Formulation of penetration resistance, softening point and viscosity of plastic modified bitumen using Genetic Expression Programming

Abstract

Use of plastic waste in bitumen is a sustainable and green technique which can improve asphalt properties and also solve the global issue of plastic waste. The focus of the present study was on the new application of Genetic Expression Programming (GenEP) to create predictive models for some of the most important traditional properties of plastic-modified bitumen, penetration, softening point, and viscosity. Overall, a large database was created by reading literature which offered eleven input parameters (plastics and blending conditions) and then applying the GenEP technique to generate explicit mathematical equations for each of the properties. Then the models were validated using R-squared (R²) value, Mean Absolute Error (MAE), Mean Square Error (MSE), and Root Mean Square Error (RMSE) and all the models were highly accurate with good generalized predictions on training, validation, and test datasets. To better understand the model predictions and also to find a value for each of the discrepancies of each input variable, the analysis included a Shapley Additive Explanations (SHAP) analysis. Information gained using the SHAP analysis also led to some interesting conclusions about the relative importance of some of the input parameters as well as increased transparency and support for close examination of the models. The study confirms GenEP with SHAP analysis as a suitable modelling predictive method for asphalt research and supports ongoing use of plastic waste in road construction as part of a sustainable infrastructure development process.