Modeling pavement performance based on data from the Swedish LTPP database : predicting cracking and rutting

(1)

Modeling Pavement Performance

Based on Data From the Swedish

LTPP Database

Predicting Cracking and Rutting

Licentiate Thesis in Highway Engineering

Markus Svensson

KTH, Royal Institute of Technology

School of Architecture and the Built Environment Department of Highway Engineering Division of Highway and Railway Engineering

Stockholm

markus.svensson@vti.se January 8, 2013

(2)

Abstract

The roads in our society are in a state of constant degradation. The reasons for this are many, and therefore constructed to have a certain lifetime before being reconstructed. To minimize the cost of maintaining the important transport road network high quality prediction models are needed. This report presents new mod-els for flexible pavement structures for initiation and propagation of fatigue cracks in the bound layers and rutting for the whole structure.

The models are based on observations from the Swedish Long Term Pavement Per-formance (LTPP) database. The intention is to use them for planning maintenance as part of a pavement management system (PMS). A statistical approach is used for the modeling, where both cracking and rutting are related to traffic data, climate conditions, and the subgrade characteristics as well as the pavement structure. Keywords: performance predictions, LTPP, rut, crack, statistical modeling.

(3)

Preface

This thesis is best viewed digitally. All readers are therefore encouraged to print this thesis only if necessary.

Thank you!

(4)

Acknowledgment

There are of course many people who have contributed to this project, and for this I am very thankful. But without the help from my fellow Ph.D. students this thesis would not exist. The project was financed by VTI and STA, ergo. . . a big "Thanks!" to the Swedish tax payers.

(5)

List of Publications

Modeling of performance for road structure rutting and cracking based on data from the Swedish LTPP database

Presented at the 8th International Conference on Managing Pavement, Santiago, Chile.

MODELING PERFORMANCE PREDICTION, BASED ON RUTTING AND CRACKING DATA

Presented at the 4th European pavement and asset management conference (EPAM), in Malmö, Sweden.

(6)

Acronyms and Abbreviations

AADT Annual Average Daily traffic..

AADTT Annual Average Daily Truck traffic..

AADT Ty Annual Average Daily Truck Traffic volume y years after the road was

opened for traffic.

AADTf Annual Average Daily Traffic.

AADToAnnual Average Daily Traffic at the time of opening the road .

AASHO American Association of State Highway and Transportation Officials . AASHTO American Association of State Highway and Transportation Officials. ABD Drained Asphalt Concrete.

ABS Stone mastic asphalt concrete. ABT Dense Asphalt conrete. ABb AC binder layer. AG Roadbase.

BG bitumen stabilizing gravel.

C-LTPP Canadian Long Term Pavement Performance. C-SHRP Canadian Strategic Highway Research Program.

Ci Crack index.

DBMS Database Management System. DBPs Deflection Basin Parameters. DSS Decision Support System.

EER Enhanced Entity Relationship Model. ER Entity Relationship Model.

ESALs Equivalent Single Axle Loads. FHWA Federal Highway Administration. FWD Falling Weight Deflectometer. GIS Geographic Information System. GIS Geographic Information Systems.

HABS Stone mastic Asphalt concrete AC with hard binder (pen 70 100)(SMA). HABS Stone mastic asphalt concrete with hard binder (pen 70 100) (SMA). HDM Highway Development and Management.

HDM-IV Highway Development and Management IV. HIPS Highway Investment Programming System. HMA Hot Mix Asphalt.

HVS Heavy Vehicle Simulator.

Heating/repaving a maintenance method. Heating the existing layer and placing a new.

(7)

MAB Asphalt concrete with soft binder (pen 180 220). MABT Dense Asphalt conrete with soft binder (pen 180 220). MAT Mean Annual Temperature.

MEPDG Mechanistic Empirical Pavement Design Guide. MMS Maintenance Management System.

MrResilient Modulus.

NVDB National Road Database.

N100Y Number of equivalent heavy vehicle axles per year.

N100Number of equivalent heavy vehicle axles.

N₁₀₀Number of equivalent standard vehicle axles.

PCC Portland Cement Concrete. PMS Pavement Management System. PSI Present Serviceability Index. PSI Present Serviceability Index. PV Previous Value.

RDBMS Relational Database Management System. RMS Road Management Systems.

RMSs Road Management Systems. RST Road Surface Tester.

Remixing recycling method.

SCI300Surface Curvature Index 300.

SHRP Strategic Highway Research Program. SQL Structured Query Language.

STA Swedish Transport Administration.

Stabinor Adhesive layer, material created by Skanska. TRB Transportation Research Board.

VTI The Swedish National Road and Transport Research Institute. WIM Weigh in Motion.

(8)

List of Symbols and Terminology

Ac- Alligator cracking

a- Parameter in the rut model, equation 2.6

b- Parameter in the rut model, equation 2.6

Ci - Crack index

Ciini - Crack index, for the crack initiation phase.

Cipropa - Crack index, for the crack propagation phase.

Lc- Longitudinal cracks

Mr- Resilient Modulus, an estimate of a materials modulus of elasticity, stress

di-vided by strain for speedily applied loads.

N₁₀₀Y - the average annual ESALs per lane.

NCri

100 - The total number of ESALs for specific section before crack initiation, in

2.3.

N₁₀₀Crp- The total number of ESALs for specific section before crack propagation, in

2.3.

N₁₀₀Cr - The total number of ESALs for specific section before cracking, in 2.3.

N₁₀₀Cr - the sum of the standard axle repetitions to crack initiations and for crack

propagation respectively up to failure.

Tc- Transversal cracks

ρX,Y = corr(X ,Y ) =cov(X ,Y )_σ_X_σ_Y =E((X −µ_σX_X)(Y −µ_σ_Y Y)) - Pearson product moment

(9)

List of Figures

1.1 The development of the number of registered vehicles in Sweden.

[56] . . . 1

2.1 Stress distribution in flexible and rigid pavement.[61] . . . 4

2.2 Composite Pavement Design. [9] . . . 4

2.3 The derivation of the N100axle, used in Sweden. [59] . . . 6

2.4 An illustration of the common types of cracks found in flexible pavements. [16] . . . 7

2.5 An illustration of the Cidivision initiation and propagation. [21, 78] 7 2.6 Illustration of common types of deformation in HMA.[16] . . . . 8

2.7 Surface defects. [16] . . . 9

2.8 Edge Defects. [16] . . . 9

2.9 PSI is a qualitative index, from the users perspective, that separates structural and functional distresses. When maintenance or rehabil-itation is performed the PSI value is often decreased compared to the start value. [73] . . . 11

2.10 An illustration of the concept of RMS and PMS. . . 13

2.11 An example from a PMS with DSS and GIS. [54] . . . 14

2.12 The conception of the life-cycle analysis in HDM-4 [63]. . . 15

2.13 Creep characteristic of bituminous mixtures. [57] . . . 19

2.14 The graphs show deformations of the pavement due to the cyclic load/unload. [16] . . . 19

2.15 Illustration of the common axle types and tire configurations. [48] 20 2.16 The graphs show axle loads from different types of axles.[49] . . . 20

2.17 The climate zones classified by the number of days when the mean temperature is below 0◦C in Sweden. [67] . . . 21

2.18 An illustration of performance of pavement structure factors. [53] 24 2.19 The RST used at VTI, and an illustration of the rut sampling and how the lasers are placed. . . 25

