Abstract
Determination of the depths of the embedment of contiguous piles requires extensive soil investigation to obtain the soil physical parameters. In addition, a large number of such piles involved in restricted access projects make that depth an essential problem. A simple mathematical model for predicting the depth of embedment using the height of the retained soil, the standard penetration test values, and the bulk unit weight of the soils encountered for 261 pile data sets was introduced using an artificial neural network approach. The coefficient of determination equals to 0.99 for the tested the data reveal that the depth of embedment was accurate against those achieved in Zakho interchange. The importance and parametric studies obtained show that the major parameter which affects the depth of embedment was the height of the retained soil whereas the effect of other parameters is relatively less.