The design of flexible pavements is in the process of adaptation, the tendency is to
gradually replace empirical methods with empirical-mechanistic methods, such as the
MeDiNa method. However, the MeDiNa software requires complex tests for the
complete characterization of the materials, which are not widely accessible, especially
in the North region of Brazil. This induces the designer to search for correlations in the
literature to obtain the MeDiNa input parameters, which adds uncertainties and
produces errors in the dimensioning. Therefore, this work aims to generate a model for
predicting Wheel Track Sink (ATR) as a function of traffic, base and subgrade
parameters. For this purpose, four parameters of the base and subgrade were varied
in five levels and the traffic in two levels, which resulted in 1250 simulation scenarios,
which were introduced in MeDiNa for the calculation of ATR. With data from 1250
simulation scenarios, a multiple linear regression model was created to predict the
results of ATR and these results were compared with those calculated by the MeDiNa
software. The main results were: the ATR prediction model presented R² = 0.693, that
is, approximately 70% of the results are explained by the model, a good result
considering that only 5 variables were used to characterize the materials; The
permanent deformation parameters Ψ4 and Ψ3 were the parameters that most
modified, directly and inversely proportionally, respectively, the ATR results; The
Poisson coefficient and the N number (traffic) were the parameters that least modified
the ATR results. Finally, it is concluded that the objective was achieved and that the
model is an important tool to assist the designer in making decisions about the design
of floors using the MeDiNa software.
|
