential scope of robot operations. However, robotics are often neglected concerning constructability. Previous studies on constructability concentrated on human-based construction methods; hence, gaps remain in assessing constructability for robotics. To minimize the barriers in robotic construction, this paper presents a method for using a rule-based framework for robotic constructability assessment checks with the help of BIM. Focusing on CANVAS—a drywall finishing robot—this paper applies a BIM-based object-oriented model integrating with ROS to utilize constructability reasoning about robotic operations. A model of rule-checking for robotics in the case study is demonstrated and tested. The availability of design information in the model containing robotics is discussed, showing the need for assessing robotics-related constructability information to support an automated review of robotic constructability assessment. This paper applies a case study to validate use of the framework for robotic constructability assessment in the design phase, leading to an automated constructability assessment of construction robotics.