ASPECTS OF CEMENT STABILIZED MOZAMBIQUE SAND APT TRAFFICKING

ASPECTS OF CEMENT STABILIZED MOZAMBIQUE SAND APT TRAFFICKING

ASPECTS OF CEMENT STABILIZED MOZAMBIQUE SAND

BASE MATERIAL PERFORMANCE UNDER MMLS3 AND MLS10

APT TRAFFICKING

Fred Hugo

1

, Eben.R. de Vos

2

and Hilário Tayob

3

1

Director, Institute for Transport Technology, University of Stellenbosch

Stellenbosch, South Africa 7600

Tel. +27 21 808-4364 Fax.+27 21 808-4361; Email: [email protected]

2

Researcher, Institute for Transport Technology, University of Stellenbosch

Private Bag X1, Stellenbosch, South Africa 7600

Tel. +27 21 808-4079 Fax.+27 21 808-4361; Email: [email protected]

3

ANE Project coordinator, National Administration of roads

Directorate of National Roads, Engineering Department

Av. de Moçambique 1225, C.P. 1439, Maputo - Mozambique

Phones +25821476163/7 - Fax +25821475862 ; Email : [email protected]

ABSTRACT

The World Bank sponsored research to support preservation and maintenance efforts in

Mozambique. The object was to develop guidelines for a mechanistic-empirical design method for cement stabilized sand bases (CTB) in Mozambique using Accelerated

Pavement Testing (APT) technology. APT encompassed scaled (one third) and full-scale

APT using mobile load simulator technology (MMLS3 and MLS10). The MLS10 expanded the MMLS3 APT investigation to full-scale pavements, including wet trafficking cycles to emulate environmental effects. Operation of the MLS10 is similar to the MMLS3. It has four sets of dual wheels running in a closed loop with a maximum speed of up to 22 kph. Loads of 60kN and 70kN were used for trafficking. Fifteen MMLS3 tests and six full-scale test sections were tested. Comparative results between MMLS3 tests and full-scale MLS10 tests were found. Distress mechanisms were similar. Related design features and material strength parameters were used as supplementary and verification tools.

INTRODUCTION

The World Bank sponsored this research project to support preservation and maintenance efforts in Mozambique. The object was to develop guidelines for a mechanistic-empirical pavement design method for cement stabilized sand bases (CTB) based on Accelerated

Pavement Testing (APT) technology. The APT program encompassed both scaled (one third) and full-scale APT using mobile load simulator technology (MMLS3 and MLS10).

The scope of the paper is limited to a presentation of aspects of Mozambique CTB material performance under the APT trafficking. The APT was focused on three primary phases:

• Exploratory scaled laboratory MMLS3 testing

• Construction of field test sections

• Full-scale and scaled field APT by means of MLS10 and MMLS3

Proceedings of the 26 th

Southern African Transport Conference (SATC 2007)

ISBN Number: 1-920-01702-X

Produced by: Document Transformation Technologies cc

821

9 - 12 July 2007

Pretoria, South Africa

Conference organised by: Conference Planners

The MMLS3 was used to evaluate the performance of the stabilized coastal sands under different trafficking conditions in the laboratory. This was to serve for identifying potential failure mechanisms for comparison with failure mechanisms observed in the field. In similar vein, this information was to provide the basis for formulation of the specifications for constructing full scale test sections to be tested by the MMLS3 and the MLS10. It was also used as input for the mechanistic modelling. Fifteen MMLS3 tests, including two field tests were completed as well as six full-scale MLS10 tests.

SCALED LABORATORY TESTING PROGRAM

The most commonly and abundent material used by local consultants namely, the reddish and yellowish coloured sands were used in this study. The sands were evaluated in terms of performance under MMLS3 APT as well as their physical, chemical and strength charactistics through supplementary and verification testing (S/VT).

