Panel data analysis of surface skid resistance for various pavement preventive maintenance treatments using long term pavement performance (LTPP) data

Various preventive maintenance (PM) treatments have been employed to restore pavement skid resistance for enhanced safety. This paper investigates the effectiveness of PM treatments using panel data analysis (PDA). Panel data analysis investigates the differences of cross-sectional information among treatments, but also the time-series changes within each treatment over time. Panel data with multiple years of friction data for four treatments (thin overlay, slurry seal, crack seal, and chip seal) at various climate, traffic, and pavement conditions are obtained from 255 long term pavement performance (LTPP) testing sections. Both fixed- and random-effects models are developed to evaluate pavement skid resistance performance and to identify the most influencing factors. Results from the PDA models are compared to those from traditional ordinary regression models. Slurry seal is demonstrated to be the most effective treatment. Five factors (precipitation, freezing index, humidity, traffic, and pavement age) are identified to be significant for pavement friction. Fixed-effects panel model is selected for the development of friction prediction models. This study not only demonstrates the capability of PDA for analyzing friction data with cross-sectional and time-series characteristics, but also can assist engineers in selecting the most effective PM treatments for the desired level of skid resistance to reduce traffic crashes. Key words: pavement skid resistance, preventive maintenance, panel data analysis, fixed-effects model, random-effects model. Dans le cadre de cette etude, on examine l'efficacite des traitements d'entretien preventif (EP) en utilisant l'analyse de donnees recueillies au moyen d'un panel (ADP). L'ADP examine les differences d'informations transversales entre les traitements, mais aussi les changements de series temporelles pour chaque traitement au fil du temps. Des donnees de panel avec plusieurs annees de donnees de friction pour quatre traitements (rechargement mince, coulis bitumineux, produit d'etancheite de fissures, enduit d'usure) a differentes conditions de climat, de trafic et de chaussee sont obtenues a partir de 255 sections d'essais de performance a long terme des chaussees (<< LTPP >>). Tant des modeles de panel a effets fixes qu'a effets aleatoires sont developpes et compares aux modeles de regression ordinaires traditionnels. Cinq facteurs (precipitation, indice de gel, humidite, trafic et vieillissement de la chaussee) sont les facteurs les plus importants influant sur la friction de chaussee. Par la suite, le modele de panel a effets fixes est choisi afin de developper des modeles de prediction de friction. Cette etude demontre non seulement la capacite de l'ADP aux fins d'analyser la friction avec des donnees de panel, mais elle peut aussi aider dans la selection des traitements les plus efficaces au niveau de la resistance au derapage desiree et de l'amelioration de la securite. [Traduit par la Redaction] Mots-cles : resistance au derapage des chaussees, entretien preventif, analyse de donnees recueillies au moyen d'un panel, modele a effets fixes, modele a effets aleatoires.
Introduction An effective transportation system should facilitate safe and efficient driving environments for vehicles. Sufficient friction (skid resistance) between tire and pavement plays a crucial role to reduce traffic accidents. [...]