Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                

Cover Sheet

Download as pdf or txt
Download as pdf or txt
You are on page 1of 18

COVER SHEET

Hunt, Philip D and Bunker, Jonathan M (2002) Road Performance Studies Using Roughness Progression & Pavement Maintenance Costs. In Proceedings Queensland Department of Main Roads Roads System and Engineering Forum, 2002, Brisbane.

Accessed from http://eprints.qut.edu.au

Road Performance Studies Using Roughness Progression & Pavement Maintenance Costs Hunt PDa and Bunker JMb (May 2002)
a
b

Queensland Department of Main Roads, Roma; pdhunt@mainroads.qld.gov.au School of Civil Engineering, Queensland University of Technology; j.bunker@qut.edu.au

Pavement management systems assist engineers in the analysis of road network pavement condition data, and subsequently provide input into the planning and prioritisation of road infrastructure works programs. The data also provides input into a variety of engineering economic analyses which assist in determining the future road network condition affected by a range of infrastructure funding scenarios. The fundamental calculation of future pavement condition is commonly based on a pavement age versus pavement roughness relationship. However, roughness - age relationships commonly used in pavement deterioration and economic modelling do not take into account the pavements historical performance, rather, an average rate of roughness progression is commonly assigned to each pavement based on its current age or current roughness measurement. This research project has undertaken a comprehensive evaluation of pavement performance by examining roughness progression over time and other related variables. A method of calculating and effectively displaying roughness progression and the effects of pavement maintenance has been developed, and has subsequently provided a better understanding of pavement performance. This understanding has led to a methodology of calculating and reporting road network performance, and measuring the performance of the pavement design and delivery system within Queensland, Australia. Methods of how this information can be used to improve the accuracy of roughness progression prediction were also investigated. Of particular interest to pavement asset managers should be the discussion and definition of poor pavement performance, the definition of network wide performance, the findings in relation roughness-age relationships, insight into pavement maintenance cost analysis, and future roughness prediction.

1. INTRODUCTION Queenslands State and National Highway Road system has an estimated pavement replacement value of $12US billion. Pavement management systems are used to assist engineers in the management of these large assets and aid in the recording, rating, and prediction of a pavements condition, which in turn assists in prudent and sophisticated management of the road network. One of the fundamental components of a pavement management system is the method of determining a pavements 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 and is subsequently the primary focus of this research. Queensland is a large State consisting of many small communities separated by vast distances. Therefore, the majority of bitumen sealed pavements exist in low traffic volume rural areas and consist of either a flexible unbound granular pavement or a semirigid modified granular pavement. This project has concentrated its investigation on the roughness progression of these pavement types.

2. RESEARCH OBJECTIVE As shown in Figure 1, the traditional roughness versus age relationship used in pavement asset management modelling can vary substantially. Regardless of the model used, each model assigns a roughness progression rate based on current age or current roughness value. Within each model, a group or category of pavements with the same age or current roughness value would be assigned the same roughness progression rate in isolation to the pavements historical performance.
6.00

PDAT QDMR(1999)

4.51

3.03

1.54

IRI scale

FIGURE 1 Comparison of the Age/Roughness Relationship for Four Roughness Progression Models(1) Even though the ARRB and HDM models also include climate, structural, age, traffic loading, and surface maintenance variables, similarly categorised pavements will also be allocated the same rate of roughness progression. Road network engineers often observe variations in pavement performance for roads that exist in the same category. The most obvious example of this is the comparison of road projects constructed adjacent to each other at the same time. Roads constructed at the same time using the same or similar materials on the same subgrade soil type in the same physical and traffic loading environments, often display varying rates of deterioration. Figure 1 also indicates that many of the roughness progression models predict an exponential increase (or rapid increase) in roughness as the pavement approaches the end of its theoretical life. Once again, road network engineers rarely observe such a rapid rate of pavement deterioration in practice. This is possibly due to the maintenance applied to ensure that the pavement is maintained in the safe and functional condition expected by road users. The above anecdotal evidence suggests that field observations do not always align well with the roughness progression models used. Therefore, the aim of this research is to better understand the performance of pavements by closely studying roughness progression and pavement maintenance history, so that improved methods of roughness progression prediction can be achieved.

