Evaluation of the Change in Undrained Shear Strength in Cohesive Soils due to Principal Stress Rotation Using an Artificial Neural Network

Prediction of Permeability Coefficient k in Sandy Soils Using ANN

The paper presents a method of application of an ANN (Artificial Neural Network) to predict the permeability coefficient k in sandy soils: FSa, MSa, CSa. To develop an ANN the results of permeability coefficients from pumping and consolidation tests were applied. The proposed ANN with an architecture 6-8-1 predicts the value of permeability coefficient k based on the following parameters: soil type, relative density ID, void ratio e and effective soil diameter d10. The mean relative error and single maximum value of the relative error for the proposed ANN are following: Mean RE = ±4%, Max RE = 7.59%. The use of the ANN to predict the soil permeability coefficient allows the reduction of the costs and time needed to conduct laboratory or field tests to determine this parameter.

Prediction of Post-Yield Strain from Loading and Unloading Phases of Pressuremeter, Triaxial, and Consolidation Test Curves for Sustainable Embankment Design

Exponential development of post-yield strain (Ԑpost) is a pivotal indicator of failure in embankments constructed on soft saturated clays. This paper characterizes saturated clay stratum comprising very soft to very stiff stratigraphy, with plasticity index (PI) ranging from 19% to 31%, by performing widely used geotechnical engineering tests, i.e., the prebored pressuremeter (PMT) test, the triaxial (TXL) test, and constant-rate-of-strain (CRS) consolidation. PMT, TXL, and CRS tests were performed at a strain rate range of 0.18%/min to 0.21%/min to explore the yield stress (σ′y), the pre-yield strain (Ԑpre), and the post-yield strain (Ԑpost). Results indicate that Ԑpost/Ԑpre for PMT, TXL, and CRS stress–strain curves range from 2.7 to 19 in the loading phase and 2 to 21 in the unloading phase. An exponential increase in Ԑpost/Ԑpre is observed in the range of 10 to 21 for very soft to soft clay which is congruent with the realistic sustainable range of 4 to 30 for embankment failure on soft clays worldwide. The evaluated Ԑpost/Ԑpre can be applied for sustainable prediction of post-failure evolution of strains in embankments on soft clays. Simplistic correlations are developed for approximation and prediction of Ԑpost as a function of σ′y, Ԑpre and maximum applied pressure (Pmax) for loading and unloading phases with reasonable accuracy. The intuitive zone of critical ℇpost is quantified for impending failure in embankments for maximum applied pressure (Pmax), ranging from 36 kPa to 100 kPa for very soft to soft clay for use in sustainable embankment design and construction. Variation in predicted versus measured results of an individual site is observed to be within ±10% of line of equality.

Shuffled Frog Leaping Algorithm and Wind-Driven Optimization Technique Modified with Multilayer Perceptron

The prediction aptitude of an artificial neural network (ANN) is improved by incorporating two novel metaheuristic techniques, namely, the shuffled frog leaping algorithm (SFLA) and wind-driven optimization (WDO), for the purpose of soil shear strength (simply called shear strength) simulation. Soil information of the Trung Luong national expressway project (Vietnam) including depth of the sample (m), percentage of sand, percentage of silt, percentage of clay, percentage of moisture content, wet density (kg/m3), liquid limit (%), plastic limit (%), plastic index (%), liquidity index, and the shear strength (kPa) was collocated through a field survey. After constructing the hybrid ensembles of SFLA–ANN and WDO–ANN, both models were optimized in terms of complexity using a population-based trial-and error-scheme. The learning quality of the ANN was compared with both improved versions to examine the effect of the used metaheuristic techniques. In this phase, the training error dropped by 14.25% and 28.25% by applying the SFLA and WDO, respectively. This reflects a significant improvement in pattern recognition ability of the ANN. The results of the testing data revealed 25.57% and 39.25% decreases in generalization (i.e., testing) error. Moreover, the correlation between the measured and predicted shear strengths (i.e., the coefficient of determination) rose from 0.82 to 0.89 and 0.92, which indicates the efficiency of both SFLA and WDO metaheuristic techniques in optimizing the ANN.