Prediction of pile settlement using artificial neural networks based on standard penetration test data

In recent years artificial neural networks (ANNs) have been applied to many geotechnical engineering problems with some degree of success. With respect to the design of pile foundations, accurate prediction of pile settlement is necessary to ensure appropriate structural and serviceability performance. In this paper, an ANN model is developed for predicting pile settlement based on standard penetration test (SPT) data. Approximately 1000 data sets, obtained from the published literature, are used to develop the ANN model. In addition, the paper discusses the choice of input and internal network parameters which were examined to obtain the optimum model. Finally, the paper compares the predictions obtained by the ANN with those given by a number of traditional methods. It is demonstrated that the ANN model outperforms the traditional methods and provides accurate pile settlement predictions.     

Explaining the hydroclimatic variability and change in the Salmon River basin

Climate change in the Pacific Northwest and in particular, the Salmon River Basin (SRB), is expected to bring about 3–5 °C rise in temperatures and an 8 % increase in precipitation. In order to assess the impacts due to these changes at the basin scale, this study employed an improved version of Variable Infiltration Capacity (VIC) model, which includes a parallel version of VIC combined with a comprehensive parameter estimation technique, Shuffled Complex Evolution (SCE) to estimate the streamflow and other water balance components. Our calibration (1955–1975) and validation (1976–1999) of the model at the outlet of the basin, White Bird, resulted in an r² value of 0.94 which was considered satisfactory. Subsequent center of timing analysis showed that a gradual advancement of snowmelt induced-peak flow advancing by about 10 days in the future. Historically, the flows have shown a general decline in the basin, and in the future while the magnitudes might not be greatly affected, decreasing runoff of about 3 % over the next 90 years could be expected and timing of peak flow would shift by approximately 10 days. Also, a significant reduction of snow water equivalent up to 25 %, increased evapotranspiration up to 14 %, and decreased soil moisture storages of about 2 % is predicted by the model. A steady decline in SWE/P from the majority of climate model projections for the basin was also evident. Thus, the earlier snowmelt, decreasing soil moisture and increased evapotranspiration collectively implied the potential to trigger drought in the basin and could affect the quality of aquatic habitats and their spawning and a detailed investigation on these impacts is warranted.     

ANN-based model for predicting the bearing capacity of strip footing on multi-layered cohesive soil

In reality, footings are most likely to be founded on multi-layered soils. The existing methods for predicting the bearing capacity of 4-layer up to 10-layer cohesive soil are inaccurate. This paper aims to develop a more accurate bearing capacity prediction method based on multiple regression methods and multi-layer perceptrons (MLPs), one type of artificial neural networks (ANNs). Predictions of bearing capacity from the developed multiple regression models and MLP in tractable equations form are obtained and compared with the value predicted using traditional methods. The results indicate ANNs are able to predict accurately the bearing capacity of strip footing and outperform the existing methods.     

Neural network prediction of the reliability of heterogeneous cohesive slopes

The reliability of heterogeneous slopes can be evaluated using a wide range of available probabilistic methods. One of these methods is the random finite element method (RFEM), which combines random field theory with the non‐linear elasto‐plastic finite element slope stability analysis method. The RFEM computes the probability of failure of a slope using the Monte Carlo simulation process. The major drawback of this approach is the intensive computational time required, mainly due to the finite element analysis and the Monte Carlo simulation process. Therefore, a simplified model or solution, which can bypass the computationally intensive and time‐consuming numerical analyses, is desirable. The present study investigates the feasibility of using artificial neural networks (ANNs) to develop such a simplified model. ANNs are well known for their strong capability in mapping the input and output relationship of complex non‐linear systems. The RFEM is used to generate possible solutions and to establish a large database that is used to develop and verify the ANN model. In this paper, multi‐layer perceptrons, which are trained with the back‐propagation algorithm, are used. The results of various performance measures indicate that the developed ANN model has a high degree of accuracy in predicting the reliability of heterogeneous slopes. The developed ANN model is then transformed into relatively simple formulae for direct application in practice. Copyright © 2016 John Wiley & Sons, Ltd.     

A method for generating virtual soil profiles with complex,multi-layer stratigraphy

This paper presents a framework for generating multi-layer, unconditional soil profiles with complex stratigraphy, which simulates the effects of natural erosion and sedimentation processes. The stratigraphy can have varying degrees of randomness and can include features such as lenses, as well as sloped and undulating layers. The method generates the soil comprising the layers using local average subdivision (LAS), and a random noise component that is added to the layer boundaries. The layers are created by generating coordinates of key points in the simulated ground profile, which are then interpolated with a customised, 2D, linear interpolation algorithm. The resulting simulations facilitate more accurate probabilistic modelling of geotechnical engineering systems because they provide more realistic geologies, such as those usually encountered in the ground. Fortran code implementing this framework is included as supplementary material.