Road Section Length Variability on Pavement ...

3 downloads 0 Views 1MB Size Report
Road Section Length Variability on Pavement Management Decision Making for Ontario. 1. Highway Systems. 2. 3. Gulfam-E- Jannat, M.A.Sc. 4. PhD Candidate.
1 2

Road Section Length Variability on Pavement Management Decision Making for Ontario Highway Systems

3 4

Gulfam-E- Jannat, M.A.Sc

5

PhD Candidate

6

Department of Civil and Environmental Engineering

7

University of Waterloo

8

200 University Ave. W., Waterloo, ON, Canada, N2L 3G1

9

Email:[email protected]

10 11

Theuns F.P Henning, PhD, MIPENZ

12

Director, Transportation Research Centre (TRC)

13

Department of Civil and Environmental Engineering

14

The University of Auckland, Auckland 1142, New Zealand

15

Ph: 64 9 923 8181

16

Email: [email protected]

17 18

Cheng Zhang, MASc.

19

Department of Civil & Environmental Engineering

20

University of Waterloo

21

200 University Ave. W., Waterloo, ON, Canada, N2L 3G1

22

Ph: 519-721-5866

23

Email: [email protected]

24 25 26 27 28 29 30

2

Jannat et al. 2015 1

Susan L. Tighe, PhD, PEng

2

Professor

3

Department of Civil and Environmental Engineering

4

University of Waterloo

5

200 University Ave. W., Waterloo, ON, Canada, N2L 3G1

6

Phone:1- 519 888-4567 ext. 33152

7

Email: [email protected]

8 9

Li Ningyuan, P.Eng., Ph.D.

10

Senior Pavement Management Engineer

11

Ministry of Transportation of Ontario

12

1201 Wilson Avenue, Downsview

13

Ontario Canada M3M 1J8

14

Ph: 1-416-235-3518

15

Email: [email protected]

16 17 18 19 20 21 22 23 24 25 26 27

Word Count: 4,742 words Text + 6 Tables x 250 words+ 5 Figures x 250 words= 7,492 words Revised Paper Submission Date: November 15, 2015

28 29 30 31 32 33

Submitted for presentation at the 95th Annual Meeting of Transportation Research Board, January 10-14, 2016, Washington D. C. on the topic AFD10- Pavement Management Systems

3

Jannat et al. 2015 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19

ABSTRACT In a pavement management system (PMS) the performance evaluation indices and their prediction methods are important aspects for assessing the overall pavement condition. So, an accurate location reference system is necessary for managing pavement evaluations and maintenance. In this regard, the length of the pavement section selected for evaluation may also have significant impact on the assessment irrespective of the type of performance indices. This study investigates the variability in pavement performance evaluation and maintenance decisions due to change in pavement section lengths. It considers rut depth, Pavement Condition Index (PCI) and International Roughness Index (IRI) as performance indices. Data from twenty seven road segments of Ontario with a total length of 172.5 km are selected for empirical investigation. The distributions these indices are compared by grouping various segment lengths ranging from 50m, 500m, 1000m and 10000 m. The variations of performance assessment due to changing section length are investigated based on their impacts on maintenance decisions. A Monte-Carlo simulation is carried out by varying section lengths to estimate probabilities of the necessity of maintenance works. Results of such empirical investigation reveal that most of the longer sections are evaluated with low rut depth and the shorter sections are evaluated with higher rut depth. Monte-Carlo simulation also reveals that 50m sections have higher probability of maintenance requirement than that for the 500m section. Although the results are related to the Ontario’s highway system, these can also be applied elsewhere with similar conditions.

20 21 22

Key Words: Key Performance Index (KPI), PCI, IRI, rut depth, and distribution

23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43

1. INTRODUCTION Evaluation of pavement performance is sophisticated due to the complexity of factors and associated interactions which influence pavement performance. An effective PMS is possible only when pavement performance is evaluated accurately and efficiently. The PMS analysis results vary significantly with the length of each pavement section and the selected performance indicators used for condition assessment. The PMS was first established for Ontario highways in 1980’s. At that time the computers were not cost effective and the pavement rehabilitation and maintenance were based on qualitative engineering experience (1). The Ministry of Transportation Ontario (MTO) has a long history of using PMS and is currently using a second generation PMS, which is known as PMS-2. Over the past 30 years, several key performance indices (KPIs) are used in the PMS to make rational maintenance and rehabilitation (M&R) decisions. These include PCI, Distress Manifestation Index (DMI), IRI, and Riding Comfort Index (RCI) etc. (2, 3). Pavement sections for assessment are pre-defined for all individual provincial highways managed by the MTO and summary of the pavement condition data and evaluation results are reported on the basis of such pre-defined pavement sections. Typical lengths of such sections range from 500 m to 50,000 m. Although the long sections are homogenously defined or classified by pavement structure and geographic locations, those do not represent homogeneous performance conditions, including rut depth, IRI, and PCI. For this reason, this study focuses on investigating the influences

4

Jannat et al. 2015 1 2 3

of section lengths on the performance evaluation and subsequent maintenance management and investment decisions generated by the PMS-2. The study investigates the impact of section lengths on both performance monitoring and maintenance decisions as well.

