837
Analysis of Ultimate Bearing Capacity of Single Pile Using the Artificial Neural
Networks Approach: A Case Study
Analyse de la capacité portante ultime d’un pieu unique à l'aide de la méthode des réseaux de
neurones artificiels : une étude de cas
Wardani S.P.R.
Civil Eng. Dept, Engineering Faculty, Diponegoro University & Indonesian Road Development Association, Indonesia
Surjandari N.S.
Civil Eng. Department, Engineering Faculty, Sebelas Maret University, Indonesia
Jajaputra A.A.
Professor Emeritus at Institut Teknologi Bandung & Visiting Professor at Diponegoro University, Indonesia
ABSTRACT: Degree of certainty, accuracy, complexity, and non-linearity are things that are adhere to geotechnical problems.
Solutions using conventional approaches, although were still used in geotechnical problems require a large number of assumptions for
the determination of geotechnical parameters. Currently new approaches emerge, including the "artificial intelligence", one of which
is a neural network (NN).This study aims to apply NN model for prediction of ultimate bearing capacity of single pile foundation, was
named NN_Qult model. The results of analysis model were then compared with Meyerhof, 1976 and Briaud ,1985 formulas. At the
stage of modeling, data from full-scale pile load test and SPT were used. The selected input variables are: d (pile diameter), L (length
of the pile embedded), the N60 (shaft) value, and the N60 (tip) value.
The study generates design Charts that are expected to predict
the ultimate bearing capacity of a single pile foundation.
The results showed that neural networks can be used for prediction of
ultimate bearing capacity of single pile foundation. This is particularly due to the sensitivity analysis results indicated the suitability
of artificial neural network model with existing theories.
RÉSUMÉ : Degré de certitude, précision, complexité et non-linéarité sont des difficultés inhérentes aux problèmes géotechniques.
Les approches conventionnelles, bien que toujours utilisées dans les problèmes géotechniques nécessitent un grand nombre
d'hypothèses pour la détermination des paramètres géotechniques. Actuellement de nouvelles approches émergent, notamment «
l'intelligence artificielle », dont l'une des formes est le réseau de neurones (NN). Cette étude vise à utiliser le modèle de réseau de
neurones pour la prévision de la capacité portante ultime de fondation sur pieu unique, elle a été dénommée le modèle NN_Qult. Les
résultats du modèle d'analyse ont ensuite été comparés avec les formules de Meyerhof, 1976 et de Briaud, 1985. Lors de l'étape de la
modélisation, des données provenant d’essai de chargement de pieux grandeur nature et de données SPT ont été utilisées. Les
paramètres retenus sont les suivants: d (diamètre du pieu), L (longueur du pieu), les valeurs N60 (frottement latéral et résistance de
pointe).
L'étude a abouti à des graphiques de conception prévus pour prédire la capacité portante ultime d'une fondation sur pieux
unique. Les résultats ont montré que les réseaux neuronaux peuvent être utilisés pour la prédiction de la capacité portante ultime de
fondation sur pieu unique. Cela est notamment dû aux résultats de l'analyse de sensibilité qui a indiqué la cohérence du modèle de
réseau de neurones artificiel avec les théories existantes.
KEYWORDS: Ultimate bearing capacity, a single pile foundation, the neural network models, design Chart.
1 INTRODUCTION.
Mathematical model (white box model) is a form that has been
established in the field of science. This model was created using
the basic principles of physics and mechanics followed by a
series of observations, used for simulation, prediction, and
analyze the behavior of a system. Appropriate mathematical
model when the underlying condition of a system are known,
the measured uncertainty and inaccuracy did not reduce the
accuracy of the model (Grima, 2000; Rahman and Mulla, 2005).
Problems in geotechnical engineering are generally complex, so
that its exact solution is the probability (Djajaputra, 1997;
Griffith et al., 2002). Uncertainty and inaccuracy is almost
always found as to seek geotechnical parameters. There are
many factors that are not known with certainty because only a
limited number of sampling used. This condition leads to the
use of mathematical models for the solution in a difficult
geotechnical problems (Rahman and Mulla, 2005; Prakoso,
2006).
Artificial neural network model has been started in the field
of geotechnical engineering. The difference
between neural
network model and matematical model is the artificial neural
network model does not require the initial assumption of
physical laws (a priori any physical law) of a system, when new
data are found, so the ability to predict can be upgraded with
relative ease (Javadi et al., 2001; Hashash et al., 2004; Right
and Faez, 2004).
The purpose of this study is to make an artificial neural
network model for calculating the limit bearing capacity of a
single pile foundation and then its ability is compared with
some existing methods.
2. BASIC THEORY
2.1 Ultimate bearing capacity of single pile foundations.
The axial limit bearing capacity (ultimate) of the pile
foundation (
Qult
) is assumed to be the result of 2 (two)
mechanisms i.e. the side friction resistance of foundation (
Qs
)
and end bearing resistance of foundation (
Qt
) so that the net
ultimate bearing capacity due to the axial load pressure is as in
Eq. 1 (Bowles , 1988).
(1)
by:
Q
ult
= ultimate bearing capacity
Qt
= end bearing resistance
Qs
= friction resistance
W
= weight of pile foundation.