Two binder types were selected for stabilisation namely, cement and a fifty percent cement-lime blend. This decision was based on past and current practise of using cement, as well as exploration of performance of a possible alternative namely, a blend of cement and lime. These two binder types were used for the construction of the scaled and fullscale test sections as well as specimen preparation for material strength testing.

The test load characteristics of the MMLS3 are summarised in Table 1 together with the load characteristics of the MLS10. The findings of the MMLS3 tests are summarized in

Table 2. A set-up view of the MLS10 trafficking on the test site at Manhiça in Mozambique is shown in Figures 1.

The results from the laboratory and field MMLS3 tests enabled:

♦ Comparison of the relative performance of the stabilising agents.

♦ Understanding the manifested distress mechanisms

♦ Adjudicating the expected pavement performance in order to predict/estimate the full-scale performance characteristics of the test sections

Both the red and yellow coastal sands performed well after stabilizing with five percent cement or a blend of 5% blend of cement and lime (ratio 50:50). The load applications were respectively 1.3 and 1.4 million. The relationship of these axles to the full-scale

MLS10 is discussed later. The seven percent sections withstood trafficking in excess of three million axle load applications.

Environmental trafficking conditions were simulated by:

• initially trafficking at ambient temperature for 50k followed by

• alternate heated trafficking at 50C at 20mm depth for the next 200k and

• 50k wet trafficking cycles at ambient temperature with water sprayed on the pavement surface.

Figure 1: Set-up view of the MLS10 trafficking on the test site at Manhiça in Mozambique

822

Table1: Comparison of MMLS3 and MLS10 APT Machine Parameters and Load

Characteristics

Item

2

3

5

6

8

Tread path width

Size of test section

Wheel load

Wheel suspension

Load propulsion

9 Power

10 Power consumption

11 Temperature control

12 Trolley description

13 Data acquisition system

14 Load control

15 Mass

16 Data acquisition

18 No of wheels

19 Tyres

20 Tyre pressure

21 Speed of operation

22 Rut depth (max)

23 Repetition /h

24 Trafficking conditions

25 Trafficked width

26

Lateral Traffic Load

Distribution

27 Lifting mechanism(s)

28 Lateral displacement

29 Short haul mobility

30 Long haul mobility

1/3 Scale MMLS3 m 1 mm 80 m 1 x 0.4 (min)

Single kN 1.9 - 2.9

Spring m/s 2.5

Electric Drum drive

Electric kW 1.5

Yes (Hot and Cold)

Four bogies with spring suspension

Standard MLS10

4

610

5m x 1.6

Single/Dual

30 - 75

Pneumatic Hydraulic

7.2

Linear Induction Motors

(LIM)

Electric - On board generator

140 max

Heating feasible

Four bogies with pneumatic hydraulic suspension

Digital (Electrical – delete) Digital

Mechanical Pneumatic Hydraulic ton 0.7

L*B*

H

Flexible per design

2.5 x 0.7 x 1.2

29

Flexible per design

10.7 x 2.4 x 3.1

(can raise 1 meter for access below)

4 4 dual / super single

Vredestein 6 ply pneumatic kPa 400 - 800

Kph 8.7 simulating 26

Continental R22.5 295/65

; super single 385/65

500 - 1000

26 mm 10

7200

50

7200 max 6000 operational

Dry, heated, wet heated Dry/Wet Heated mm 80 - 240 610 – 1 600 (optional) mm

Gaussian or channelised

+/- 80

Screw jacks

Electric motor

Hand drawn

Trailer

Gaussian or channelised

Programmable +/- 500

Automated hydraulic jacks

Automated hydraulic jacks

Hydraulic motor

Low bed trailer/container

823

Table2: MMLS3 test results with laboratory and field trafficking

MATERIAL

TYPE

Red

STABILISING

AGENT

SURF-

ACING

3 % CEM II 32.5 HMA

AXLES TO

FAILURE

150 000

DISTRESS MECHANISMS

·Pumping >>Crushing

Red

Red

Red

3 % CEM II 32.5

7 % CEM II 32.5

+ 1.5 % Lime

(ILC)