2.20 A description of the FWD measurements. [60, 11] . . . 26

2.21 An illustration of the wire surface principle in rut depth sampling. [20] . . . 27

2.22 Illustration of the variance in the rut data. The data plotted comes from object D-RV53-2. [21] . . . 28

(10)

2.23 Example of size and development of a crack between three dif-ferent inspections of the same road that is carried out at difdif-ferent times. This object is 100 [m], and divided into smaller 10 [m]

sections.[77] . . . 29

2.24 Civariation in an object. [21] . . . 30

3.1 The red dots in the map mark the geographical locations of the

objects. . . 32

3.2 The geographical distributions of road objects in the Swedish LTPP

database. The red dots are active objects, and the orange objects

retired from data monitoring. . . 33

3.3 Traffic estimations for object H-RV40-2. The black dots are the

traffic measured and the red lines are the estimations used in the

project. . . 35

3.4 Traffic estimations for object T-2051-1. The black dots are the

project. . . 36

3.5 Traffic estimations for object F-RV31-1. The black dots are the

project. . . 37

3.6 Traffic estimations for objects H-RV34-1. The black dots are the

project. . . 38

3.7 Traffic estimations for objects D-RV53-2. The black dots are the

project. . . 38

3.8 Traffic estimations for objects C-292-1. The black dots are the

project. . . 39

3.9 Scatter plots for the Ci factors SCI300, Mrand N100100,

represent-ing subgrade, structure and traffic. . . 40

3.10 Scatter plots for the Ci factors precipitation and temperature,

rep-resenting climate and age. . . 41

3.11 Scatter plots for the Cifactors previous value (PV). . . 41

3.12 Scatter plots for the rut factors SCI300 Mr and N100, representing

subgrade, structure and traffic. . . 42

3.13 Scatter plots for the rut factors precipitation, temperature and age. 43

3.14 Scatter plots for the rut factor PV and the interaction effect,

be-tween standard axles and precipitation. . . 44

3.15 An illustration of how the factors in the models are expressed as

(11)

4.1 The plots show the results for the objects D-RV53-2 and T-205-1. The black dots are the measured data, and the red line is estimated.

The Ciis measured as a function of time. . . 50

4.2 The plots show the results for the objects F-RV31-1, RV34-1, H-RV40-2 and C-292-1. The black dots are the measured data, and the red line is estimated. The Ciis measured as a function of time. 51 4.3 The plot shows the results for the objects D-RV532-2, H-RV34-1. The black dots are the measured data, and the red line is estimated. The Ciis measured as a function of time. . . 53

4.4 The plot shows the results for the objects T-205-1, F-RV31-1 and H-RV40-2. The black dots are the measured data, and the red line is estimated. The rut is measured as a function of time. . . 54

4.5 The Cimodels used on object G-RV23-1. The small black dots are the measured values, the dots on the red line are the predicted values. 55 4.6 The Cimodels used on object W-RV80-1. The small black dots are the measured values, the dots on the red line are the predicted values. 56 4.7 The rut model used on object W-RV80-1. The small black dots are the measured values, the dots on the red line are the predicted values. 56 4.8 The Ci models used on object Z-E45-4. The small black dots are the measured values, the dots on the red line are the predicted values. 57 4.9 The rut model used on object Z-E45-4. . . 57

A.1 An illustration of the concept of random variables. . . 65

A.2 Image of an r.v. variable. . . 66

A.3 A box plot displays the differences between populations without any assumptions of the statistical distribution, i.e. a non-parametric method. The box plot helps identify outliers by indicating the level of dispersion and skewness in the data.[50] . . . 67

A.4 An illustration of a general controlled experiment. . . 69

A.5 A "typical" Ci, data from the object D-RV-2. . . 70

A.6 "Typical" rut values, data from the object H-RV34-1. . . 70

B.1 Relational database terminology. . . 73

B.2 Illustration of the normal forms. . . 73

C.1 The Ciand rut data presented in box plots. . . 83

D.1 The red dots mark the geographical location of the validation objects. 84 D.2 The Ciand rut data presented in box plots. . . 87

(12)

List of Tables

2.1 The values for PSI. . . 11

2.2 The periods length in days during one year, [67]. . . 22

2.3 Temperature, [◦C], in the bitumen bounded layers, [67]. . . 22

2.4 Deflection Basin Indexes. [11] . . . 26

3.1 General object information. . . 31

3.2 General construction and maintenance information. . . 31

3.3 Material information continued. . . 32

3.4 Material information continued. . . 32

3.5 Civariable structures. . . 45

3.6 Civariable structures. . . 45

3.7 Rut variable structures. . . 45

D.1 General info of the validation road objects. . . 84

D.2 Material info for the validation objects. . . 84

D.3 Civalidation factor data. . . 85

D.4 Civalidation factor data. . . 85

(13)

INTRODUCTION

The value of a society’s infrastructure is difficult to measure, but the benefits of a functioning road network are easy to understand. Our roads are in a state of con-stant deterioration due to factors such as climate and heavy traffic loads. Therefore they are constructed to have a certain lifetime before being maintained or rehabil-itated. The transport infrastructure in Sweden costs every year several billions, in 2010 the government decided to spend more than 480 billion SEK on infrastructure development and maintenance over the time period 2010 to 2021. To minimize this cost, effective models for predicting performance are needed. A Pavement Man-agement System (PMS) is therefore used for managing the road network. The PMS requires condition data of the network, which allows an overview to enhance the sustainability of transportation infrastructure systems.

Road deterioration is a complex process involving numerous variables and their interaction. Road performance is affected by external loading, and is influenced by factors such as material properties, the environment, and construction practices. In the past, pavement design procedures have relied mainly on empirical relationships based on long term experience, and field tests. An estimated length of 420 000 km of asphalt road exists in Sweden out of which 100 000 km is owned by the state, municipalities own 40 000 km and there are 280 000 km privately owned. The number of vehicles on our road network has historically increased, and this development is shown in Figure 1.1.

(18)

1.1. AIMS CHAPTER 1. INTRODUCTION A Pavement Managing System (PMS) is a tool used to plan and perform mainte-nance on the road network in a cost efficient manner. Usually a PMS consists of five main parts:

◦ Pavement condition studies.

◦ Database containing all relevant

pavement condition information.

◦ Analysis plan.

◦ Decision strategy.

◦ Implementation procedures.

The analysis and decision making process require high quality distress prediction models; and two of the most common distress types are: rutting and cracking. [37, 33, 46, 25, 3, 56, 2]

1.1 Aims

Two of the most common distresses for flexible pavements are cracks and rut. This project has developed performance prediction models for flexible pavement struc-ture cracks in the bound layers and rutting for the whole pavement strucstruc-ture. The performance prediction models used today are obsolete, due to reasons such as changes in the climate, new methods and technical advancements in the construc-tion and data gathering process. The aim for this project is to develop predicconstruc-tion models for initiation and propagation of cracks and rutting, that deteriorates roads with a gravel/bitumen superstructure. The intention is to use them for planning of maintenance activities. The data used in this project have mainly been taken from the Swedish Long Term Pavement Performance, (LTPP). The traffic data were complemented with data from Swedish Road Administration.

This project investigated which variables are key factors in the deterioration pro-cess, such as type of construction, volume of heavy traffic, climate, subgrade, and then use this information to develop new deterioration models, based on data from the Swedish LTPP database for flexible pavements. The intention, for the models, is to use them in planning maintenance as a part of the PMS. This will enable the Swedish Transport Administration (STA) to make reliable distress predictions in their pursuit to prudently manage the road network. Models for this already exist, but they can be further optimized, or replaced with more accurate models.

