Yusuf Erzin

The prediction of the critical factor of safety of homogeneous finite slopes using neural networks and multiple regressions

This study deals with development of artificial neural network (ANN) and multiple regression (MR) models that can be employed for estimating the critical factor of safety (Fs) value of homogeneous finite slopes. To achieve this, the Fs values of 675 homogenous finite slopes having different soil and slope parameters were calculated by using the simplified Bishop method and the minimum (critical) Fs value for each slope was determined and used in the ANN and MR models. The results obtained from ANN and MR models were compared with those obtained from the calculations. The values predicted from ANN models matched the calculated values much better than those obtained from MR models. Additionally, several performance indices such as determination coefficient (R^2), variance account for (VAF), mean absolute error (MAE), and root mean square error (RMSE) were calculated; the receiver operating curves (ROC) were drawn, and the areas under the curves (AUC) were calculated to assess the prediction capacity of the ANN and MR models. ANN models have shown higher prediction performance than MR models based on the performance indices and the AUC values. The results demonstrated that the ANN models can be used at the preliminary stage of designing homogeneous finite slope.

Artificial neural network (ANN) models for determining hydraulic conductivity of compacted fine-grained soils

This study deals with development of artificial neural networks (ANNs) and multiple regression analysis (MRA) models for determining hydraulic conductivity of fine-grained soils. To achieve this, conventional falling-head tests, oedometer falling-head tests, and centrifuge tests were conducted on silty sand and marine clays compacted at different dry densities and moisture contents. Further, results obtained from ANN and MRA models were compared vis-a-vis experimental results. The performance indices such as the coefficient of determination, root mean square error, mean absolute error, and variance were used to assess the performance of these models. The ANN models exhibit higher prediction performance than the MRA models based on their performance indices. It has been demonstrated that the ANN models developed in the study can be employed for determining hydraulic conductivity of compacted fine-grained soils quite efficiently.

Artificial neural networks approach for swell pressure versus soil suction behaviour

In this study, the swell pressure versus soil suction behaviour was investigated using artificial neural networks (ANNs). To achieve this, the results of the total suction measurements using thermocouple psychrometer technique and constant-volume swell tests in oedometers performed on statically compacted specimens of Bentonite–Kaolinite clay mixtures with varying soil properties were used. Two different ANN models have been developed to predict the total suction and swell pressure. The ANNs results were compared with the experimental values and found close to the experimental results. Moreover, several performance indices such as correlation coefficient, variance account for (VAF), and root mean square error (RMSE) were calculated to check the prediction capacity of the ANN models developed. Both ANN models have shown a high prediction performance based on the performance indices. Therefore, it can be concluded that the initial soil suction is the most relevant state of suction that characterizes the pot...

The use of neural networks for the prediction of the settlement of one-way footings on cohesionless soils based on standard penetration test

In this study, artificial neural networks (ANNs) were used to predict the settlement of one-way footings, without a need to perform any manual work such as using tables or charts. To achieve this, a computer programme was developed in the Matlab programming environment for calculating the settlement of one-way footings from five traditional settlement prediction methods. The footing geometry (length and width), the footing embedment depth, the bulk unit weight of the cohesionless soil, the footing applied pressure, and corrected standard penetration test varied during the settlement analyses, and the settlement value of each one-way footing was calculated for each traditional method by using the written programme. Then, an ANN model was developed for each method to predict the settlement by using the results of the analyses. The settlement values predicted from each ANN model developed were compared with the settlement values calculated from the traditional method. The predicted values were found to be quite close to the calculated values. Additionally, several performance indices such as determination coefficient, variance account for, mean absolute error, root mean square error, and scaled percent error were computed to check the prediction capacity of the ANN models developed. The constructed ANN models have shown high prediction performance based on the performance indices calculated. The results demonstrated that the ANN models developed can be used at the preliminary stage of designing one-way footing on cohesionless soils without a need to perform any manual work such as using tables or charts.

The use of neural networks for CPT-based liquefaction screening

This study deals with development of two different artificial neural network (ANN) models: one for predicting cone penetration resistance and the other for predicting liquefaction resistance. For this purpose, cone penetration numerical simulations and cyclic triaxial tests conducted on Ottawa sand–silt mixes at different fines content were used. Results obtained from ANN models were compared with simulation and experimental results and found close to them. In addition, the performance indices such as coefficient of determination, root mean square error, mean absolute error, and variance were used to check the prediction capacity of the ANN models developed. Both ANN models have shown a high prediction performance based on the performance indices. It has been demonstrated that the ANN models developed in this study can be employed for predicting cone penetration and liquefaction resistances of sand–silt mixes quite efficiently.