3.5% CEM II

32.5 + 3.5 % Lime

+1.5 % ILC

SEAL

HMA

HMA

·Horizontal & vertical shear

·Flexural cracking

3 100 000

·Interface distress

145 000 ·Shear followed by bottomup cracking to neutral axis

·Local distress induced by artificial transverse cracking

3 100 000

·Longitudinal flexural cracks

(longitudinal growth)

·Transverse cracks

Red

Red

7 % CEM II 32.5+

1.5 % ILC

5% CEM II 32.5

SEAL

HMA

·Longitudinal flexural cracks

3 500 000 ·No apparent distress @

2.9 m

Red

Yellow

Red

Drilled cores reconstituted

Red Field

2.5 % CEM II

32.5 + 2.5% Lime

7 % CEM II 32.5

+

HMA

HMA

5 % CEM II 32.5 SEAL

1 300 000 ·Long and transverse cracking ·Ageing had an important effect

1 400 000 ·Longitudinal flexural crack

·Secondary transverse cracking

2 400 000 ·No apparent distress

125 000 Longitudinal flexural cracking - Interface distress

Red Field

Red Field

Red Field

Red Field

1.5 % CEM II

32.5 + 4.0% SS60

5.5% SS60

1.5 % CEM II

32.5 + 4.0% SS60

5.5% SS60

SEAL

SEAL

SEAL

SEAL

200 000*

Field test

5 % CEM II 32.5 SEAL 1 700 000*

Field test

·Rutting ·Secondary cracking

19 000*

Field test

200 000*

Field test

·Rutting ·Seal failed due to debonding and interface distress

·Rutting ·Secondary cracking

80 000* ·Rutting ·Seal failed due to debonding and interface distress

·No debonding

824

UNDERSTANDING THE MANIFESTED DISTRESS MECHANISMS UNDER MMLS3

TRAFFICKING

The performance of the cement-lime blend was in general better than the plain Portland cement in terms of MMLS3 trafficking and loss- of-stiffness. Distress mechanisms were explored by dissecting extracted specimens from the pavement. Crack patterns and rupture planes were mapped. Micro-cracks, were tracked as well as the manner and sequence in which distress occurred. The performance was found from the tests in the laboratory and in the field:

• Surface distress under the asphalt surfacing is evident. The distress is aggravated when water ingresses during trafficking. The extent of damage is related to the extent to which water can infiltrate. Stiffness ratios (relating the stiffness after trafficking to that before) as low as 0.50 were found where the surface of the structure had been pulverised and sand was being pumped out during trafficking.

• Horizontal shear at the interface between the asphalt surfacing and the underlying

CTB occurs during trafficking of thin asphaltic surfacing (seal and HMA) when the

CTB has a low strength (3 % cement content). This leads to delamination at spots resulting in the erosion of the underlying CTB. Sections containing 5% cement content maintained the bond better than the 3 % sections.

• Pumping of the CTB material occurs when the CTB surface becomes pulverised or crushed. The pumping appears to be initiated by water infiltrating through cracks in the asphalt.

• Longitudinal cracking in the CTB was found in the scaled pavements and in the fullscale test pavements (as well as the cores that were tested in the test bed). This phenomenon was also observed on old and new sections of highway EN1, where such cracks had migrated to the pavement surface.

• The P-SPA was successful in monitoring degradation of the pavement structure. It yielded stiffness results in terms of the ratio between trafficked and untrafficked response that reflected the state of the in-situ pavement. As distress and failure increased the ratio progressively reduced.

• The worst damage appeared in the locations with the highest in-situ moisture content, evidence of the major impact of water ingress.

• Surface rutting appeared to be related to densification of the asphalt and pulverising and/or crushing of the CTB. No densification of the sand subbase subgrade was evident.

• The 7% yellow sand performed very much the same as the red sand under trafficking.