Specific aims:

◦ Investigate which variables are a key factors in the deterioration process,

such as type of structure, volume of heavy traffic, climate and subgrade.

◦ Develop new deterioration models, based on LTPP data, for flexible

pave-ments.

This requires an extensive literature study, followed by extracting and analyzing data from the Swedish LTPP program.

(19)

Chapter 2

LITERATURE REVIEW

This chapter contains a review of some of the current knowledge in this area of science.

2.1 Pavement Types

The oldest findings of engineered roads are dated from about 4000 BC. Roads paved with stone were found at Ur, in modern day Iraq, and old timber roads pre-served in a swamp have been found in Glastonbury, England. The first usage of asphalt in a road occurred in 1824 in the Champs-Élysées in Paris. The more mod-ern type of asphalt was created by Edward de Smedt at Columbia University in New York City, and used in Battery Park in New York City in 1872 [4].

Today a highway pavement is a structure consisting of superimposed layers of processed materials above the natural soil subgrade. The purpose of a pavement structure is to provide a surface with:

◦ adequate skid resistance.

◦ favorable light reflecting characteristics.

◦ acceptable riding quality.

◦ low noise pollution.

◦ long design life with low construction and maintenance cost.

This is done by distributing the applied vehicle loads to the subgrade, and providing waterproofing. The body of the road is divided into:

◦ Surface layer.

◦ Base course, usually unbounded.

◦ Subgrade, reinforcement to the base course and the "naturally" occurring

soil. Normally the layer with the largest thickness.

Pavement structures can generally be categorized into two groups, flexible or rigid. A third group; composite pavement, is a mixture of both. This type is more expen-sive and seldom used [27]. They are all presented in the following text.

2.1.1 Flexible Pavement Structure

The most common type of pavement structure in Sweden is flexible pavements, i.e. flexible due to the fact that the structure bends under loading. The surface is covered with a hard waterproof bituminous material. The top layers are designed

(20)

2.1. PAVEMENT TYPES CHAPTER 2. LITERATURE REVIEW to withstand the stresses from the vehicles and to distribute the loads to the lower layers over a larger area. This design method modifies the stress distribution area in such a way that less expensive weaker materials can be used [27]. This is illustrated in Figure 2.1.

Figure 2.1: Stress distribution in flexible and rigid pavement.[61] [27]

2.1.2 Rigid Pavement Structure

Rigid Pavements are composed of a Portland Cement Concrete (PCC) surface course. This gives a higher elastic modulus, a stiffer construction, suitable for areas with high traffic loads and/or heavy vehicles i.e. airports, bus stops. This type of construction tends to distribute the load over a wider area of subgrade due to its rigidity. (See Figure 2.1) [27].

2.1.3 Composite Pavement Structure

Composite Pavements are a combination of both flexible and rigid design. This is the most expensive type of structure and seldom used in Sweden. There are two types of composite pavements, one with a layer of AC over PCC or a layer of PCC over PCC, this is further illustrated by Figure 2.2 [27].

(a) AC over PCC (b) PCC over PCC

Figure 2.2: Composite Pavement Design. [9] [80, 9]

As most roads in Sweden are flexible pavements this project concentrates on ana-lyzing and predicting performance for flexible pavement structures.

(21)

2.2. TRAFFIC CHAPTER 2. LITERATURE REVIEW

2.2 Traffic

One of the functions of a pavement is to protect the subgrade by distributing the load from the traffic through the upper layers. The loads magnitudes and their frequencies vary with time and place, which gives a highly stochastic factor. The Annual Daily Traffic (ADT) is a measure of the average traffic flow per day, for a certain year for a certain part of a road [67]. ADT is given as the number of vehicles per day for a specific vehicle class per day, can be given with indices as:

◦ ADT_tot - the total number of vehicles in both directions.

◦ ADT_l - the traffic flow in one lane. ◦ ADT_tot

heavy - the total flow of heavy vehicles in both directions. ◦ ADT_l

heavey- the flow of heavy vehicles in one direction.

ADT includes:

◦ cars.

◦ single-unit trucks and buses.

◦ multiple-unit trucks.

Annual Average Daily Traffic (AADT) is the yearly mean value of ADT. Average Daily Truck Traffic (ADTT) is an index describing the heavy traffic. The traffic volume is measured on a more or less regular basis, and to estimate the equivalent

heavy vehicle axles (N100) for other years a simple model was used, see Equation

2.1 [59]. N₁₀₀= AADTl∗ 3.65 ∗ phv∗ ¯n ∗ yn

∑

y0 (1 + r 100) y_, _(2.1) where:

◦ AADT_l is the AADT in one lane.

◦ p_hvis the part of heavy vehicles (ADT T

ADT ), given in %.

◦ n¯is the equvivalent number of standard axles per heavy vehicle.

◦ y∈ [1, 2, 3, ..., n], n ∈ N

. _y

0is the starting year of the calculation period.

. _y

nis the ending year of the calculation period.

◦ ris the traffic growth, in %, for heavy vehicles.

The standard axle is an imaginary axle with dual tires and a 100 kN load, homo-geneously distributed over the axle. The tire/pavement contact area is circular, and each area distributes a pressure of 800 kPa. The tires in each pair have a relative distance of 300 mm [67]. This is further illustrated by Figure 2.3.

(22)

2.3. DISTRESS TYPES CHAPTER 2. LITERATURE REVIEW

Figure 2.3: The derivation of the N100axle, used in Sweden. [59]

The Equivalent Single Axle Load (ESAL), is a concept which converts all axle configurations and axle loads of various magnitudes and repetitions (mixed traffic) to an equivalent number of standard or equivalent axle loads. A traffic estimation approach, based on the idea of fixed vehicles but varying ADT. The result is a number of repetitions of standard vehicle. The load on this axle varies internation-ally; in many parts of the world it is 18 kP. This allows the cumulative number of repetitions of standard axles during the design life to be expressed as the traffic parameter for design purpose [58, 27].

N100 is the same as N₁₀₀rut, N₁₀₀Cr, N₁₀₀Crini and N Crpropa

100 , which are appellations used in

the literature.

2.3 Distress Types

A distress in a road is not the same as a failure/fracture in many other fields of engineering. They can have many levels, and when it becomes unacceptable a failure occurs. Distresses can manifest in many different ways, and to avoid only treatment of the symptoms the cause needs to be found and dealt with in the proper manner [16]. The most common types are presented here.

2.3.1 Cracks

A fracture in the pavement leads to the formation of a crack. There are many types of cracks identified, and here follows a more detailed description. Illustrations of different types are found in Figure 2.4 [16].

Fatigue

When a pavement is subjected to repeated traffic loadings a set of interconnected cracks in the initiation phase can develop. They can usually be described as many-sided, sharp-angled pieces, often no more than 0.3 m on the longest side. These cracks are usually classified as bottom-up, but top-down cracks are also reported in the literature. The fatigue cracks are mainly generated in the wheel path. If unattended they can lead to Crocodile cracks.

Meandering Cracks

Cracks not connected, propagating in any direction. Transversal Cracks

Cracks similar to longitudinal cracks, mainly found in aged asphalt where the tem-perature differences cause the surface layer to contract and dilate.

(23)

2.3. DISTRESS TYPES CHAPTER 2. LITERATURE REVIEW Longitudinal Cracks

Cracks mainly caused by fatigue in the material. Diagonal Cracks

Single cracks formed across the lane. Block Cracks

Block Cracks are interlinked cracks that more often than the crocodile cracks, can appear on areas of the road not being subjected to traffic loads. Ruts can often be seen if crocodile cracks are present.

