An ANN Based Sensitivity Analysis of Factors Affecting Stability of Gravity Hunched Back Quay Walls
Abstract: This paper presents
Artificial Neural Network (ANN) prediction models that relate the safety
factors of a quay wall against sliding, overturning and bearing capacity
failure to the soil geotechnical properties, the geometry of the gravity
hunched back quay walls and the loading conditions. In this study, a database
of around 80000 hypothetical data sets was created using a conceptual model of
a gravity hunched back quay wall with different geometries, loading conditions
and geotechnical properties of the soil backfill and the wall foundation. To
create this database a MATLAB aided program was written based on one of the
most common manuals, OCDI (2002). Comparison between the results of the
developed models and cases in the data bank indicates that the predictions are
within a confidence interval of 95%. To evaluate the effect of each factor on
these values of factor of safety, sensitivity analysis were performed and
discussed. According to the performed sensitivity analysis, shear strength
parameters of the soil behind and beneath the walls are the most important
variables in predicting the safety factors.
Keywords: Quay Wall; Hunched
Back; Safety Factor; Sliding; Overturning; Bearing Capacity; Artificial Neural
Network
Author: Samir
Karimnader-Shalkouhi, Mehran Karimpour Fard, Sandro Machado
Journal Code: jptsipilgg170030