In summary the degradation process consisted of the follow mechanisms:

♦ Bottom up longitudinal cracking

♦ Transverse cracking

♦ Horizontal shear plane formation in the middle of the combined CTB and HMA pavement structure

♦ CTB/HMA interface distress in the low cement content sections.

825

The order and direction of development of cracks was dependent on structural composition, material properties and construction history. A comparison of MMLS3 failure mechanisms with those found under MLS10 trafficking is discussed later. It is note worthy that the trends in relative seismic stiffness ratio under MMLS3 and MLS10 trafficking in the laboratory and the in the field respectively are similar in nature (see Figure 2).

1.1

1

0.9

0.8

1.1

1

0.9

0.8

0.7

0.6

0.5

0.7

0.6

0.5

R

2

= 0.589

2

= 0.589

0.4

0.3

0.2

0.1

0

0

0.4

0.3

0.2

0.1

200

0

0

400

200

R

2

= 0.820

R

2

= 0.820

600

400

AXLE LOAD IN 1 000's

600 800

1000

MLS10 70 kN 800 kPa

MLS10 70 kN 800 kPa

Expon. (MMLS3 2.7 kN

Expon. (MMLS3 2.7 kN

700 kPa)

Expon. (MLS10 70 kN

800 kPa)

1200

800 kPa)

1200

Figure 2: Relative stiffness ratios for MMLS3 and MLS10 – 7% CTB in lab and field

SUPPLEMENTARY AND VERIFICATION TESTING (S/VT)

Testing comprised exploration of physical, chemical and strength charactistics.

Physical and chemical characteristics of sands

Engineering properties normally determined on site or central laboratories could not identify a significant difference in the properties of reddish or yellowish coloured sands.

Electron microscope analyses were conducted by Energy Dispersive Spectroscopy (EDS) in the geology department of the University of Stellenbosch. The photos were taken with back scatter detector which shows differences in chemistry in shades of grey on polished slide sections. Differences in coatings can be seen.

The findings yielded a number of differences in the characteristics of the two sands.

1. Particle sizes differed: a. Yellow sand - single sized and larger particles b. Red sand - smaller in size with some apparent grading

2. Surface texture differed: a. Yellow sand – smoother with less of a surface layer b. Red sand – textured and covered with a surface deposit

3. Chemical analysis a. Yellow sand – significantly higher percentage of Si (double) b. Red sand – slighter higher percentage of Fe and Al

826

Two comparative photos taken during the investigation are shown in Figures 3 a and b.

Chemical composition of the layers of surface material on the sand particles was determined by comparing grey scales of polished discs using the backscatter detector.

Yellow Red

(a) 200μm (b) 200μm

Figure 3: Electron microscope photos of yellow sand vs. red sand

During construction of the scaled pavement it was observed that the red sand dried and drained very slowly after being wetted to above optimum moisture content. It retained the moisture for an extended period of time. Soil suction was also high. These phenomena caused the sand to be water sensitive impacting on constructability of the pavement.

Wet, dry sieve and hydrometer analyses as well as particle size distributions from the sieve analysis were performed. The results in Table 3 indicate that the silt content of the red sand is five times more than the yellow sand (6.8 vs. 1.3). Furthermore the percentage passing 0.075 is three times more for the red sand. This confirms the results reported from the field testing by the contractor.

Maximum Dry Density (MDD) of the red and yellow sand were reported from construction records to be in the ranges of 1920 to 1950 kg/m3 and 1820 to 1850 kg/m3 respectively. It was concluded that these differences in the materials were the cause for the difference in strength values found with the two types of sand.

