Search

In this research, a new idea was implemented to combine different models of artificial intelligence (AI) for evaluating the strength parameters of concrete samples in laboratory. To do this, a series of laboratory data related to concrete samples containing eight different parameters were collected. The system was developed based on various AI models including multi-layer perceptron (MLP), random forest (RF), the k-nearest neighbors (KNN), and tree. The final model developed in this research was designed as a stacking-tree-RF-KNN-MLP structure. This new structure includes various characteristics of four different models that were optimized to achieve a stable structure to evaluate the compressive strength of concrete samples. Each of the basic models involves different parameters that affect the final system performance. By investigating and optimizing each of them, the stacking-tree-RF-KNN-MLP system was updated and finally the final model was obtained. This model covers the weaknesses of each of the basic AI models and uses their best performance in order to get higher prediction capacity. The results of coefficient of determination (R2) showed that the developed model has an accuracy of 0.995 and 0.962 for training and testing data, respectively, which is able to create a suitable structure to predict the compressive strength of concrete samples. In addition, the stacking-tree-RF-KNN-MLP model developed in this research, compared to other models, receives a lower level of system error. The newly developed model can be used in other fields related to construction and material for solving relevant problems.

Low-temperature cracking of asphalt pavements is a major issue in highway construction. The deterioration of fracture resistance in asphalt pavements is accelerated by traffic loads when water infiltrates the cracks. However, previous studies have largely overlooked the effects of moisture content on the low-temperature cracking behaviour of asphalt concrete, particularly under different fracture modes. To address this gap, we investigated the fracture toughness and crack distribution of asphalt concrete using semicircular bending tests under five low-temperature conditions (-10 °C, −5 °C, 0 °C, 5 °C, and 10 °C), five moisture contents (0 %, 25 %, 50 %, 75 %, and 100 %), three fracture modes (mode I, mode II, and mixed mode I&II), and two types of asphalt binders (base asphalt binder and styrene–butadiene–styrene modified asphalt binder).

Ceramsite concrete is currently one of the most widely used types of lightweight concrete. Although its durability (in terms of permeability, shrinkage, etc.) has been extensively demonstrated, fatigue resistance has not been addressed in previous studies, particularly the impact of ceramsite characteristics on fatigue resistance. Hence, in this study, the fatigue life of ceramsite concrete at different stress levels is investigated using four-point bending fatigue tests. The Weibull distribution test reveal, the effects of ceramsite size and content on the equivalent fatigue life under different failure probabilities. The effects of ceramsite on fatigue resistance are also analysed using SEM. Finally, a uniform fatigue equation considering the ceramsite content, ceramsite size, and stress level is established to predict the fatigue life of ceramsite concrete. The results show that: the equivalent fatigue life of ceramsite concrete increases with 5 – 10 mm and 10 – 20 mm ceramsite contents, especially for 5 – 10 mm ceramsite. Concerning 20 – 30 mm ceramsite, it has a negative impact on the fatigue resistance of ceramsite concrete, particularly at low stress levels. Ceramsite size and content have less influence on fatigue life under stress level variation. The propose fatigue prediction equation concurrently considers the effects of ceramsite content, ceramsite size, and stress level on the fatigue life. The R2values of the prediction equation for different failure probabilities exceed 0.94, thereby proving the reliability of the proposed fatigue equation.

Uncertainty is an essentially challenging for safe construction and long-term stability of geotechnical engineering. The inverse analysis is commonly utilized to determine the physico-mechanical parameters. However, conventional inverse analysis cannot deal with uncertainty in geotechnical and geological systems. In this study, a framework was developed to evaluate and quantify uncertainty in inverse analysis based on the reduced-order model (ROM) and probabilistic programming. The ROM was utilized to capture the mechanical and deformation properties of surrounding rock mass in geomechanical problems. Probabilistic programming was employed to evaluate uncertainty during construction in geotechnical engineering. A circular tunnel was then used to illustrate the proposed framework using analytical and numerical solution. The results show that the geomechanical parameters and associated uncertainty can be properly obtained and the proposed framework can capture the mechanical behaviors under uncertainty. Then, a slope case was employed to demonstrate the performance of the developed framework. The results prove that the proposed framework provides a scientific, feasible, and effective tool to characterize the properties and physical mechanism of geomaterials under uncertainty in geotechnical engineering problems.

In order to better understand the mechanical properties of graded crushed rocks (GCRs) and to optimize the relevant design, a numerical test method based on the particle flow modeling technique PFC2Dis developed for the California bearing ratio (CBR) test on GCRs. The effects of different testing conditions and micro-mechanical parameters used in the model on the CBR numerical results have been systematically studied. The reliability of the numerical technique is verified. The numerical results suggest that the influences of the loading rate and Poisson's ratio on the CBR numerical test results are not significant. As such, a loading rate of 1.0–3.0 mm/min, a piston diameter of 5 cm, a specimen height of 15 cm and a specimen diameter of 15 cm are adopted for the CBR numerical test. The numerical results reveal that the CBR values increase with the friction coefficient at the contact and shear modulus of the rocks, while the influence of Poisson's ratio on the CBR values is insignificant. The close agreement between the CBR numerical results and experimental results suggests that the numerical simulation of the CBR values is promising to help assess the mechanical properties of GCRs and to optimize the grading design. Besides, the numerical study can provide useful insights on the mesoscopic mechanism.