To assist in achieving this aim, pavement roughness data for 26,557 pavement segments each 1km long has been supplied by the Queensland Department of Main Roads. This data includes physical pavement attributes and pavement maintenance cost information, and is supplemented with climate and soil type data. This data has been used to investigate : the effect of pavement maintenance on roughness progression; roughness progression trends and its effect on pavement performance; a macroscopic view of road network performance; the effect of independent variables on pavement performance; and the use of the above information to improve roughness progression modelling.

3. ROUGHNESS Roughness is literally the measure of how uneven or irregular a road surface is. It is a useful term for the condition of a pavement, because it is a condition directly experienced by motorists. It is commonly reported by either the NAASRA Roughness Measurement (NRM) method which is measured using the NAASRA Roughness Car, or by the International Roughness Index (IRI) which is calculated by applying a computer quarter car model to road profile data collected via laser profilometer. NRM can be reliably converted to IRI by a linear equation (and vica versa), where required. Historically, the Queensland Department of Main Roads (QDMR) has collected NRM using the Roughness Car (a dynamic response type device) and reported both NRM and IRI, with NRM being the most readily used. QDMR implemented laser profilometry in 2001 but still maintains the reporting of both NRM and IRI. NRM is measured in roughness counts with one count equalling 15.2mm of accumulated vertical movement between the vehicles differential (unsprung mass) and the body (sprung mass). Table 1 is useful in understanding the relativity of roughness values, and provides a descriptive condition of NRM and the corresponding IRI value currently used for reporting the condition of various traffic volume roads in Queensland.
Traffic Ranges : AADT Descriptive Condition Excellent Very Good Good Poor Very Poor < 500 <80(2.43) 81-95(2.88) 96-130(3.92) 131-180(5.40) >180 501-1000 <60(1.84) 61-95(2.88) 96-110(3.30) 111-130(3.92) >130 1001-10,000 <60(1.84) 61-80(2.43) 81-95(2.88) 96-110(3.30) >110 >10,000 <60(1.84) 61-70(2.14) 71-80(2.43) 81-95(2.88) >95

TABLE 1 Roughness NRM(IRI) Values for Varying Traffic Ranges(2)

A 1996 Coopers & Lybrand survey, undertaken in the United States, showed that pavement smoothness (opposite of roughness) is the primary concern of the travelling public (3). A recent RACQ (4)survey conducted in Queensland Australia, also confirmed that road roughness is a primary concern for road users. Surprisingly, roughness received a higher number of complaints/concerns than that recorded for narrow roads which was previously believed to be a more important issue. Roughness is seen as an important road condition measure right across the world. Martin (5) stated that roughness is the most widely used condition parameter because roughness data is relatively inexpensive to capture, is an objective measure, correlates well with road user costs, and is accepted as the most relevant measure of the long term functional behaviour of a pavement network. 4. CAUSES OF ROUGHNESS The study of roughness progression with time is a complex phenomenon and is viewed by Paterson (6) as a composite distress comprising components of deformation due to traffic loading and rut depth variation, surface defects from spalled cracking, potholes, and patching, and a combination of ageing and environmental effects. A great deal of effort has been invested in the study of roughness progression of pavements over time. Three major efforts include the development of the World Banks HDM-3 model between 1974 and 1987, the current Long Term Pavement Performance (LTTP; ex SHRP) program in the USA, which commenced in the early 1990s, and the Australian Road Research Boards (ARRB) LTPP sites monitoring since the early 1990s. Each of these studies indicates that roughness progression is complex and that considerable variability in the rate of roughness progression between similar pavement types is experienced. Consequently it is difficult to define parameters that can reliably predict the roughness of a pavement. Figure 2 was developed in this study from a review of the literature and experience in pavement asset management to illustrate the complexity of influences on the roughness, and hence performance, of a pavement. Influences relating to construction quality, material quality, and maintenance quality, are not contained in the figure; however these will also influence pavement performance. There are also many material characteristics such as microscopic particle behaviour, stone size and shape, permeability, capillary rise, etc, that influence pavement performance. However, it is difficult and typically not cost effective to measure these characteristics for all pavements making up a network.

