PREDICTING LOAD CAPACITY OF PRECAST CONCRETE PILES USING SPT AND ARTIFICIAL NEURAL NETWORK

Autores

  • Juliele N. Jesus
  • Maria do Socorro C. São Mateus
  • Anderson de S. M. Gadéa

Palavras-chave:

Estimate the bearing capacity, Precast concrete piles, Semi-empirical methods, Artificial Neural Networks (ANN)

Resumo

One of the great challenges of foundation engineering is the calculation the bearing capacity of piles
because it requires, in theory, the estimation of soil properties, its changes by the execution of the foundation and
the knowledge of the soil-pile interaction mechanism. The semi-empirical methods of Aoki-Velloso [1] and
Décourt-Quaresma [2] are the most widely used to estimate the bearing capacity of concrete piles in Brazil.
However, these methods were developed for a group of soils from a specific region, so it is necessary to adjust
them to adequately represent the soil-pile interaction mechanism in soils from regions different from those initially
studied. In the geotechnical engineering, Artificial Neural Networks (ANN) have shown potential in determination
of the bearing capacity of deep fundations. In this paper, an ANN model is implemented to predict the bearing
capacity of precast concrete piles based on data from 126 Standard Penetration Test (SPT) and their respective
load tests results, static and dynamic pile testing. Based on the results obtained, the ANN model may represent a
promising solution for the design of precast concrete piles.

Downloads

Publicado

2024-05-30

Edição

Seção

Artigos