4 5 6 7 8 9

2. LITERATURE REVIEW Most of the studies on pavement performance indices available in literature are mainly on the existing practice of performance indices used in PMS by different agencies. In Ontario highway systems, different KPIs are used for performance monitoring. Table 1 summarizes these indicators (2).

10

TABLE 1 Performance Measures and KPIs Used in Ontario PMS Performance Measure

Assessment of overall pavement condition, structural strength and functional serviceability of both network and individual pavement sections

KPI Technical Measures PCI

Overall pavement surface condition for individual pavement sections and network

DMI

Evaluation of pavement riding quality in terms of roughness or smoothness

IRI

Evaluation of pavement riding quality in terms of user comfortableness Pavement surface skid resistance

Riding Comfort Rating (RCR) Skid Number

Transverse profiles and rutting measurement

Rut depth in mm measured in both left and right wheel paths

Assessment of pavement structural strength or service life

Structural Adequacy Index (SAI) or deflection value measured by FWD equipment

Maintenance costs and service life, traffic accidents and costs, travel time, travel cost and travel time reliability, etc.

Economic measures Highway agency cost, social and economic impacts, Road user costs

Specific Use

PCI is currently used by regions to generate an annual pavement maintenance program and investment planning strategies DMI is used to support PCI and is favoured by regions that have lower class roads Pavement roughness data is collected by use of high-speed inertial profilers and are used in M&R decision RCR was collected prior to IRI data Measured at project level on a request basis and used in project level Rutting is measured at network level using high-speed equipment and used as both distress and KPI in PMS Currently used at project level

For inclusion in the PMS-2 analysis function

11 12 13 14 15 16

In the USA and Europe, different PMS are used. Van Der Lei et al. (4) presented a comparison of performance management practices in the USA and Europe. They found that KPIs were normally developed and chosen once the goals of the corresponding agency was determined. In a similar study, Lea et al. (5) studied the initial configuration of California's PMS focusing on aggregation of pavement data from their automated surveying.

5

Jannat et al. 2015 1 2 3 4 5 6 7 8

Henning et al. investigated the effectiveness of practicing KPIs. In this study, Rutting Index (RI) is developed with the objective to effectively quantify the structural performance of pavement in New Zealand (6). For Ontario highways, effect of using alternative KPIs on pavement condition assessment is investigated (7). Molenaar and Navarro, conducted case studies to determine how contractor performance is managed during construction. This study showed that although performance measures were used in the majority of the case studies, formal KPIs were not (8). For small communities, PMS is dissimilar from that in highways which considers different modules (9).

9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

Pavement performance index model is also found to be developed from the combination of both experimental data and field data (10). American Association of State Highways Officials (AASHO) road test data and Minnesota Road Research Project (MnRoad) are used in this study. ‘Linear mixed effects model’ is developed to predict future conditions of pavement. This model is developed by a weighted combination of the average deterioration trend and the past conditions of pavement (11). Non-linear model is also developed for KPIs for realistic prediction of pavement condition (12). In this study, KPI model is developed considering traffic characteristics, pavement structural properties, and environmental conditions. Moreover, effect of KPI model accuracy is also investigated on optimal design and lifecycle costs by using regression model, probabilistic model (13). In a similar study, effectiveness of the estimation of rutting models is also investigated (14). In this study, combined estimation is found to identify parameters by combining the information from two data sources, the AASHO and the WesTrack road tests. Furthermore, effect of different statistical assumptions and estimation techniques of KPI models on predictive capabilities are also explored by using AASHO Road Test data (15). An approach of applying mathematical techniques of fuzzy sets are found to categorize the subjective rating of pavement distress severity and extent (16).