Drying Evaporation Oxidation of bitumen Temperature Variance

Rain during life Rain during construction

Traffic mix Past/Future loading Routine maintenance Gravel Issues: Strength/Moduli Type of Source Rock Gravel Processing Permeability Depth
FORCE

ve me nt Ag e

Seal width Seal age Resurfacing maintenance

BASE SUBBASE SUBGRADE

Water Table Reactive Soil FORCE

Non-Reactive Soil

DRAINAGE

It is hypothesized that the wide variation in pavement performance is attributable to the chaotic system in which pavements operate, as evidenced by Figure 2. This system promotes different proportions of influences that exist in seemingly similar environments, thus producing different behaviours. Because roughness is a measure of the effect of the manifestation of these influences, it is considered that consistent time series monitoring of roughness is useful. However, the prediction of future roughness is still considered a somewhat imprecise science and one that is difficult to predict across a population of pavements.

Cross Section

Pa

Cross Section

Long Section

FIGURE 2 Influential Factors on the Roughness Progression of an Unbound Granular Pavement Interestingly, current Australian granular pavement design methods do not use roughness as a direct pavement design input parameter, however, the parameters used in pavement design are also often used in detailed roughness progression modelling. Thus, there is an indirect link between pavement design theory and roughness progression modelling. 5. ROUGHNESS PROGRESSION MODELS The major model types currently used throughout the world have been categorised into the following four groups; Causal Models, Family Group Data-Fitting Models, Site Specific Data-Fitting Models and Pattern Recognition Models. 1) Causal Models attempt to define the root cause or parameters of roughness progression. Equations are developed by subjecting the causal parameters to a variety of statistical techniques and mechanistically derived equation forms. Examples of this type of model include the HDM-3 Incremental Model, HDM-3 Aggregate Model, ARRB TR Project Model, and ARRB TR Network Model. 2) Family Group Data-Fitting Models predict future roughness progression based on the average deterioration curve for a series of similar type pavements.

3) Site Specific Data-Fitting Models base the future prediction for each individual pavement segment on the actual history of progression. 4) Pattern Recognition Models such as Artificial Neural Networks (ANNs) can store and recognise complex patterns described by many independent variables. When establishing a future prediction, the ANN relies on past similar patterns to predict the performance of the pavement. A review of these model types has shown that, because of the varying individual performance displayed by many pavements, the Site Specific Data-Fitting Models fit well into a pavement management environment and tend to provide a greater accuracy of roughness progression prediction when compared to multi-variable regression, family grouping techniques and causal models. 6. THE EFFECT OF PAVEMENT MAINTENANCE Pavement maintenance works has the potential to upset or mask the true roughness progression rate of a pavement, and thus it was important to apply some effort into the investigation and understanding of this issue. Ideally, the study of pavement maintenance costs and the interaction with pavement roughness would involve the measurement of the change in roughness with the associated pavement maintenance works. However, due to several data limitations the ideal situation was not a practical one to achieve. Although the interrelationship between pavement maintenance and roughness could not practically be studied, it was relatively easy to investigate the rate of pavement maintenance costs for any individual pavement segment. Therefore a study of the pavement maintenance costs, in isolation, was undertaken. This type of analysis could only provide an indication of those pavements that perform poorly with respect to pavement maintenance costs over time and as such provide a flag against those pavements where a roughness progression relationship may be either difficult to calculate or potentially unreliable. A study of pavement maintenance data was undertaken and a process developed to identify pavements significantly affected by pavement maintenance. The masking effect of maintenance does not always directly show itself as a disturbance in the roughness progression data. Two examples of pavement segments that have been identified with excessive maintenance expenditure are shown in Figure 3. A representative value for roughness progression is not sought where excessive maintenance expenditure has occurred.

200 180 160 140 120

Time (Years) V Roughness (NRM)

Two direct effects of pavement maintenance

101

105

107 109 108 91

100

82

90

80 60 40 20 0

1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000

200 180 160 140 120

Time (Years) V Roughness (NRM)

The invisible effect of excessive pavement maintenance

101

103

80 60 40 20 0

94

Data Error

1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000

FIGURE 3 Masking Effect of Pavement Maintenance on Roughness Progression Five-year average pavement maintenance costs (5yr-APMC) were used following previous research (7) which had indicated that smoothing the data over time provided a better indication of the pavement maintenance regime being demanded by the pavements current performance. The 5Yr-APMCs were investigated with many independent variables including Sealed Width, AADT, Annual Traffic Loading, Treatment (Pavement) Age, and Subgrade Classification. No distinct and reliable trends could be represented using a parametric relationship. A large scatter of results was recorded for all of these parameters. This result indicates that there are pavements in all categories requiring large amounts of funds to survive, whilst others require very little. It is suggested that this finding is an indication of the unique nature of pavement performance. Further investigation using predefined engineering categories of AADT (low to high) enabled a better understanding of the 5Yr-APMC, refer to Figure 4. Although large standard deviations were encountered, the analysis enabled high cost threshold values to be defined and be used in the development of a set of rules that define excessive pavement maintenance.

