ESTIMATING THE FRICTION PERFORMANCE OF HOT MIX ASPHALT PAVEMENTS BASED ON AGGREGATE PROPERTIES AND ROUTE CHARACTERISTICS: ANALYSIS, MODELING AND VALIDATION

Abstract

Traffic accidents are one of the major causes of death in the United States. In 2008 alone, more than 37,000 fatalities occurred, accounting for one fatality every thirteen minutes. More than one tenth of fatal accidents occur when pavements are wet

and slippery. In wet conditions, a water film is created between the pavement surface and the tire, thereby reducing the amount of available friction.

There are several factors that affect the level and type of friction between tires and a wet pavement surface. Some of these factors are microtexture and macrotexture, age of

pavement, seasonal and environmental factors, traffic level and composition, individual and blend aggregate properties, binder used in mix, and road location/geometry. The

research presented in this dissertation explores the impact of aggregate and mixture properties as well as the role of route characteristics, such as traffic intensity and

composition, on the friction performance of Hot Mix Asphalt (HMA) pavements.

In the research, various databases for construction, material, pavement management and traffic condition were examined. The data included 5 years of pavement friction readings,

construction and material data, and traffic monitoring data. The research included reviewing aggregate quality requirements and friction measurements, and compiling,

categorizing and examining the various databases to develop a working dataset/s. In addition, a methodology was developed to isolate and analyze data specific to a given

roadway constructed using a known type of aggregate and mix material. The results were then used to estimate pavement friction service life in terms of cumulative traffic loading.

Multivariate Regression methods were employed to establish the relationship between Friction Number (FN) and cumulative AADT, for specific aggregates.

The research also included establishing relationships between materialproperties/route characteristics and pavement friction, and investigating/developing a

model that can be used to predict the friction performance of pavements based on these factors. Partial Least Squares (PLS) Regression, a type of Structural Equation Modeling

(SEM) method, was used to extract factors from datasets in order to formulate, test and validate several models out of which the most significant model was selected.