Table 3: Summary of hydrometer tests results

Fraction Name Red Sand

%

Fraction Size

2.0 mm > % >

0.425 mm

0.425 mm > %

0.05 mm > % >

0.005 mm

Coarse Sand

Fine Sand

> 0.05 mm

0.05 mm > % > Clay

0.005 mm

0.005 mm > % Silt + Clay

% Passing 0.075

15.3

72.8

5.1

11.9

14.0

Yellow Sand %

61.8

32.9

4.0

5.3

5.3

827

Strength and related tests

Extensive testing was done in Stellenbosch (Masondo, 2005) (de Vos 2007). The tests included the following:

• Indirect Tensile Strength (ITS)

• Unconfined Compression Strength (UCS)

• Shear Split Test (SST)

• Semi-circular bending strength (SCB) [subsequently discont]

• Three point bending [subsequently discont]

• Tensile

ε b

• Insitu Dynamic cone penetrometer measurements (DCP)

Tests were conducted on laboratory compacted briquettes. Cored specimens (100 mm diameter) from the full scale and laboratory scaled test sections were also tested to determine the one and a half year strength of in-situ pavements constructed and cured according to current construction practice. Extracts from the results are contained in Tables

4, 5 and 6. Supplementary to the above tests the change in stiffness with increased number of load applications was measured to monitor structural performance. The

Portable Seismic Pavement Analyser (P-SPA) was utilized for this purpose.

The shear split test was incorporated to assess material shear strength since interface distress between the HMA and CTB occurred in sections tested (see later). The shear split test comprises vertical splitting along the diameter of a 60 by 100 mm diameter specimen or core at a displacement rate of 50.8 mm per minute while monitoring the load applied.

The shear split test method was developed by Lorio (1993, 1997). The testing rig was modified and methodology adapted for use in the Materials Testing Machine (MTS), the test setup is illustrated in Figure 4. The following discussion relates to the strength and related test results.

Figure 4: Shear split test setup (side and frontal view)

828

Table 4: Summary of material test results conducted on laboratory prepared specimens

SAND

TYPE BINDER

ITS ITS UCS SHEAR

7 DAY 28 DAY 28 DAY 28 DAY

Red

Red

Red

3 % CEM

5 % CEM

7 % CEM

200 8 290

310 11 540

15 1500 23 270

8 2220 14 540

9

3

350 15 550 16 2500 25 650 11

Red

Red

2.5 % CEM 2.5

% Lime

3.5 % CEM 3.5

% Lime

130 15 200 13 880 21 400 1

210 13 450 15 1490 23 470 12

Yellow 7 % CEM 390 18 870 5 3600 8

*COV - Coefficient of Variation = Standard Deviation / Average x 100

690 11

Table 5: Summary of test results obtained from tested full-scale field test section cores at

an age of 1.5 years

SAND

TYPE

BINDER

TYPE 2 TYPE 1

Red

Red

Red

Red

Red

Yellow

3 % CEM

5 % CEM

7 % CEM kPa kPa kPa kPa

-

360

680

2.5 % CEM 2.5 % Lime 938

3.5 % CEM 3.5 % Lime

7 % CEM

-

1094

-

596

-

-

-

-

-

4879

5190

4720

-

-

270

360

-

-

3866 370

From comparison with SAMPD standards (TRH 14, 1985) the CTB material conforms to

C2 and C3 material class standards according to ITS and UCS results respectively.

However, it should be noted that the reference standard relates to G5 or G6 material as

CTB material which is of course different in structure and composition from the

Mozambiquen sand.

Table 6: Summary of ITS test results of cores taken from scaled laboratory test sections at

an age of 1.5 years

SAND

TYPE

Red

BINDER

HMA SEAL

Red

Red

Red

3 % CEM

5 % CEM

7 % CEM

-

350

-

10

-

-

1700 15 1000

2.5 % CEM 2.5 %

Lime 550

-

-

17

3.5 % CEM 3.5 %

Red

Yellow 7 % CEM 530 2 - -

Tensile and shear strengths obtained from laboratory compacted and specimens cored from full-scale and scaled test sections do not correlate well. This was attributed to differences in the mixture quality, compaction and curing methodology and environmental circumstances. Field sections were subjected to environmentally induced distress and traffic loading of paving and surface compaction machinery during early life strength gain.