65

97

104

100

73

79

99

114

5 Yr Avg Annual Expenditure ($) per Segment

20,000 15,000

Median Mean 90%ile

10,000 5,000 0 0-100 100-500 500-1000 1000-3000 3000-5000

1st Test Values 98%ile 2nd Test Values

AADT Category

FIGURE 4 Maintenance Cost Analysis Summary Based on this analysis and engineering judgement, the following two rules were developed as an indicator of poor pavement maintenance cost performance : Rule 1. Any 3 years out 5 that have a pavement maintenance cost greater than or equal to the 1st Test Value for its associated AADT range. Rule 2. Any 2 years out of 5 have a pavement maintenance cost greater than or equal to the 2nd Test Value for its associated AADT range. The current rules indicate that a very reasonable 3.1% of the road network has an excessive pavement maintenance cost regime, and should be considered a Poor performing pavement. It was concluded that the developed method, in principle, should be included in the overall rating of pavements in addition to a roughness progression rating, however, the actual rule values should be the subject of ongoing research and consideration by engineering practitioners. 7. STUDY OF ROUGHNESS PROGRESSION Some of the important conclusions from an initial study of the roughness progression of many 1km pavement segments included: many pavements displayed a consistent increasing trend over time; a large proportion of pavements tended to exhibit a linear roughness progression over time; isolated points with extreme variance from the surrounding points were most likely from data collection error; obvious unrecorded treatments to pavements - displaying a logical decrease (or step) in roughness data; some records have extreme data variance and will need to be handled in a manner such that the research is considered reliable; and

even the most consistent roughness data displayed a slight degree of fluctuation from year to year.

It was noted that the linearised trends of roughness against treatment (pavement) age were similar to that experienced by other international studies (6, 8). They were similar, in that, a wide variety of roughness progression rates were experienced for similar pavement types of similar ages. This highlights that pavement behaviour and the prediction of pavement roughness is not only complex, but is also unique to any one homogeneous pavement segment. This initial work led to the development of a methodology that calculated a Linear Roughness Progression Rate (LRPR) for each pavement segment. In essence, the methodology consists of two methods of fitting a line through time-series roughness data, rating the accuracy of each line, and making a decision about which line should be adopted to represent the roughness progression of the pavement segment. This concept for handling time-series roughness data is summarised in the flowchart shown in Figure 5.

Line Fitting

Line Fitting

Method 1
(All Points) Feed Lines into Matrix

Method 2
(Filtered Points)

Decision Matrix Consider : Maintenance Cost Analysis Line Accuracy Several Facts about the Roughness Data
Pass Test : YES Pass Test : NO
Is Default Roughness Progression Method OK?

ADOPTED Roughness Progression

No Roughness Progression Result Available

FIGURE 5 Roughness Progression Calculation Methodology The first method plotted a line through all points with a treatment age greater than zero (All Points LRPR), and the second being plotted through the data points that have been passed through an Engineering Filter which eliminated potential outliers and logical problems with the data (Filtered LRPR). An expert system approach, using a combination

of statistics and engineering judgement, was applied to the regression lines to select the most accurate and reliable representation. The LRPR chosen to represent the pavement segment is termed the Adopted LRPR. An example is shown in Figure 6.

FIGURE 6 Example of LRPR Determination

In the event that neither LRPR is considered an accurate representation, a Default LRPR is assigned to the pavement and is calculated by taking the most recent roughness value and subtracting an assumed start of life roughness value and dividing the result by the treatment age. The purpose of the Default LRPR calculation is to ensure that a complete road network representation of a roughness progression rate is achieved. This is important for practicing engineers working in road asset management and is discussed further in section 9. However, for the purposes of research and study of relationships between LRPR and independent variables, the Default LRPR was disregarded. The results of this process for the 15,802 individual 1km pavement segments analysed, indicated that 84% of pavement segments were able to be represented by a line fitted through historical roughness data. 12% of pavement segments were unable to be represented by a reasonable LRPR, and reverted to the Default LRPR calculation. These results are summarised in Figure 7.

