Roughness, or ride quality, is one of the most important pavement performance parameters. For DOT engineers, it is a measure of pavement surface distortion or variation in pavement surface elevation; it can be used to trigger appropriate maintenance treatments, approve new and rehabilitated roadways, and determine contractors’ performance incentives. For the traveling public, roughness is an indicator of a comfortable ride; it is directly used by the public to judge pavement condition, it also affects user costs, including fuel, repairs, and vehicle depreciation.
Since 1986, the International Roughness Index (IRI) has been widely used to quantify pavement roughness, and is the parameter by which pavement smoothness is defined within the Mechanistic-Empirical Pavement Design Guide (MEPDG). IRI is measured during state DOTs’ routine pavement surveys, and is required to be reported to the Highway Performance Monitoring System (HPMS) for the National Highway System (NHS) on an annual basis.
Although Federal Highway Administration (FHWA) has several recommended IRI thresholds, they are not locally calibrated and thus do not reflect the ride smoothness perceived by the traveling public in North Carolina. This research project was conducted to address this issue.
To ensure a geographically balanced sampling, a total of 9 counties, with at least 2 counties in each region of North Carolina (Mountains, Piedmont, and Coastal), were selected. Of the counties selected in each region, at least one was a rural county, and another one was an urban county. In each selected county, two roadway loops, one urban and one rural, were identified. These loops were selected based on the goal that each loop should include sections that have different pavement types (flexible or ASP and Jointed Concrete Pavement (JCP)), smoothness, and speed limits.
Research participants were recruited through face-to-face interviews. The recruitment process was designed to randomly enroll local drivers in selected counties at locations such as senior centers, community colleges, universities, grocery stores, supermarkets, hospitals, and public health service centers. A total of 241 participants were recruited and participated in this research project. Their perception of the smoothness of roadways were collected in both numerical (0-5) and categorical (Acceptable and Unacceptable) format. A total of 3,539 observations were collected. Among these observations, 158 were deleted due to missing values. Eventually, a total of 3,381 valid observations were used in further analyses.
The quality of the survey data was satisfactory. Sixty-five out of 88 Kendall’s W coefficients are larger than 0.5 and also have a p-value that is less than 0.025, indicating that approximately 74% of participants agreed with each other when assessing the same roadway sections, and the null hypothesis that there is no agreement between ratings of all participants was rejected at a 0.05 level.
Through analyses of participants’ perceived ride quality ratings, the following conclusions were obtained: The most influential factors in affecting perceived ride quality of a particular roadway section were measured IRI values, the speed limit, and participants’ seating positions in the survey vehicles. Other factors, including types of survey vehicles, geographic regions the roadway section is located, and the type of pavement of the section, were not statistically significant.Roadways that had greater measured IRI values and higher speed limits were more likely to be rated as “Unacceptable”.Participants seating in window positions were more likely to rate roadways favorably.Asphalt urban and rural sections were more likely to be rated as “Acceptable” than JCP sections.In North Carolina, if the measured IRI value of a roadway section is less than 103 inches/mile, most likely this section would be rated as “Acceptable” by the general driving public; most likely it would be rated as “Unacceptable” if its measured IRI value is greater than 151 inches/mile.The target initial IRI value for a new construction project should be between 60 and 70 inches/mile. For a “perfect” roadway section, its IRI value should be between 50 and 60 inches/mile. If the IRI value is greater than 182 inches/mile, the roadway section is considered as “Very Unacceptable”.IRI index was derived using a linear relationship (IRI vs. AGE) developed in this study. It was assumed that: 1) when IRI = 55 inches/mile, IRI Index value = 100 (perfect condition); and 2) when IRI = 182 inches/mile, IRI Index value = 0 (Very Unacceptable).IRI models were developed using the IRI index values determined in this study. It should be noted that currently the NCDOT uses distress models (distress index vs. pavement age) and performance models (PCR vs. pavement age) to make treatment decisions, and IRI models are not considered because they have not been developed before. Therefore, it is suggested that the developed IRI models should not be used to determine pavement service lives before these models are added to the decision tree as a separate branch.It was concluded that smooth pavements (smaller initial IRI values) deteriorated at a slower rate and therefore had longer service lives. In addition to initial IRI, other factors such as traffic volume and environmental factors also work jointly to impact how the network performance IRI values change over time.It should be noted that the results from the JCP sections are not sufficient to draw explicit conclusions because of the limited number of JCP sections that were surveyed in this study.
The following recommendations are proposed for future research: It is recommended that the developed IRI models to be included in the NCDOT treatment decision-making process for increased PMS performance. A separate branch can be developed using these models and added to the decision tree. IRI trigger values can be determined to suggest appropriate treatments.It is recommended that more JCP sections to be studied in order to develop IRI limits and targets for network management and construction approval purposes.It is recommended that balanced sampling (fixed sample size and stratification) of varying types of roadways (e.g., flexible, rigid, and composite) should be required whenever possible in order to obtain informative research results. In this study, some contrasts were not performed because of unbalance samples.The following data collection methods have proven to be effective in this research project, and are recommended for future studies in this research area:The face-to-face recruiting method is more effective than the phone call method.It is necessary to over-recruit participants in order to avoid “no-shows”. Essentially, this is to ensure that the appropriate sample sizes are achieved, and that the findings are statistically significant.Locations of roadway loops should be close to each other. It took approximately 2 hours to complete one survey (including 3 roadway loops: one ASP urban, one ASP rural, and one JCP) in this research project, which caused uncertainties in logistics.
It was concluded in this research project that initial IRI, traffic levels, and environmental factors are important factors that can affect long-term network IRI. Other factors, such as pavement designs, are recommended to be considered in future studies.