829

Compressive strength results compare favourably by indicating increased compressive capacity with ageing. It would appear that the compressive strength of the field CTB is less susceptible to early age damage caused by heavy loading of the construction equipment than the shear and tensile characteristics. The results further indicate that the retarded initial trength gain of the cement – lime blends relative to the cemented material increased with time to surpass the materials stabilized with cement only. The above results compare favorably with other reported studies (Melis et al. 1978 and Croney and Croney, 1991).

ITS strengths of lime stabilized granular pavement materials were studied in-depth by

Melis et al (1978). One of the outcomes was the relationship between UCS, Modulus of

Rupture (MR) and ITS It was concluded that MR was about 50 percent greater than ITS strength. This correlation was then applied to the ITS values measured on the

Mozambican sands, with a further correction to allow for ageing after research on stabilized materials by Croney and Croney (1991) [ Figure 13.3]. The results were subsequently used to determine the ratio of tensile stress under loading relative to tensile strength at failure to adjudicate fatigue performance [Figure 13.9 Croney and Croney

(1991)].

CONSTRUCTION OF FULL-SCALE FIELD TEST SECTIONS

The pavement structure comprised insitu sandy subgrade, 150 mm imported red sand subbase, 150 mm CTB and a surfacing of HMA or a double seal. Two different methodologies for compaction of the CTB were specified:

♦ Type 1 : Compation by vibrating padfoot roller for six passes followed by pneumatic tyre rolling for four hours. Compation had to be completed six hours after mixing of the stabilising agent had started.

♦ Type 2 : Compation by vibrating padfoot roller for six passes followed by pneumatic tyre rolling for six passes only. Compation had to be completed three hours after mixing of the stabilising agent had started.

Construction Type 1 is conventionally used by contrators to attain the high design densities prescibed – 97 percent of Modified AASHTO (MAASTHO). This requires a relatively long compaction period resulting in the break down of initial cementing action by the heavy construction equipment. Over compaction and destruction of the surface of the base layer occurs with some rollers when the number of passes exceeds a critical level. From construction experiments the steel drum roller caused surface shear while the pneumantic tyre roller has a kneading action on the fresh CTB material and smooth surface finish.

Figure 5 depicts the test arena adjacent to the existing rehabilitated highway EN1. The results of the respective laboratory tests for material characterization of the respective test sections are shown in Table 7. Construction was completed in June 2005.

The protocol for full-scale trafficking with the MLS10, was based on the performance of the respective MMLS3 test sections. Maximum tensile stress at the underside of the scaled

CTB pavement was calculated and compared to the maximum tensile stress under the fullscale pavement. The expected full-scale trafficking yielding similar distress, was then calculated on the basis of a “fourth power law” of 4.2. Material characteristics and test conditions were assumed similar. For example, the 5% CTB the MMLS3 trafficking was terminated at 1.3 million load applications and the expected full-scale performance was found to be 100 000 using 60kN loads. For the 7 percent CTB the expected full-scale yielded 230 000 60 kN loads. It was concluded that 1 million 60 kN load applications would

830

be sufficient for realistic diagnostic analyses. For environmental conditioning of the full scale tests, trafficking was alternated between 200 000 ambient dry and 50 000 wet load cycles with water sprayed onto the pavement surface (de Vos, 2004). Channelized loading was applied with no lateral wander.

FULL-SCALE AND SCALED FIELD APT BY MEANS OF MLS10 AND MMLS3

During this phase of the study the following data was collected:

♦ Dynamic surface deflections during trafficking using an extended Benkelman Beam in lieu of Falling Weight Deflectometer measurements

♦ Seismic stiffness using the PSPA

♦ Surface deformation by means of a electronic profilometer

♦ Surface cracking and related distress through photos, diagnostic, trenching and coring

♦ Diagnostic trenching after completion of traffic loading in conjunction with coring

Figure 5 Plan view of Test Sections constructed in June 2005

EVALUATION OF PAVEMENT PERFORMANCE TO ESTIMATE THE FULL-SCALE

PERFORMANCE CHARACTERISTICS OF THE TEST SECTIONS.