LRPR Study Data : 15,802 km Length of Road


Default (Treatment Age) LRPR 1941 (12% ) No Calc 51
(0.3%)

Poor Mtce Pavements 616 (4% )

7031 (45% ) 6163 (39% ) Engineering Filter Regression LRPR All Points - Regression LRPR

FIGURE 7 Roughness Progression (LRPR) Calculations It was concluded that pavement roughness progression tended to be linear, and that a sound procedure for calculating the LRPR had been developed. It is suspected that the level of pavement surfacing maintenance (routine and programmed/periodic) required to hold the road system at an acceptable level of service and safety, maintains the rate of roughness progression in this linear zone. 8. ROUGHNESS PROGRESSION (LRPR) WITH INDEPENDENT VARIABLES A review of the literature identified several key independent parameters that have been used in previous roughness progression research. These include pavement age, traffic loading, loading by heavy vehicle type, climate zones, rainfall zone, temperature zone, maintenance costs, subgrade type, seal width, pavement structure and structural number. With the exception of pavement structure and structural number, linear regression was applied between 14 parameters and LRPR, each resulting in very low R2 values. Given the considerable spread in the data between LRPR and each independent variable, it was concluded that LRPR (or roughness progression) could not be represented as a twodimensional function. Several data mining techniques including Neural Networks, Rule Induction (or Decision Tree Analysis), Memory Based Reasoning, and Logistic Regression were used to determine whether any combination of the independent variables could be used to predict roughness progression. All methods failed to provide any useful predictive model. The data indicates that each pavement segment is unique in its roughness progression, and there is no one formula, or set of parameters, that can accurately predict roughness progression across a population. This questions the reliability and suitability of all formula based roughness prediction methods previously proposed in the literature.

9. ROAD NETWORK PROFILE Section 7 outlined the process used to characterise a pavement segments roughness progression by way of a Linear Roughness Progression Rate (LRPR). The LRPRs for the 15,135 km of pavement segments able to be calculated are displayed as a frequency distribution in Figure 8. Adding those pavements with Poor performance defined by excessive pavement maintenance cost (refer to Section 6) to the LRPR profile, provides a method of identifying the overall performance of pavements within the network. The graph showing both the cumulative and probability density functions is termed the Network Profile.

Probability Density Function


4000 3655 3329

Cumulative Density Function


0 8 10 15 25 23 2456 Poor Mtce Pavs No Meaningful Default LRPR Default LRPR 50 Filtered Points Regression LRPR 1353 All Points Regression LRPR Network Profile Percent >= LRPR (2nd Y axis) Percent >= LRPR Value 6 5 5 4 4

3000

2000 1446

1000 Frequency 412 51 0


n/a

765 719 417 616 262 132 76 47 67

75

100
>=-3,<-2 >=-2,<-1 >=-1,<0 >=0,<1 >=1,<2 >=2,<3 >=3,<4 >=4,<5 >=5,<6 >=6,<7 >=7,<8 >=8,<9 >=9,<10 >=10 Poor Mtce

Good 60%

Fair 25% LRPR Category (Counts/km/yr)

Poor 15%

FIGURE 8 LRPR Network Profile Figure 8 also displays a descriptive pavement performance rating in terms of Good/Fair/Poor. These ratings were derived from analysis that rated the ability of a pavement to exceed/achieve/under achieve an expected design life, respectively. The Network Profile indicates that 60% of pavements deteriorate at a slow rate, of between 3 counts/km/year and 2 counts/km/year, and should remain in service well in excess of the design life of 20 years. It is worth reiterating that a negative rate of roughness progression, which suggests a slow rate of deterioration, is most likely due to inherent system error in the annual measurement of roughness. It is not suggested that these pavements drastically improve in roughness over time, rather, some pavements appear to exhibit stagnant roughness. Of the network examined, 25% of pavements performed in the Fair performance range. 15% of pavements are in the Poor performance range, with 11% exhibiting roughness progression rates that are likely to see pavements exceed their roughness intervention

