Estimation of the allowable bearing capacity of soil in some municipalities of the province of Pampanga using neural networks

Urbanization is evident in some municipalities in Pampanga specifically in San Fernando and Santo Tomas. Those municipalities being developed, requires a prediction of the load-bearing capacity of soil. Predicting the bearing capacity of soil will help the construction of foundations that can resist geotechnical failures of soil. The bearing capacity was calculated using the local shear failure equation of the Terzaghi’s bearing capacity formula. Also, the bearing capacity was predicted using Artificial Neural Network for those locations which has available data with the N60 value, friction angle, unit weight, and footing width as dependent parameters. In addition, weighted k-Nearest Neighbor was used to estimate the value of the bearing capacity for those locations without available data with the latitude and longitude of the points as the basis to calculate the weighted distances between the points. The results show a coefficient of correlation of approximately equal to 1 and a mean squared error of at least 0 for a hidden layer of 10 which proves that ANN is an efficient way in predicting the bearing capacity. Based on the sensitivity analysis, it was found out that the unit weight is the most significant parameter affecting the value of the bearing capacity. Five nearest neighbors were considered in the prediction of the allowable bearing capacity using weighted k-NN. A relationship between the N60 value, soil classification, and the bearing capacity was observed. It was concluded that the N60 value and soil classification are the two determining factors how will the value of the bearing capacity be because it affects the consistency of the soil and most of the parameters are dependent on those two variables. Both municipalities have a dominant soil classification of SM. The value of the bearing capacity ranges from a minimum of 20 kPa to a maximum of 630 kPa for a specific area. The values are all estimated based on the considerations and limitations provided by the equations and correlations of all the parameters and the equations