Crocodile (Alligator) Cracks

Are interconnected, forming a pattern that resembles the back of a Crocodile. These are caused by fatigue failures in the asphalt layers as a result of an high repetition of traffic loading. In pavements with thin layers the cracks normally initiate from the bottom and propagate upward; however the opposite is true for pavements with thicker layers. This enables moisture to infiltrate and can lead to potholes.

Crescent shaped Cracks

Crescent shaped cracks are usually the result of lateral and shear stresses from vehicles.

Figure 2.4: An illustration of the common types of cracks found in flexible pave-ments. [16]

Crack Index

The Crack Index (Ci) is an index that describes the severity of a crack. The growth

rate for the cracks differs in the early years compared to later. Generally the lifetime of a crack is separated into: initiation: 5 < Ciand propagation: Ci≥ 5 (See Figure

2.5) [78].

(24)

2.3. DISTRESS TYPES CHAPTER 2. LITERATURE REVIEW

2.3.2 Permanent Deformation

Asphalt consists of a mixture composed of aggregate, binder and air voids. The volume ratios are different in each mixture, giving different viscoelastic properties in permanent deformation. The permanent deformation can occur in every layer, but is only visible on the surface. An illustration of different types of deformation is found in Figure 2.6 [16].

Rutting

Permanent surface depression in the wheel path. Sometimes the pavement along the track can suffer uplift. This is a consequence of the compression in the HMA and/or subgrade layers, caused mainly by heavy traffic. Rutting can also be caused by wear from studded tires; studded tires causes abrasion of the wearing course, which results in rutting. The studs wear the bitumen in the wearing course and thereby the contact between aggregates in the wearing course and the tires in-creases. After repeated traffic passes the aggregates in the asphalt layer abrade and this will develop rutting in the wearing course [16].

Many previous studies have shown that flexible pavement smoothness is signifi-cantly affected by rut depth variance [22].

Shoving

Shoving is horizontal plastic movement of the surface. Usually caused by acceler-ating or retarding of vehicles.

Depressions

Depressions are local surface areas with a negative shift in the elevations compared to the surrounding pavement.

Corrugation

Corrugation is a form of plastic movement, seen as ripples on the surface.

Figure 2.6: Illustration of common types of deformation in HMA.[16] 2.3.3 Surface Defects

These types of distresses are mainly caused by asphalt concrete surface fatigue [16]. An illustration of different types is found in Figure 2.7.

Pot holes

The result of disintegration in the different layers. Besides from increasing the vul-nerability in the pavement structure they can cause damage to the vehicles passing. Patches

Patches are the result of repairing smaller local areas in distress. However a patch is considered a defect, as it generally increases the roughness which can impact the

(25)

2.3. DISTRESS TYPES CHAPTER 2. LITERATURE REVIEW fuel consumption and ride quality.

Delamination

Delamination sections of a surface layer that have come loose from the pavement. Polishing

The result of heavy traffic polishing the aggregates in the top layer creating a smooth surface with a low friction.

Raveling

Weakening of the bond between the aggregates and binder occurs at the surface of the layer the result is often the dislodgement of aggregate particles.

Flushing/Bleeding

Can occur when the aggregates at the surface are totally covered by binder. Result-ing in a softer surface with low friction.

Stripping

The weakening of the bond between the binder and aggregates. It initiates at the bottom of the Hot Mix Asphalt (HMA) layer and progresses upward.

(a) (b)

Figure 2.7: Surface defects. [16] 2.3.4 Edge Defects

Edge defects are the defaults found along a joint or the pavement edge. Usually there are two types of classified defects edge: break and drop off [16] (see Figure 2.8).

Figure 2.8: Edge Defects. [16]

2.3.5 Maintenance

The methods needed for repairing differ depending on type of distress, but the same strategy applies for all types of distresses;

(26)

2.4. PAVEMENT PERFORMANCE CHAPTER 2. LITERATURE REVIEW 1. Identify and find the cause.

2. Rectify the design for the cause of problem. 3. Repair the symptoms.

A road can of course have several different types of distresses without being main-tained. However when a road is maintained due to an unacceptable level of a distress in one type, several of other types can also be treated at the same time [42].

2.4 Pavement Performance

The Transportation Research Board (TRB) is a division of the American National Research Council, which has defined pavement performance as "A function of its relative ability to serve traffic over a period of time". The establishment of criteria for this task is vital. At the end of the 1950’s more objective measures started to appear [72]. This enabled the condition and performance of a pavement to be quantified, often described in terms of:

◦ Cracking

◦ Permanent deformation

◦ Surface Defects

◦ Edge Defects

[72]

The American Association of State Highway and Transportation Officials (AASHO) Road Test is a famous experiment carried out during the latter part of 1950. The tests had the aim of determining the effect of traffic in highway deterioration. The concept of Present Service Index (PSI) was derived during this study, based on data regarding the longitudinal roughness caused by: patch work, rutting, cracking, etc [1].

PSI is a qualitative index, from the users perspective, that separates structural and functional distresses in an pavement. The PSI relates to the quality of the ride from the users perspective. In order to prolong the service life of a pavement construction it can be used as an indicator of when to perform maintenance. This is described in Figure 2.9 [1].

(27)

2.4. PAVEMENT PERFORMANCE CHAPTER 2. LITERATURE REVIEW

Figure 2.9: PSI is a qualitative index, from the users perspective, that separates structural and functional distresses. When maintenance or rehabilitation is per-formed the PSI value is often decreased compared to the start value. [73]

The allowed PSI value, described in Ta-ble 2.1, before maintenance is decided by the road owner, and usually economy is a factor. The PMS concept is interna-tionally recognized to be a highly cost effective tool in planning road construc-tion and maintenance [1].

Table 2.1: The values for PSI.

0 - 1 Very Poor

1 - 2 Poor

2 - 3 Fair

3 - 4 Good

4 - 5 Very Good

2.4.1 Present Serviceability Rating

The Present Serviceability Rating (PSR) is the mean of the individual rating, de-rived from a specific panel. The PCR is calculated as:

PSR≡ 5.03 − 1.91log10(1 + SV ) − 1.38RD − 0.01

√

C+ P, (2.2)

where:

◦ PSRis the Present Serviceability Rating.

◦ SV is the average slope variance.

◦ RDis the average rut depth in inches.

◦ C+ P is the cracking and patching in [ f t

2

1000 f t2].

However, this was considered unpractical since the panel was supposed to consist of 12 trained persons [1, 55, 80, 30, 44, 38, 17, 8, 12, 16].

2.4.2 Swedish Models

The performance models used in Sweden for crack and rut predictions were created by Wågberg and Göransson [78, 19].

Wågbergs crack model

In the crack propagation model developed by Wågberg, [78], the accumulated

heavy traffic loading is represented by equivalent standard axle, N100, loading

(28)

2.4. PAVEMENT PERFORMANCE CHAPTER 2. LITERATURE REVIEW sum of the standard axle repetitions to crack initiations and for crack propagation respectively at failure as:

N₁₀₀Cr = NCrini

100 + N Crpropa

100 . (2.3)

Whereas the number of axle repetitions for crack initiations is given by Equation 2.4:

NCrini

100 = 10

7.24−0.0052∗SCI300−5010000∗ 1

SCI300∗NY100, (2.4)

and for propagation by Equation 2.5.