25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45

Although different KPIs and methods are used for evaluation of pavement, accurate location referencing is often overlooked. However, this is a very important aspect for pavement evaluation and management. Accurate location referencing enables the sectioning of the network in to homogenous pavement sections. The procedure of forming homogenous sections are known as sectioning or segmentation (17). Different agencies use different road sectioning methods. Generally, three sectioning methods are used commonly in PMS. These are: (a) fixed length sectioning method, (b) dynamic sectioning method, and (c) static sectioning method (18). Generally, fixed length sectioning methods divide each road length into fixed lengths of typically between 100m to 500m. Although this method is considered as suitable one as it overcomes most of the statistical limitations of other methods, however, it has complexity in determining appropriate lengths of treatments. Dynamic sectioning method uses road condition data being stored at a relatively short interval and amalgamates sections into variable lengths while ensuring the condition and inventory data falls within certain homogeneity rules. In this method, sections are analysed in the homogeneous lengths. This method has limitation that the sectioning changes every year pending the outcome of condition surveys. Static sectioning method uses a level of dynamic sections with the difference being that once homogenous sections are being amalgamated it stays constant for a while. This is a popular method

6

Jannat et al. 2015 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19

used by road agencies. This method has limitation that the sectioning is only valid for a short time period and if agencies don’t review the sections on a regular basis it may result in some illogical outcomes of maintenance decisions.

20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46

4. EXISTING KPIs USED IN ONTARIO HIGHWAYS As discussed in previous section, different KPIs are used with various purpose of pavement evaluation for Ontario highway systems. PCI is used generally as an overall index in pavement M&R decision tree. The decision trees use a series of logical tests to capture the decision making process in selecting appropriate M&R treatments based on the status of KPIs (19).

So, the review of existing literature clarifies that KPIs and sectioning methods are used by different agencies in their PMS; however, there is a gap in understanding on the impact of section length on the performance indices. So, this paper focuses on investigation of the impact of section length on pavement performance indices.

3. SCOPE AND OBJECTIVE OF STUDY This study will analyse the impact of section length on performance evaluation and maintenance decision of pavement. Implication of section length is investigated in two steps, firstly, on performance monitoring and secondly, on maintenance decisions. It uses data from the MTO PMS2 to evaluate the impacts of varying length of sections on overall outcome of the analysis. For empirical investigations, pavement performance data of Ontario’s central zone highway network are used. For this analysis, four types of section lengths are considered. These are: 50m, 500m, 1,000m, and 10,000m.

The DMI which is a component of the PCI, is also an important index. MTO has classified the distresses into fifteen categories and assigned weight to them to calculate DMI (20). The PCI is composed of two sub-indices respectively representing DMI and IRI. In Ontario, the formula used to calculate PCI from DMI is defined as follows (20): For Asphalt Concrete (AC) Pavement, 𝑃𝐶𝐼 = 𝑀𝑎𝑥𝑖𝑚𝑢𝑚 𝑣𝑎𝑙𝑢𝑒 𝑜𝑓( 0, 𝑜𝑟 𝑚𝑖𝑛𝑖𝑚𝑢𝑚 𝑜𝑓 (100, 13.75 + 9 × 𝐷𝑀𝐼 − 7.5 × 𝐼𝑅𝐼))

(1)

Where, PCI= Pavement Condition index in score DMI= Distress Manifestation Index in score IRI= International Roughness Index in mm/ m Recently, Ontario highway systems are moving from manual rating system to an automated survey regime. The formula to calculate new PCI is being changed for automated survey data which is as follows:

7

Jannat et al. 2015 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45

𝑁𝑒𝑤 𝑃𝐶𝐼 = (0.7 × 𝐼𝑅𝐼 𝑆𝑐𝑎𝑙𝑒𝑑) + (0.2 × 𝐷𝑀𝐼) + (0.1 × 𝑅𝑢𝑡 𝑆𝑐𝑎𝑙𝑒𝑑)

(2)

Where, 𝐼𝑅𝐼 𝑆𝑐𝑎𝑙𝑒𝑑 = [100 × {1 − 𝐼𝑅𝐼 𝑖𝑛 𝑚𝑚 𝑝𝑒𝑟 𝑚 /5}] 𝑅𝑢𝑡 𝑆𝑐𝑎𝑙𝑒𝑑 = [100 × {1 − (𝐴𝑣𝑒𝑟𝑎𝑔𝑒 𝑅𝑢𝑡 𝐷𝑒𝑝𝑡ℎ 𝑖𝑛 𝑚𝑚 /30 )}] 5. ROAD PERFORMANCE DATA AND SELECTED ROAD SECTIONS As discussed in previous section that the pavement sections for assessment are pre-defined for the Ontario highways. The pavement condition data and evaluation results are reported on the basis of such pre-defined pavement sections. A detailed rut depth and performance data is collected from the Ontario’s Central region for different section lengths for the year of 2013. These section lengths are 50m, 500m, 1,000m and 10,000m. To evaluate the performance of the certain pavement segment, the road sections are surveyed separately for 50m, 500m, 1,000m and 10,000m respectively. Although these sections are overlapping in one segment, they are surveyed unconnectedly so that their performances can be compared. From available performance records of highways, only twenty seven road segments from two highways (Highway 401 Eastbound and Westbound) are found that contain these four types of section lengths that are listed along with the number of total sections in Table 2. Road segments with maximum available length are selected where all four types of section lengths are available for that certain segment. Finally, an experimental design is selected which contains twenty seven road segments (covering 172.5 km length from the Ontario Central zone) which consists of three thousand four hundred fifty one 50m sections, three hundred forty six 500m sections, one hundred fifty two 1,000m sections, and fifteen 10,000m sections.

