TY - JOUR
T1 - Multivariate chemo-rheological framework for optimizing laboratory aging protocols of paving binders
AU - Khalighi, Sadaf
AU - Ma, Lili
AU - Mosleh, Yasmine
AU - van Lent, Diederik
AU - Varveri, Aikaterini
PY - 2024
Y1 - 2024
N2 - This study aims to improve laboratory aging procedures for bituminous materials to better replicate field conditions. Two binders and mixtures were subjected to various levels of humidity, temperatures, pressures, film thicknesses, and aging durations. By comparing these lab-aged samples to field-aged samples, the study aims to simulate real-world aging more accurately. Fourier-transform infrared (FTIR) spectroscopy and frequency sweep tests were employed to analyse these samples. Multivariate techniques—Principal Component Analysis (PCA), Multiple Linear Regression (MLR), and Support Vector Regression (SVR)—were used to explore chemical and rheological relationships, evaluate the interchangeability of aging factors, and quantify the equivalency between laboratory and field aging. The findings revealed that increased temperature, pressure, and duration lead to more oxidative products. The PCA distinguished between two binders and aging trends, highlighting the importance of both FTIR and rheological measurements. The SVR model demonstrated strong predictive performance for rheological properties, identifying critical FTIR region, 710–912 -1cm. By MLR model, optimal aging conditions to simulate nine years of field aging for porous asphalt and stone mastic asphalt were back-calculated. The Euclidean distance found laboratory conditions that closely match field-aged samples. SVR models provided predictions of simulated field aging time for various laboratory aging conditions.
AB - This study aims to improve laboratory aging procedures for bituminous materials to better replicate field conditions. Two binders and mixtures were subjected to various levels of humidity, temperatures, pressures, film thicknesses, and aging durations. By comparing these lab-aged samples to field-aged samples, the study aims to simulate real-world aging more accurately. Fourier-transform infrared (FTIR) spectroscopy and frequency sweep tests were employed to analyse these samples. Multivariate techniques—Principal Component Analysis (PCA), Multiple Linear Regression (MLR), and Support Vector Regression (SVR)—were used to explore chemical and rheological relationships, evaluate the interchangeability of aging factors, and quantify the equivalency between laboratory and field aging. The findings revealed that increased temperature, pressure, and duration lead to more oxidative products. The PCA distinguished between two binders and aging trends, highlighting the importance of both FTIR and rheological measurements. The SVR model demonstrated strong predictive performance for rheological properties, identifying critical FTIR region, 710–912 -1cm. By MLR model, optimal aging conditions to simulate nine years of field aging for porous asphalt and stone mastic asphalt were back-calculated. The Euclidean distance found laboratory conditions that closely match field-aged samples. SVR models provided predictions of simulated field aging time for various laboratory aging conditions.
KW - Field aging
KW - Laboratory accelerated aging
KW - MLR
KW - Multivariate analysis
KW - PCA
KW - SVR
UR - http://www.scopus.com/inward/record.url?scp=85211129485&partnerID=8YFLogxK
U2 - 10.1016/j.matdes.2024.113520
DO - 10.1016/j.matdes.2024.113520
M3 - Article
AN - SCOPUS:85211129485
SN - 0264-1275
VL - 248
JO - Materials and Design
JF - Materials and Design
M1 - 113520
ER -