N₁₀₀Crpropa= 195 ∗ 10

5

4.39 +7.1∗106

N₁₀₀Crini

. (2.5)

◦ N₁₀₀Y is the average annual N₁₀₀per lane.

◦ SCI₃₀₀is the surface curvature index in [µ m] based on FWD measurements

carried out at temperature 20°C on the structure recently after the structure was built.

Göranssons rut model

A similar approach, as previously described in section 2.4.2, has been used to pre-dict the number of accumulated standard axles to reach a certain failure rutting depth. In the model developed by Göranssons, [19], the total number of ESAL’s for specific segment is given by Equation 2.6:

N₁₀₀rut = 1 0.9533 ∗ rut−0.0209∗ ( rut a ) 1 b_. _(2.6) Where: ◦ Nrut

100is the average annual (ESALs) per lane.

◦ rutis the total rutting in [mm] on the surface used to define failure.

◦ aand b are parameters estimated from Falling Weight Deflectometer (FWD)

test as the surface curvature index SCI300in µm measured during the autumn,

first time after the pavement structure (section) is built or rehabilitated. The parameters a and b can be estimated as:

a= 0.0001579 ∗ SCI300+ 0.034322,

b= 0.0005695 ∗ SCI300+ 0.296.

(29)

2.5. ROAD MANAGEMENT CHAPTER 2. LITERATURE REVIEW

2.5 Road Management

Road Management Systems (RMSs) are used to store and analyze road data as a part of the infrastructure of a society. The work method is to attain the data needed to systematically analyze and prioritize and thereafter take action needed for maintenance and planning of everything related to the road network [74] (see Figure 2.10).

Figure 2.10: An illustration of the concept of RMS and PMS.

A PMS can be described as "A set of tools or methods that can assist decision making in funding cost effective strategies for providing, evaluating, and main-taining pavements in a serviceable condition". A PMS is a part of the RMS, ad-dressing only the road structure explicitly. The Federal Highway Administration (FHWA), U.S. Department of Transportation, uses the software Pavement Health

Track[74].

2.5.1 History

The first systems appeared in the latter part of the 1970s, and in 1985 AASHTO published "Guidelines on Pavement Management". In 1990 the AASHO gave the PMS guidelines: A Pavement Management System is designed to provide objec-tive information and useful data for analysis so that highway managers can make consistent, cost-effective, and defensible decisions related to the preservation of a pavement network. A PMS, sometimes refereed to as a Maintenance Management System (MMS) is a tool for planning the maintenance or rehabilitation of roads. These strategies require prediction models, and LTPP programs are a powerful tool when deriving them. [74]

2.5.2 Decision Support System & Geographic Information System Decision Support System (DSS) is a software based information system that sup-ports the decision making process, and a Geographic Information System (GIS) is a multipurpose system designed to store and retrieve relevant information that can be connected to a geographical position. An implementation of such system is shown in Figure 2.11.

(30)

Figure 2.11: An example from a PMS with DSS and GIS. [54]

Using a PMS with an integrated DSS and GIS is a helpful tool in any PMS [47, 80, 16, 31].

2.5.3 Highway Development and Management Model-IV

The World Bank has developed a set of tools for highway development and man-agement to be used in PMS. The area of use for the HDM-IV tools has expanded beyond the traditional project appraisal and management to a potent system for road management and investment analyzing. The gathered data for these models are seen in Figure 2.12:

(31)

Figure 2.12: The conception of the life-cycle analysis in HDM-4 [63]. The maintenance strategy is to use a road asset as long as possible, by conducting planned maintenance before the allowed amount of deterioration considered ac-ceptable is reached. The allowed deterioration level is decided by the road owner; usually economy is the most important factor [63, 64, 28, 74].

PMS Methodologies

There are generally two levels for the usage of a PMS: network and project. The network methodology results in solutions optimized for the entire road network. Using large quantities of combined data the optimal strategy is found, the next

(32)

2.6. LTPP CHAPTER 2. LITERATURE REVIEW step here is to target the smaller objects. The project approach solves the problem using a fundamentally different technique, working from the low to the higher level [63, 64, 28, 74].

2.5.4 PMS in Sweden

The Pavement Management System in Sweden started with a high focus on the quality of road condition data in order to attain reliable and repeatable informa-tion for analysis and research. Sweden uses average values for 100 m secinforma-tions, the standard for maintenance is based on posted speed and aloud traffic classes. The software "PMS 95" is currently used (but will be replaced by "PMS 2012") using data from the National Road Database (NVDB). "PMS 95" focuses on the information demand of pavement engineers, and is intended to aid the planning of maintenance. In Sweden all the maintenance projects are outsourced. A contrac-tor in Sweden is obliged to follow national asphalt norms. The Swedish Transport Administration (STA) is divided, and the section Society operational area consists of six regions [35].

2.6 LTPP

The original Long Term Pavement Performance (LTPP) was a project initiated in the beginning of 1980 by the American TRB and AASHTO and the project began with monitoring the deterioration of the North American highways, and storing the results in a database. One of the outputs was the Strategic Highway Research Program SHRP, with a focus on a LTPP monitoring program, among others. The ambition of this original program was to:

◦ analyze existing design methods.

◦ develop improved design methods and strategies for rehabilitating existing

pavements.

◦ develop improved design equations for new and reconstructed pavements.

◦ determine the effects from:

. _loading . _environment . _{material properties} . _{material variability} . _{construction quality} . _{maintenance levels}

on pavement performance and maintenance

◦ establish a national database of pavement inventory and performance

infor-mation [74].

The main purpose on any LTPP program is to prolong the life period of any pave-ment structure by monitoring various pavepave-ment designs and rehabilitated objects, using different types of construction methods and materials; subjected to different loads, environments and climate. USA. and Canada collaborates in this project, and have a joint LTPP program, with approximately 1500 test sections. The C-LTPP project was established in 1989. In the U.S. and Canada each State/Province De-partment of Transportation collects pavement condition data on a regular biannual basis. The data is then reported to one of the six Regional LTPP Data Managers. A pre-investigation is then performed to control the quality. The final result is then

(33)

2.6. LTPP CHAPTER 2. LITERATURE REVIEW accessible by anyone in the transportation community [74].

In Canada this is executed by Transport Canada, the department responsible for transportation, in Transport, Infrastructure and Communities Portfolio [23, 69]. 2.6.1 General and Specific Pavement Studies

In the "original" LTPP program two different types of studies are preformed: Gen-eral Pavement Studies (GPS) and Specific Pavement Studies (SPS). The GPS ex-periment targets "ordinary" structural designs of pavement versus the impacts of climate, geology, maintenance, rehabilitation, traffic, and other factors. The SPS has the aim of investigating the more "pavement engineering" factors. The objects in the SPS are designed to provide a different set of design factors, enabling com-parison of the performance of dissimilar design factors, both within and between sites [74].

The GPSs were agency nominated to fit into broad categories but were paved as part of normal paving operations within the agency. The GPS sites entered the LTPP program with a variety of initial ages, were in a variety of conditions, and were made of a variety of materials. Because they were paved as part of longer paving projects, the GPS sites were believed to be more representative of standard paving processes and ride quality than were the 500-ft test sections in the SPS pro-gram [13, 14, 74, 45, 51, 76].

