الفهرس 
IntroductionOnly 14 pages are availabe for public view
 |  | from | 186 |  |  |
| | | from | 186 | | |
Abstract
The most precise way to determine the axial load capacity is to drive a full-size prototype pile at the site of the proposed production piles and load it to failure. All other methods determine the capacity indirectly, and therefore are less precise. However, load tests also are much more expensive, and thus must be used more judiciously. In this study, reviewed the pile load test and develop two types of models to predict the pile capacity but with less expense. The first one is centrifuge model with the facility of Mansoura Faculty of Engineering which simulate pile load test in the laboratory. The second one is the Artificial Neural Networks (ANNs) model which uses the data of pervious pile load tests to develop the ANN model. Centrifuge modeling is used to simulate piles embedded in Delta soil. Actual soil materials are used. Their physical and mechanical properties are pre-determined before testing. Soil-pile interaction is studied especially with regards to the phenomenon of negative skin friction. Finally, the physical model results were calibrated with the behavior of 1-g model in an area having a ethological section analogous to one of the centrifuge tests. Error back propagation Neural Networks was utilized to predict the bearing capacity of piles. The data of pervious pile load tests in Delta zone are used for learning and testing the ANNs. The ANN output (predicted pile capacity) showed that the maximum error of prediction did not exceed 10 %. Thus, the use of Neural Networks to predict pile capacity seems to be feasible for practical purpose.
OBJECTIVES OF THE STUDY
-Investigate the traditional methods and techniques used to estimate axial pile capacity of single pile.
-Collect the data and results of large numbers of full-scale pile load tests.
-Use the centrifuge modeling to simulate full-scale pile load test.
-Develop an ANN model to predict axial pile capacity.
CONCLUSIONS
- Intersection of all available data concerning pile design loads and corresponding factors of safety shows that the factors are generally conservative.
- It is recommended to perform pile load test to failure, so various methods of estimating ultimate pile capacity could be applied.
- The results of centrifuge models used to simulate pile load tests show that they are close to actual results of pile load tests.
-The centrifuge modeling is a relatively recent tool that is built on a solid mathematical base. It leads to acceptable results when compared with the most dependable pile load test.
-The ANN is tangible in predicting pile bearing capacity but it needs continuous up dating to data-base.