Factors influencing the crushing strength of some Aegean sands

Engineering properties of sands mainly depend on the integrity of the particles, which in turn has a strong bearing on their crushing strength. Seven different Aegean sands were tested for mineralogy, particle shape, size and specific gravity and the influence of aspect ratio, particle composition, particle shape and size on the crushing strength was examined. As the Aegean sands have a small range of sphericity and roundness, crushing strength tests were also performed on five Anatolian sands. A multiple regression analysis was carried out and an equation proposed to determine the crushing strength value of the Aegean sands. The computed values were found to be in good agreement with those obtained from the experimental investigations. It is concluded that the equation is sufficiently accurate to be a useful, time- and cost-effective way of obtaining crushing strength estimations at the preliminary stage of site investigations.RésuméLes propriétés géotechniques de sables dépendent principalement de l’intégrité des particules qui est fortement en rapport avec leur résistance à l’écrasement. Sept sables égéens différents ont été analysés quant à la nature minéralogique des grains, la forme et la taille de particules et leur densité. Les influences de la composition des particules, de leur forme et de leur taille sur la résistance à l’écrasement ont été étudiées. Comme les sables égéens présentent des sphéricités et des émoussés peu variables, des essais d’écrasement ont aussi été réalisés sur cinq sables anatoliens. Une analyse par régression multiple a été réalisée et une équation proposée pour déterminer la résistance à l’écrasement des sables égéens. Les valeurs calculées sont apparues en bon accord avec les valeurs mesurées. On conclut que l’équation est suffisamment précise pour fournir des estimations utiles et peu coûteuses de la résistance à l’écrasement dans le contexte d’études préliminaires de site.

Use of neural networks for the prediction of the CBR value of some Aegean sands

This study deals with the development of an artificial neural network (ANN) and a multiple regression (MR) model that can be employed for estimating the California bearing ratio (CBR) value of some Aegean sands. To achieve this, the results of CBR tests performed on the compacted specimens of nine different Aegean sands with varying soil properties were used in the development of the ANN and MR models. The results of the ANN and MR models were compared with those obtained from the experiments. It is found that the CBR values predicted from the ANN model matched the experimental values much better than the MR model. Moreover, several performance indices, such as coefficient of determination, root-mean-square error, mean absolute error, and variance, were used to evaluate the prediction performance of the ANN and MR models. The ANN model has shown higher prediction performance than the MR model based on the performance indices, which demonstrates the usefulness and efficiency of the ANN model. Thus, the ANN model can be used to predict CBR value of the Aegean sands included in this study as an inexpensive substitute for the laboratory testing, quite easily and efficiently.

The use of neural networks for the prediction of cone penetration resistance of silty sands

In this study, an artificial neural network (ANN) model was developed to predict the cone penetration resistance of silty sands. To achieve this, the data sets reported by Ecemis and Karaman, including the results of three high-quality field tests, namely piezocone penetration test, pore pressure dissipation tests, and direct push permeability tests performed at 20 different locations on the northern coast of the Izmir Gulf in Turkey, have been used in the development of the ANN model. The ANN model consisted of three input parameters (relative density, fines content, and horizontal coefficient of consolidation) and a single output parameter (normalized cone penetration resistance). The results obtained from the ANN model were compared with those obtained from the field tests. It is found that the ANN model is efficient in determining the cone penetration resistance of silty sands and yields cone penetration resistance values that are very close to those obtained from the field tests. Additionally, several performance indices such as the determination coefficient, variance account for, mean absolute error, root mean square error, and scaled percent error were computed to examine the performance of the ANN model developed. The performance level attained in the ANN model shows that the ANN model developed in this study can be employed for predicting cone penetration of silty sands quite efficiently.

Development of relationships between swelling and suction properties of expansive soils

In this paper, an attempt is made to correlate swelling properties of expansive soils with their suction properties. To achieve this, both swelling, using a conventional oedometer apparatus, and suction, by dew point potentiameter, WP4, properties were measured on different soils used in the study. With the help of data obtained from experimental investigations, plots were drawn against individual swelling parameters (viz., swelling potential, S, swelling pressure, Sp, & free swell index, FSI) vs. suction property like (a) air-entry value, AEV, (b) initial suction, ψi, (defined as suction measured at natural water content that is determined at the time of collection of a soil from the field) and (c) suction measured at optimum water content, ψOMC, in order to develop generalised empirical relationships between swelling and suction properties. It has been observed that AEV and ψOMC exhibits a linear relationship with S and Sp, whereas ψi exhibits a tri-linear relationship with S and a linear relationship with Sp, respectively. The study finds that AEV and ψi, the former parameter is to estimate Sp & FSI, and the latter one is to estimate S of a soil, are the best useful suction parameters to relate them with the swelling properties. Additionally, efforts were also devoted to predicting the swelling properties (viz., swelling potential and swelling pressure) from suction properties by adopting to an artificial neural network (ANN) modelling tool. Overall, the results demonstrate that resorting to suction properties is a quite promising option for predicting the swelling properties of expansive soils.

Prediction of standard penetration test (SPT) value in Izmir, Turkey using radial basis neural network

Site exploration, characterization and prediction of soil properties by in-situ test are key parts of a geotechnical preliminary process. In-situ testing is progressively essential in geotechnical engineering to recognize soil characteristics alongside. In this study, radial basis neural network ( RBNN ) model was developed for estimating standard penetration resistance ( SPT-N ) value. In order to develop the RBNN model, 121 SPT-N values collected from 13 boreholes spread over an area of 17 km 2 of Izmir was used. While developing the model, borehole location coordinates and soil component percentages were used as input parameters. The results obtained from the model were compared with those obtained from the field tests. To examine the accuracy of the RBNN model constructed, several performance indices, such as determination coefficient, relative root mean square error, and scaled percent error were calculated. The obtained indices make it clear that the RBNN model has a high prediction capacity to estimate SPT-N .