Canada

The Canadian LTPP (C-LTPP) monitors 24 test sites in Canada that deals with re-habilitation of an asphalt concrete pavement. Furthermore, each test site contains two to four adjacent test sections, each with a different rehabilitation strategy en-abling a focused study on the design and optimization of overlay. The C-LTPP started in 1989 by the C-SHRP in order to compliment the U.S. LTPP, targeting factors more relevant to Canada. The C-SHRP monitors the construction and per-formance of 24 highway test sites, distributed across the major provincial system. The focus is upon asphalt concrete overlays, constructed on existing AC pavement having a granular base course. This gives information concerning prior condition of the pavement. The climate and the lower (compared to the US) traffic flow were also important factors when selecting objects suitable for monitoring. The C-LTPP test objects are monitored throughout an entire life cycle (∼ 15 years). The aims of this project were to:

◦ analyze the efficiency of the current rehabilitation methods and strategies.

◦ create new or calibrate performance models to suit the local conditions.

◦ institute common methods for LTPP evaluation, and build a base for future

research projects.

(34)

2.7. DISTRESS FACTORS CHAPTER 2. LITERATURE REVIEW The highway agencies in Canada were given technical guidelines containing de-tailed information in how to collect and report different types of performance data [7, 23].

2.6.2 LTPP Database

Gathering and storing the data require several millions of $ in funding, and this demonstrates the importance of transport. One of the most common ways to store the LTPP data is in a relationship database, i.e. a multipurpose storage concept. The data is simply stored in tables (row/column) with different types of relations to each other. This can be a very powerful tool, and at the same time highly user-friendly with a good Relational Database Management System (RDBMS) creating SQL queries combined with a logical designed Enhanced Entity Relationship Model (EER) [74]. More information about the EER database can be found in appendix B.

2.6.3 International LTPP

The interest in this project rapidly spread throughout the world. Some of the im-plementations are further described here. [74]

The Nordic Countries

The original SHRP LTPP program was not designed for the cold climate in the north (Finland, Norway, Sweden); therefore a different set of data samples are used. "Cold Climate" can be defined by the Freezing Index (the number of days times degrees below freezing point) at least 100. Frost is an extremely important design factor resulting in thin overlays and thick unbound layers [35].

New Zealand and Australia

The New Zealand LTPP project started in 2000, and after performing studies on several existing LTPP projects the aim was to:

◦ establish a representative sample of LTPP sections across New Zealand,

re-lating to the interaction effects of climate and sub-soil moisture sensitivity.

◦ reinforce the current road data collection program.

◦ collect data at the precision level needed for the models used.

Australia’s Austroads (Australia’s peak road agency organization), funded an LTPP study in 1994/95 with the aim of improving the existing pavement performance prediction models [26, 24].

2.7 Distress Factors

In this section the main factors causing the pavement distresses are presented, though there are many more.

2.7.1 Factor Traffic

Based on the data from [56] the number of vehicles is increasing and highways are often constructed to deal with this. The vehicles give a dynamic load/unload stress, that cumulative gives permanent strain shown in Figure 2.13. Here the strain

(35)

2.7. DISTRESS FACTORS CHAPTER 2. LITERATURE REVIEW increases differently depending on how many axle loads the site has been subjected to [57].

Figure 2.13: Creep characteristic of bituminous mixtures. [57] The results from a dynamic repeated load test are shown in Figure 2.14.

Figure 2.14: The graphs show deformations of the pavement due to the cyclic load/unload. [16]

The geometrical placement of the wheel/pavement interface differs for different axle types; the most common types and tire configurations are shown by Figure 2.15. The current rules in the EU are a maximum length of 18.75 m and a maximum weight of 40 tons, however in Sweden 60 tons are allowed. The method used to sample vehicle weight is often static weigh stations, which produce data with a

(36)

2.7. DISTRESS FACTORS CHAPTER 2. LITERATURE REVIEW very high bias [62].

Figure 2.15: Illustration of the common axle types and tire configurations. [48] This can be avoided by using Weigh in Motion (WIM) technique, which enables the weight of the axles to be measured when the vehicle is in motion so that the traffic can be divided into an axle load spectrum. This results in a spectrum of load pulses with different frequencies. Each vehicle type contributes to the distress of a pavement; the load from the wheels creates a unique stress wave that travels through the pavement (see Figure 2.18) [49].

Figure 2.16: The graphs show axle loads from different types of axles.[49] In Sweden the following axle loads are allowed:

◦ 11.5 ton on a steering axle.

◦ 10 ton axle load.

◦ 18 tons on a tandem axle.

◦ 19 tons on a driving tandem axle.

◦ 60 ton total weight.

(37)

2.7. DISTRESS FACTORS CHAPTER 2. LITERATURE REVIEW Tire/Pavement interface

The shape of the contact area between tire/pavement depends on the tire pressure; this is one of the elements in the pavement response. According to the rules in Sweden studded tires are required from the first of November to the end of April. The studded tire can give a mechanical dislodging of aggregates, due to wear, sim-ilar to the pavement deformation caused by heavy loading [27, 70, 20]. The rut given by the LTTP database does not give the source [20].

2.7.2 Climate

The climate is a key factor in road deterioration, and several different types of data are available, such as the number of days of frost, per year and rain amounts etc. These factors are strongly dependent on their geographical placement, as seen in Table 2.2 and Figure 2.17. The climate has also a significant impact on the design of pavements. The strength, durability and load bearing capacity of the pavement are affected, adding traffic and complex interaction effects [36, 67].

Figure 2.17: The climate zones classified by the number of days when the mean

temperature is below 0◦C in Sweden. [67]

The climate plays an important part in the selection of materials and for the thick-ness dimensioning of the layers. Table 2.3 gives the temperatures. The thickthick-ness and material are selected to protect the road against frost heave/thaw.

(38)

2.7. DISTRESS FACTORS CHAPTER 2. LITERATURE REVIEW Table 2.2: The periods length in days during one year, [67].

Climate zone : 1 2 3 4 5 Winter : 49 80 121 151 166 Frost heave : 10 10 Frost thaw : 15 31 45 61 91 Spring : 46 15 Summer : 153 153 123 77 47 Fall : 92 76 76 76 61

Table 2.3: Temperature, [◦C], in the bitumen bounded layers, [67].

Climate zone : 1 2 3 4 5 Winter : -1.9 -1.9 -3.6 -5.1 -7 Frost heave : 1 1 - - -Frost thaw : 1 2.3 4.5 6.5 7.5 Spring : 4 3 - - -Summer : 19.8 18.1 17.2 18.1 16.4 Fall : 6.9 3.8 3.8 3.8 3.2

To represent the climate the Mean Annual Temperature (MAT) in◦C, and the

per-ception, measured in [mm], was used. The values are calculated as the mean value over a longer period. The climate is believed to be an important factor, however the lack of control over the observations is believed to be the cause for the unexpected low correlation values.

2.7.3 Aging

This term describes the effects of long term exposure to the failing factors and how this changes the material properties by altering the bond strength in and between molecules, the chemical structure of the bitumen and the settings for the exposure. This is measured in years from opening [60, 18].

2.7.4 Material

The Pavement body is the engineered part, where material and geometrics can be designed; the subgrade is the natural occurring soil. The properties and the settings change with time and the subgrade can be composed of a wide range of different materials; some are less desirable when it comes to road construction.

Structure

An important explanatory variable for crack initiation of flexible pavements is the

Surface Curvature Index (SCI300), used as a value of the strength of the pavement

(39)

2.7. DISTRESS FACTORS CHAPTER 2. LITERATURE REVIEW Subgrade

The quality of a subgrade can be described by four highly interconnected factors: 1. Load bearing capacity.

