| Home

Overview


Original Research

FWD DATA MODELING TO PREDICT THICKNESS USING ANN APPLICATIONS TO CALCULATE THE REMAINING SERVICE LIFE OF ROAD PAVEMENT SYSTEMS

DADAN RUSLI 1, SOFIA W ALISJAHBANA 2, DADANG MA’SOME 3, NAJID NAJID 4, and SIEGFRIED SYAFIER 5.

Vol 18, No 01 ( 2023 )   |  DOI: 10.17605/OSF.IO/6Z9R3   |   Author Affiliation: Civil Engineering, Faculty of Engineering, Universitas Tarumanegara, Jakarta, Indonesia 1,2,3,4; Civil Engineering, Faculty of Engineering of Universitas Langlang Buana, Bandung, Indonesia 5.   |   Licensing: CC 4.0   |   Pg no: 694-708   |   Published on: 18-01-2023

Abstract

Significant efforts have been made to develop low-cost NDT assays to measure pavement properties. These properties are collected and maintained by a Pavement Management System (PMS). The information contained in PMS is often used by engineers to assess pavement integrity and to determine the salvage value of the pavement life (future useful). In the road management system, the main parameters considered are the functional and structural performance of each section of the road. Functional performance is related to the comfort of road users when passing through it, while structural performance is related to the strength of the pavement system in accommodating the traffic load that passes. In general, functional performance is identified by the International Roughness Index (IRI), while structural performance is usually manifested by the Structural Number (SN). The Falling Weight Deflectometer (FWD) test is one of the most widely used tests to assess the structural integrity of pavements in a non-destructive manner. The modulus of elasticity of the individual pavement layers "recalculated" from the FWD deflection measurements is an effective indicator of the condition of the layer. Most counter-calculation programs currently in use do not take into account the non-linearity of unbound granular materials and fine-grained cohesive soils and therefore do not produce realistic results. The main objective of this research is to develop a tool to calculate the non-linear pavement layer modulus from FWD data using an Artificial Neural Network (ANN). A multi-layer, feed-forward network that uses an error-backpropagation algorithm is trained to estimate the back computation of the FWD function. The synthetic database generated using the Ken Pave nonlinear pavement finite element program was used to train the ANN. Using ANN, we managed to predict the AC modulus and the module ground. The final product is used in calculating the pavement layer modulus from the actual field data obtained.


Keywords

Artificial Neural Network (ANN), FWD Deflection, Ken Pave, Multilayers.