level in less than 17 years, with a majority of these exceeding their functional level in less than 10 years. The remaining 4% of pavements required excessive pavement maintenance funds to remain serviceable. It could be reasonably assumed that these pavements displayed the worst signs of distress because consistent and immediate attention is being applied. It was concluded that the frequency distribution and cumulative distribution of LRPRs and High Pavement Maintenance Cost pavements provides a useful measure of road network performance. Such a performance measure is commonly used to evaluate the acceptability of the current pavement design and construction delivery system and to determine broad scale funding scenarios based on the expected average life of pavements. 10. SUMMARY NETWORK PROFILE WITH INDEPENDENT VARIABLES With the regression between LRPR and independent variables not able to produce any useful insight into the way in which pavements perform, another approach was developed in a bid to better understand roughness progression and pavement performance. Using the Good/Fair/Poor pavement performance rating outlined Section 9, a Summary Road Network Profile was defined. This profile was then applied to engineering based categories of the independent variables listed in Section 8. This resulted in a very useful method of understanding the pavement performance of the road network. An example for traffic volume (AADT) is shown in Figure 9.
100% Segments in Category (Percent) 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 0-50 50-100 100250 250500 5001000 10003000 30005000
N=615 N=1759 N=3627 N=3671 N=2181 N=3073 N=904

24

20

12 26

15 24

16 25

13 19 68

14 18 67
Poor Fair

32

29 62 61 59
Good Trend Line

44

51

AADT Category (Perform ance based on LRPR + High Pav Mtce Cost)

FIGURE 9 Example of Summary Road Network Profile v Independent Variable

Figure 9 illustrates that there is a higher representation of Poor performing pavement segments in the lowest traffic volume category. It also indicates a general trend of an increasing number of Fair and a decreasing number of Good pavement segments as

AADT declines. It would appear that this outcome is a function of the engineering management strategy adopted to manage the large rural road network. Further, the trends indicate that all pavement categories (as defined by the independent variables), regardless of where they exist in the Queensland State Controlled Road Network, display a proportion of Poor performing pavements. This result provides a method of rating and understanding the road network performance. The current results suggest that the Pavement Design and Construction Delivery Environment used in the traditional high cost pavement design areas (high traffic loading, reactive soil, specification materials & high rainfall areas) are performing better than the practices currently being used in the traditional low cost Pavement Design and Delivery areas (low traffic loading, reactive soil, non-specification materials & low rainfall areas). Relatively, it appears that the risk profile of the various Pavement Design and Delivery Environments are probably of the correct order. However, the asset manager will need to determine : 1. Whether the proportion of Poor and Fair performing pavements is considered acceptable; 2. Whether the Design and Delivery profile needs to be adjusted to match a Desired Performance Profile; and 3. The definition and basis of a Desired Performance Profile. It is concluded that the study of Rated Pavement Performance (Good/Fair/Poor) does provide an understanding of the performance of the road system and can assist the management of the pavement design and delivery system. 11. PREDICTION OF ROUGHNESS PROGRESSION The underlying aim of the research works outlined in Sections 6 to 10 is to better understand roughness progression in order to improve the prediction capability of the roughness progression of a pavement segment. The literature highlights, that where annual roughness data is collected, a Site Specific prediction approach (or individual pavement segment approach) is more successful than existing mechanistic-empirical formulae or average family deterioration curve fitting (912). This research indicated that at least six data points were required to confidently project a fitted line through historical data and into a predicted future five-year timeframe.

12. INDUSTRY APPLICATION As previously outlined, the key goals of a pavement management system are to monitor pavements in order to provide performance information that can be applied to decision making processes in strategic planning, asset management, current and future network performance, pavement design (checking of current processes), and identification of future rehabilitation works (13). In accordance with these goals, it is believed that this research of roughness progression will assist in the following ways : Asset Management, the Identification of Future Rehabilitation Works, and Research Works will be aided by the use of the Poor rated pavements identified by high Linear Roughness Progression Rate (LRPR) and excessive pavement maintenance expenditure. Future Performance Prediction (5yr) is considered more reliable if a Site Specific predictive approach is used. Network Performance is defined via a Road Network Profile, based on LRPR and pavement maintenance costs, and displayed for any defined network. The Network Profile can be translated into a Summary Network Profile that uses a Good/Fair/Poor rating system based on the values of LRPR and excessive maintenance costs. Assessment of the Performance of the Pavement Design System is aided by studying the relationships between various independent parameters using the Summary Road Network Profile. A better understanding of roughness progression on a network of pavements will assist Strategic Planning, by ensuring a justifiable average roughness progression value is used in Network Analysis for the various categories and locations of pavements.

