Analysis of roughness deterioration of bitumen sealed unbound granular pavements for use in road asset management modeling

The road system under the care and control of the State of Queensland, Australia has an estimated replacement value of $23 Billion (US$12 billion), excluding structures. Pavement management systems for recording, rating, and predicting a pavement's condition are used to prudently manage this large asset. Tne of the fundamental components of a pavement management system is the method of determining a pavement's rate of deterioration over time. Throughout the world, "pavement roughness" is one of the most widely used methods of measuring the performance of a pavement. Queensland is a large State, having an approximate area of 1.7 million square kilometres, and incorporates many small communities separated by vast distances. The majority of bitumen sealed pavements exist in low traffic volume rural areas, and consist of either a flexible unbound granular pavement or a semi-rigid modified granular pavement. This study has focussed on the roughness progression of these pavement types. Roughness data has been collected at varying frequencies since 1987, across the State's road network. In order to understand roughness progression behaviour, this research has considered 16,000 pavement segments (each lkm) from all parts of the State. This number is significant, as many previous research efforts, which form the basis of the current roughness progression models, have been based on the study of typically between forty and several hundred pavement segments. This extensive database has been used in this study to investigate trends in roughness pogression over time (~R). Traditionally, roughness progression has been represented in many road asset management models as an exponential relationship, providing rapidly increasing roughness values as the pavement approaches the end of its theoretical life. However, this relationship has not been commonly observed in the State's pavement asset. Investigations revealed that a linear relationship, termed the Linear Roughness Progression Rate (LRPR), represented the roughness progression of the pavement segments very well, and thereby assists in the identification of poor performing pavement segments. Pavement maintenance costs can also be used to identify poor performing pavement segments, based on excessive use of pavement maintenance funds to maintain functionality. This information assists in 'unmasking' pavements that may possess an umepresentative LRPR. For this reason, the effects of pavement maintenance costs (~PMC) on roughness progression have also been examined. Independent variables which are typically considered to affect roughness progression, have been examined with LRPR in this study. These variables include traffic volume, traffic loading, pavement type, subgrade soil classification, climate (rainfall, temperature, Thomthwaite Index), pavement age, and width of bitumen surface (seal width). It has been established that the roughness progression of each individual pavement segment is unique. No particular formula or model has been identified, which can accurately predict the roughness progression of any particular pavement segment from a population, based upon known values of the independent variables. In order to develop an understanding of the effects of these variables on pavement performance, a means of rating a pavement's condition using a combination of LRPR and excessive pavement maintenance expenditure has been established. Subsequently, a discrete rating scale from "Good", "Fair" to "Poor" has been used. The effects of the independent variables on pavement rating have been examined across the population of study pavement segments, using summary charts. This has enabled the proportions of pavement rated within each category to be examined with changes in each independent variable, to establish whether that variable has a global influence on pavement rating. For instance, whether the proportion of pavements rated "Poor" varies with Annual Average Daily Traffic. These summary charts provide the basis of a Network Performance Profile, which has yielded a better understanding of the variables globally influencing pavement rating across the State's road network. Further, the Network Performance Profile provides insight into the inbuilt risk of the past and current pavement design and delivery system. The knowledge gained from a whole of network analysis of condition may form a catalyst for the assessment of material quality, design methods, construction technology, and contract delivery, to ensure that pavements have the best chance of performing well. Currently, road network performance is commonly defined only by a measure of the road network's 'Absolute Condition'. The Network Performance Profile developed in this study will add new measures of pavement performance to the current definition. By including a Current Network Profile, and a more robust method for predicting mid term (five year) roughness values, asset managers will have a suite of information on historic, current, and estimated future road network condition, to use in evaluating the impact of previous and current management decisions. Pavement roughness prediction analysis, based on site-specific extrapolation of LRPR, has shown that at least six time-series roughness data points are required if an accurate short term prediction (4 to 6 year) is to be achieved. It is believed that the research of roughness progression has highlighted the unique nature of pavement segment deterioration. It is interesting to note that all categories of pavements suffer from a range of good to poor performance, and this finding is considered to be one of the keys in understanding pavement behaviour. At a network level, the study of performance has been considered very successful and has not only provided a robust methodology for quantifying road network performance, but also provides information for input into the improvement of technical policies and funding distribution decisions