2. Moisture content.

3. Frost heave\thaw

4. Shrinkage and\or swelling. The load bearing capacity

A "good" subgrade is able to withstand the forces transmitted by the pavement structure without deforming. This parameter can be described as the stiffness of

the subgrade. The Resilient Modulus, (Mr), is required when for determining the

stresses, strains, and deflections in pavement design.

Mr≡

qd

εr

, (2.7)

where

◦ M_ris the Resilient Modulus.

◦ q_dis the applied deviatoric stress. ◦ εris the resilient strain.

[27]

The deviator stress is a result when there is a difference in the principal stresses σ1

and σ3, qd= |σ1− σ3| [32].

According to the STA [60] the Mrcan be estimated as:

Mr=

5200 D900

(2.8) where

◦ D₉₀₀ is the deflection measured at 900 mm from an FWD measurement,

descried in section 2.8.2.

The SCI300and Mrvalues are to be regarded as constants, for both phases in the Ci

and rut model.

Climate and Material

For most of the highways the design is, among other factors, based on typical his-toric climatic patterns, reflecting the local climate and incorporating assumptions about a reasonable range of temperatures and precipitation levels. Regions with cold climate can be subjected to frost and freezing of the roadbed. Certain sub-grade soils are particularly susceptible to frost action, though generally frost heave is limited to areas with silty soils. Frost heave is the upward motion of the sub-grade, caused by the expansion of the moisture in the soil when it freezes. If the enthalpy increases the moisture changes state from solid to liquid, resulting in a re-duced bearing capacity. The temperature in the pavement also affects the oxidation process that can increase the viscosity properties. In the presence of moisture, one of the mechanisms for the deterioration of asphalt is the debonding effect. This

(40)

2.7. DISTRESS FACTORS CHAPTER 2. LITERATURE REVIEW can cause durability problems such as stripping, release of stones, and cracking. While relatively unimportant for loose aggregate, aggregate chemical properties are important in a pavement material. In HMA, aggregate surface chemistry can determine how well an asphalt cement binder will adhere to an aggregate surface. Poor adherence, commonly referred to as stripping, can cause premature structural failure. The materials used in road construction are often described in terms of:

◦ Gradation and size

◦ Toughness and abrasion

resis-tance

◦ Durability and soundness

◦ Particle shape and surface texture

◦ Specific gravity

◦ Cleanliness and deleterious

mate-rials

Previous Ci/Rut value

The state of the object is highly dependent on the Previous Value (PV). The PV acts as the "starting point" when making predictions.

Pavement Deterioration

The reasons for a road to fail are many, and the distress a road has to withstand varies over time with frequency and severeness. The Figure 2.18 illustrates the complexity of road performance.

Figure 2.18: An illustration of performance of pavement structure factors. [53] Pavement deterioration is a negative change in performance or condition of the pavement, i.e, an increase in distress or decrease in serviceability. Pavements are often designed to maintain functionality in whichever environment they are built. However material properties vary with the seasons and temperature changes. The climate’s effect on a road also depends on the environment in which it is built, some microhealing has been noticed for pavements [74].

(41)

2.8. ROAD MEASUREMENTS CHAPTER 2. LITERATURE REVIEW

2.8 Road Measurements

The objects in the LTPP are inspected on a yearly basis and are given a Ciaccording

to the principles described in: [77]. Here follows a description of some of the data found in the LTPP and how they are measured [20].

2.8.1 The Road Surface Tester

The RST measures the state of a road, and the data is stored in the LTPP database. The sampling is done automatically when the car is driven, the RST is illustrated by Figure 2.19 [75].

Figure 2.19: The RST used at VTI, and an illustration of the rut sampling and how the lasers are placed.

The figure also gives an illustration of the sampling process. Sampling with 17 lasers is the current standard, a measuring distance of 3.2 m is reached, however the Road Surface Tester (RST) owned by VTI carries 19 laser units and that gives 3.6 m. A sampling occurs every 10 cm and using the data the rut depth is calculated. The placement of the lasers is shown in Figure 2.19. More information regarding these tests can be found in: [71].

2.8.2 Falling Weight Deflectometer

The purpose of a Falling Weight Deflectometer (FWD) is to simulate the load pulse from a heavy vehicle. This is done with a load plate of 30 [cm] being dropped

(42)

2.8. ROAD MEASUREMENTS CHAPTER 2. LITERATURE REVIEW from a height so that a force of approximate 50 [kN] is reached when it impacts the pavement surface. This is usually done three times, and the deflection is measured at the impact center and at 0, 200, 300, 450, 600, 900 and 1200 mm from the center. The FWD are spot measurements representing longer sections, and the road needs to be closed while measuring. The aim here is, of course, that this type of data can be attained without disturbing the traffic. The Figure 2.20 shows the placements of the geophones, and Table 2.4 states how the results are used [11].

Figure 2.20: A description of the FWD measurements. [60, 11]

Table 2.4: Deflection Basin Indexes. [11]

Sensor no., Di D1 D2 D3 D4 D5 D6 D7

(43)

Maximum deflection recorded - D1- Indirect proportional to the overall stiffness

of an elastic half-space.

Surface Curvature Index - SCI300 = D1− D3- Curvature of the inner portion of

the basin. Indicates the stiffness of the top part of the pavement.

Base Curvature Index - BCI = D7− D6- Curvature of the outer part of the basin.

Indicates the stiffness of the bottom part of the pavement or the top part of the sub-grade soil.

Base Damage Index - BDI = D3− D5- Curvature of the middle part of the basin.

Indicates the stiffness of the pavement.

Basin Area - Area = _D1

0∑

N

i=0((Di−1+ Di) ∗ (ri− ri−1)) - Considered a good

indi-cator of overall pavement strength during spring thaw.

Radius of Curvature of the center of the basin - R = (D1− Dx)2+ a2)

2(D1− Dx) ≈

a2 2(D1− Dx)

-Dx is the deflection measured at the sensor just outside the loading plate and a is

radius of the plate.

Tensile Strain at the bottom of the AC layer - ε = h1

2R - h1 is the thickness of the

asphalt bound layer [mm].

Subgrade Strength Index - SSI = D5t

D5s - SSI indicates a measurement during thaw

and SSI indicates measurement after thaw recovery.

The deflections are measured by accelerometers that measure the vertical displace-ment speed of the surface. The influence of passing traffic is eliminated by having a very short sampling time when the weight is dropped [52, 60, 11, 40].

2.8.3 Rut

The rut data is sampled by the RST, which measures unevenness in the horizon-tal plane. In the LTPP database rut depth, given in mm, is measured by a RST, equipped with 17 Laser sensors, the data are measured in both directions using the wire surface principle, illustrated by Figure 2.21, and the average value from all of the sections in an object are used in this study [21].

Figure 2.21: An illustration of the wire surface principle in rut depth sampling. [20]

The rut data stored in LTPP are the ruts in the left and right track, and the maximum depth for the entire sampling width.

Rut variance

The rut values in the LTPP database are given in mm, and from the objects inves-tigated they seem to have a profile with a small value of variance (see the plots in Figures 2.22) [21] .

(44)

Figure 2.22: Illustration of the variance in the rut data. The data plotted comes from object D-RV53-2. [21]

The data plotted come from the object D-RV53-2, and shows how the rut values from all sections in an object have been transformed into one value representing the whole object. The sampling is done automatically several times for each section and gives deviations each time the rut is measured. When needed, the available rut data were used to interpolate new values, in order to replace missing samples in an interval.