13. CONCLUSIONS This research has focussed on a holistic pavement management approach to the investigation and understanding of the roughness progression of pavements. Although the current analysis methodologies may be improved over time, it is anticipated that the essence of the research will form the basis of several key asset management functions with respect to pavement management in Queensland. The main conclusions are : The roughness progression of each individual pavement segment is unique. It is highly unlikely that there is one formula, or set of parameters, that can easily and accurately represent the roughness progression of all pavements across a population.

Historical Roughness Progression of a pavement segment can be defined by linear regression (LRPR) for a majority of the road network. Pavement Maintenance Costs can 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 provide a misrepresented LRPR. A combination of the Linear Roughness Progression Rate (LRPR) and excessive pavement maintenance expenditure can be used to define a Network Performance Profile, which aids in measuring the performance of a road network. Currently, road network performance is commonly defined only by a measure of the road networks Absolute Condition. The Network Profile work outlined in this research will add some measures of pavement performance to the current definition. By including a Current Network Profile, and a more robust method for predicting 5 Year Roughness values, engineers will now have a suite of information to enable historic, current, and estimated future road network condition, to assess the impact of previous and current management decisions. Study of network profiles for a variety of independent variables can assist the understanding of the in built risk of the current pavement design and delivery system. The learnings from a global analysis of condition may form a catalyst for the assessment of material quality, design methods, construction technology, and contract delivery, to ensure that as built pavements have the best chance of performing well. Prediction of pavement roughness, based on an extrapolation of the pavements LRPR, is a useful method of predicting roughness over a 5yr timeframe. A meaningful and useful Roughness Age relationship, for use in the prediction of future roughness, was not found.

14. ACKNOWLEDGMENTS The assistance of the Queensland Department of Main Roads in funding this research is gratefully acknowledged. The findings expressed in this paper are those of the author and do not necessarily reflect the views of QDMR. 15. REFERENCES

1.

2.

QDMR, A Brief Report on Pavement Deterioration Profiles (PDAT). 2000, Queensland Department of Main Roads. Asset Management Section : Road System Management Division: Brisbane, Australia. p. 15. QDMR, Queensland Department of Main Roads, Pavement Roughness Condition Reporting, Asset Management Services. 2000: Brisbane, Australia.

3. 4. 5. 6.

7.

8.

9.

10.

11.

12. 13.

Civil-Engineering-USA, Study proves Smoothness Extends Pavement Life, in Civil Engineering (USA). 1998. p. 13-14. RACQ, Unroadworthy Roads Survey 2001.. 2001, The Royal Automobile Club of Queensland Limited: Brisbane, Australia. Martin, T.C., A Review of existing pavement performance relationships. 1996, ARRB Transport Research: Vermont South, Victoria, Australia. p. 60. Paterson, W.D., The Highway Design and Maintenance Standard Series - Road Deterioration and Maintenance Effects. First ed. 1987: A World Bank Publication - The John Hopkins University Press - Baltimore. Hand, A.J., P.E. Sebaaly, and J.A. Epps, Development of Performance Models based on Department of Transportation Pavement Management System Data. Transportation Research Record, 1999. 1684(Pavement Design, Management and Performance): p. 215-222. Perera, R.W., C. Byrum, and S.D. Kohn, Investigation of Development of Pavement Roughness : Long Term Pavement Performance project. (LTPP). 1998, FHWA - Federal Highway Administration : US Department of Transportation. p. 285. Cheetham, A. Auto Adaptive Pavement Performance Prediction Methodology. in Fourth International Conference on Managing Pavements. 1998. Durban, South Africa. Hajek, J.J., et al. Performance Prediction for Pavement Management. in North American Pavement Management Conference, March 18 to 21 1985,. 1985. Toronto, Canada. Garcia-Diaz, A. and M. Riggins, Serviceability and Distress Methodology for Predicting Pavement Performance. Transportation Research Record, 1985. 997: p. 56-61. Ping, W.V. and H.E. Yunxia, Evaluation of Flexible Pavement Performance Life in Florida. 1998, Department of Civil Engineering: Tallahassee Florida. p. 231. Gordon, R. Monitoring of Pavement Performance. in South Western Division Queensland Department of Main Roads Symposium. 1984.

You might also like