8

Jannat et al. 2015 1

TABLE 2 Selected Road Segments with Four Types of Sections Name of Highway

Direction/ Bound

Chainage of Road Segment , in m Begin

End

Total Distance in m

Section ID

Number of Total Sections

LHRS1

50m

500 m

1,000 m

10,000 m

401

E2

507000

518500

11500

111390

47700

230

23

11

1

401

E

497500

506000

8500

111370

47689

170

17

8

1

401

E

378000

384000

6000

110950

47558

121

12

6

0

401

E

385000

392000

7000

110970

47560

140

14

7

0

401

E

392000

403500

11500

110990

47570

229

23

10

1

401

W3

376500

383000

6500

110960

47558

130

13

6

0

401

W

383500

391000

7500

110980

47560

151

16

8

1

401

W

392000

402500

10500

111000

47570

210

21

10

1

401

W

403000

412500

9500

111020

47580

190

19

9

1

401

W

413000

421500

8500

111040

47594

170

17

8

1

401

W

422000

424500

2500

111060

47600

50

5

2

1

401 401 401

W W W

425000 428500 434500

429000 434000 441500

4000 5500 7000

111080 111100 111120

47603 47607 47612

80 110 140

8 11 14

4 5 7

1 1 1

401

W

442000

443500

1500

111140

47625

30

3

1

0

401

W

449500

458000

8500

111180

47631

170

17

3

0

401

W

458500

462500

4000

111200

47643

80

8

3

0

401

W

463000

468000

5000

111220

47652

100

10

5

1

401

W

468500

472000

3500

111240

47660

70

7

1

0

401

W

472500

474000

1500

111260

47665

30

3

2

0

401

W

474500

476500

2000

111280

47667

40

4

2

0

401

W

476500

482500

6000

111300

47672

120

12

5

0

401

W

483000

486500

3500

111320

47675

70

7

3

0

401

W

487000

495000

8000

111340

47680

160

16

4

1

401

W

495500

497500

2000

111360

47688

40

4

2

0

401

W

498000

507000

9000

111380

47689

180

18

9

1

401

w

507500

519500

12000

111400

47700

240

24

11

1

3,451

346

152

15

Total

2 3 4 5

LHRS – Linear Highway Referencing System E- East-bound 3 W-Westbound 1 2

172,500

9

Jannat et al. 2015 1 2 3 4 5 6 7 8 9 10 11 12

6. RESEARCH METHODOLOGY The objective of this study is aimed to investigate the variability in performance evaluation of pavement due to change in section length for different KPIs used in PMS. Although for Ontario highways, PCI is used generally as an overall index in the pavement M&R decision tree, nowadays rut depth is another major concern for Ontario’s major freeways too. Moreover, the PCI is calculated from the deterioration condition measured by the three components: IRI, rut depth, and DMI. In this study, these two types of distresses (rut depth and IRI) are considered for analysis. Although these days some agencies use rut depth and IRI not only as distresses but also as overall condition index. However, DMI is not taken into consideration in this study as it contains only 20% of weight in to PCI, and DMI consists of fifteen distresses along with cracking. Analyses of all these fifteen types of cracking and distresses might complicate the detailed data analysis at this stage.

13 14 15 16 17 18 19

This study is carried out based on the statistical analysis in to three steps. Rut depth is used in first step to illustrate the impact of section length on the understanding of condition distribution. In second step, impact on overall condition of road segment is investigated by comparing the PCI and IRI for these section lengths. In third step, impact on maintenance decision due to variation of performance evaluation offered by change in section length is analysed. Figure 1 presents steps in methodology which are followed in this study. Selection of Experimental Design Consisting of 27 Road Segments covering 172.5 km (Containing a total of 3,451 nos 50m Sections, 346 nos 500m Sections, 147 nos 1,000m Sections, and 15 nos 10,000 m Sections) from Ontario Central Zone Highway Network

Comparison of Performance of KPIs for 50 m, 500m , 1000 m and 10,000 m Section Comparison of PCI Comparison of Rut Depth Comparison of IRI Distribution • Comparison of New Vs Old • Comparison of PCI • Comparison of Probability Distribution Parameters Paper Plot • Comparison of Distribution Parameters • Comparison of Distribution Parameters Maintenance Requirement Comparison for 50 m and 500 m Section by Monte-Carlo Simulation

Assuming Distribution

Considering Rut Depth as random variable

Counting Maintennace Requirement when PCI