(45)

2.8. ROAD MEASUREMENTS CHAPTER 2. LITERATURE REVIEW 2.8.4 Crack Index

Crack index (Ci), is an index derived in the EU project "PARIS", [15]. Every

section in a road object is manually inspected and given an Ci, that includes all

types of cracks. A crack is classified and given a weight in the calculation, the magnitude of the weight is dependent on the classification and size [77].

Figure 2.23: Example of size and development of a crack between three different inspections of the same road that is carried out at different times. This object is 100 [m], and divided into smaller 10 [m] sections.[77]

The index increases with the level of severity and spreading, but also with the type of crack. The crack index is empirical, based on a visual survey, and is calculated as Equation 2.9:

Ci= 2Ac+ Lc+ Tc, (2.9)

where:

Ac: Alligator cracking; Ac low[m] + 1.5 ∗ Acaverage[m] + 2 ∗ Acbad[m]

Lc: Longitudinal cracks; Lclow[m] + 1.5 ∗ Lcaverage[m] + 2 ∗ Lcbad[m]

Tc: Transversalcracks; Tclow(no) + 1.5 ∗ Tcaverage(no) + 2 ∗ Tcbad(no.)

Low, average and bad are weights defined in [77], and no. stands for the number of cracks. Cracks shorter than 1[m] are assigned a length of 1[m] [78].

The Ci used in this project is the mean value from the sections in an object.

How-ever the magnitude of the Ciincreases the standard deviation in the prediction. As

(46)

Figure 2.24: Civariation in an object. [21]

The figure shows that the standard deviation is low in the first time period, but once cracks appears in different sections they grows with a speed that increases

with time. Resulting in large variances between the sections in an object. The Ci

can be divided in to two phases: initiation (when Ci < 5) and propagation (when

Ci ≥ 5). This is due the different growth rate in the phases and creates the need of

(47)

Chapter 3

METHOD

The prediction models created are empirical-mechanistic models, based on data from a database that include the structural information, traffic volume, and con-dition data for each "homogeneous" section of a road. The road objects used are described by Table 3.1, 3.3, 3.4.

Table 3.1: General object information.

Object Name Opened [year] Sections no. Speed [km_h ] Observation Length [year]

C-292-1 Gimo 1994 9 90 16 D-RV53-2 Nyköping 1987 10 90 22 F-RV31-1 Nässjö 1988 11 90 20 H-RV34-1 Målilla 1987 10 90 16 H-RV40-2 Vimmerby 1980 12 90 30 T-205-1 Laxå 1994 8 80 16

Table 3.2: General construction and maintenance information.

Object Name Surface layer ∆Thickness [mm] Maintenance

C-292-1 Gimo AG, Stabinor 85 1994

ABb, ABS 43 2009 D-RV53-2 Nyköping MABT 40 1987 HABS 28 1993 ABS 43 2008 F-RV31-1 Nässjö AG 85 1988 Remixing, ABS 24 2007 Remixing, ABS 24 2007

H-RV34-1 Målilla AG, MAB 75 1987

HABS, Heating 28 2002 HABS, Heating 28 2002 H-RV40-2 Vimmerby BG, MABT 75 1980 ABT 13 2008 ABT 12 2008 ABT 8 2008 Remixing, ABS 28 2009 Remixing, ABS 28 2009

(48)

CHAPTER 3. METHOD The materials in the surface layer are further described by [68].

Table 3.3: Material information continued.

Object Subgrade Type Subbase

C-292-1 clay 3 rock

D-RV53-2 silty clay / silt 3 gravel and sand

F-RV31-1 silty moraine 6

-H-RV34-1 sand 1 gravel and sand

H-RV40-2 s moraine/s moraine on rock 6

-T-205-1 blockr Si Mn 6 crushed/uncrushed mtrl

Table 3.4: Material information continued.

Object Base Course Thickness [mm]

C-292-1 crushed rock 160 D-RV53-2 gravel 160 F-RV31-1 gravel 115 H-RV34-1 gravel 125 H-RV40-2 gravel 80 T-205-1 crushed soil/rock 150

Figure 3.1: The red dots in the map mark the geographical locations of the objects. The subgrade is classified in the same way as in the EU project PARIS;

1. sand 2. silty sand 3. clay 4. peat 5. bedrock 6. other [15]

(49)

3.1. THE SWEDISH LTPP CHAPTER 3. METHOD

3.1 The Swedish LTPP

The Swedish LTPP project started 1984, and has continued to grow ever since. It began with a limited number of objects but has expanded with time. The LTPP database contains relevant pavement information with focus on road deterioration caused by heavy traffic and exposure to climate. Today it contains data from over 650 sections, distributed in over 65 road objects in the stately road network, the object are represented as dots the map in Figure 3.2. The objects are selected from the national road network to ensure that they are constructed according to national standards. Each object is divided into smaller, 100 m sections but the number of sections in an object varies. The performance monitoring is mainly focused on road deterioration caused by heavy traffic.

Figure 3.2: The geographical distributions of road objects in the Swedish LTPP database. The red dots are active objects, and the orange objects retired from data monitoring.

Most of the road objects in the LTPP database are concentrated toward the southern part of the country as most of the traffic is there. The Swedish LTPP has a database containing almost all relevant pavement information. The Swedish Transport Ad-ministration (STA), has given the National Road and Transport Research Institute (VTI) the assignment to collect a large amount of data concerning the state of several objects in Sweden. The main objective for this project is to create road de-terioration models for a PMS, used for conducting road maintenance. An overview of the stored data in the Swedish LTPP database is given by the list:

◦ road construction and structure

◦ material properties

◦ road state

◦ climate

◦ traffic

◦ pavement distresses

The Swedish LTPP database is accessible to the public and has been used in var-ious research projects for pavement performance analysis and management deci-sions. The Swedish pavement design software "PMS Objekt" has been verified

(50)

3.2. TRAFFIC ESTIMATIONS CHAPTER 3. METHOD with the LTPP database. The database can also be used for calibrating and vali-dating models, that require detailed data regarding local conditions of pavement structures, bound and unbound material characterization, environmental condi-tions, traffic loading, and distress data, such as rutting, cracking, that stretches over the lives of pavement structures. The database is stored in a Microsoft Access format. [65, 20, 60, 49]

3.2 Traffic Estimations

From the measured ADT and ADTT the change in the ADT and ADTT has been has been estimated with models given by equations 3.1 and 3.2:

ADTi+1= ADTi∗

(1 + r)Y− 1

r , (3.1)

ADT Ti+1= ADT Ti∗

(1 + r)Y_{− 1}

r , (3.2)

where:

◦ ADT_i- is the measured ADT for year i.

◦ ADT_i+1- is the estimated ADT for the next year.

◦ ADT T_i- is the measured ADTT for year i.

◦ ADT T_i+1- is the estimated ADTT for the next year.

◦ Y - is the number of years from the measured.

◦ r- is the traffic volume growth, estimated to 3%.

◦ i∈ [0, 1, 2, ..., n], n ∈ N

The ADT data used in this project from data comes from [66], and the models from [27].

Results are shown in the plots in Figures 3.3, 3.4, 3.5, 3.7 and 3.8. The traffic volume is plotted as a function of time in the ADT and ADT T estimations. In all the plots are the black dots are the traffic measured; the red lines gives the estimations used in the project.

(51)

3.2. TRAFFIC ESTIMATIONS CHAPTER 3. METHOD

Figure 3.3: Traffic estimations for object H-RV40-2. The black dots are the traffic measured and the red lines are the estimations used in the project.

Modeling pavement performance based on data from the Swedish LTPP database : predicting cracking and rutting