Pavement behaviour at different stages during performance phases under trafficking were similar for the tested sections. The pavement surfaces started to roll dynamically in wave form under loading tyres, with deflections visible with the naked eye when debonding occurred. Six comprehensive MLS10 tests were completed. The details are shown in Table7.

831

832

Table 8: Summary of MLS10 Tests

Section 8a 4a 5a 5b 8b 4b 8c 7

Stabilizing

Binder

Sand for CTB

(150mm)

Surfacing

(40mm)

Axle load X

1000 applications

7 %

Cem

Red Red Red Red Red Red Red Yellow

HMA HMA HMA HMA HMA HMA HMA HMA

50

5 %

Cem

330

2.5%

C

2.5%

L

100

2.5%

C

2.5%

L

1,080

7 %

Cem

84,5

5 %

Cem

150

7 %

Cem

730

7 %

Cem

1 050

Axle Load kN

60 60 60 60 60 70 70 70

833

MONITORING OF RESPONSE AND PERFORMANCE DURING FIELD APT

Dynamic deflection performance

The dynamic defections measured with the MBB were used in lieu of FWD tests.

Geometric features of the deflection bowl were determined relative to FWD norms.

Results for 70 kN are shown in Figure 6d. If deflection bowls are transposed to

40kN wheel loads the traffic life is likely to extend well beyond 10 million axles

(TRH12 (1986). This compares favourably with the other performance predictions that were established. TRH12 (1986) is of course based on granular and not sandy materials.

6

TIME IN s

8

1.4

1.6

1

1.2

0.6

0.8

0.2

0.4

0

0 2 4 10 12 14

0

30000

50000

100000

(a) MBB set-up under the MLS10

3

2.5

2

1.5

1

0.5

R

2

= 0.984

R

2

= 0.998

(b)Deflection change as applications increase

R

2

= 0.996

R

2

= 0.994

R

2

= 0.995

SECTION 4B *

SECTION 5A

SECTION 5B

SECTION 8 *

SECTION 7 *

Power (SECTION 5B )

Power (SECTION 8 *)

Power (SECTION 7 * )

Power (SECTION 5A )

0

0 200 400 600 800 1000 1200

APPLIED AXLE LOADS IN 1000's

(c) Peak deflections for different pavement sections as load applications increase

0

0 200 400 600 800

OFF-SET in mm

1000 1200 1400 1600 1800 2000

100

200

300

400

5 % CEM RED 4B

7 % CEM RED 8

7 % CEM YELLOW 7

500

600

700

(d) Deflection bowls of sections 4B, 7 and 8 under 70kN wheel loads

Figure 6 Graphic depiction of dynamic deflection results under moving wheel loads

834

Seismic stiffness changes during APT using P SPA measurements

Seismic stiffness monitoring was conducted with the PSPA device developed by

Nazarian et al. (1993). The operating principle of the PSPA is based on generating and detecting stress waves in a medium. The Ultrasonic Surface Wave (USW) interpretation method (Nazarian et al., 1993) is used to determine the modulus of the material. Surface waves (or Rayleigh, R-wave) contain two-thirds of the seismic energy. Accordingly, the most dominant arrivals are related to the surface waves making them the easiest to measure. This method utilizes the surface wave energy to determine the variation in surface wave velocity (modulus) with wavelength (depth). The variability of tests results with PSPA is less than 3% without moving the device and around 7% when the device is moved in a small area (Celayn et al. 2006).

The surface wave velocity, VR, is converted to modulus, E, using:

E= 2 [ρ (1.13 – 0.16)V

R

Where:

]

2

(1 + ν

E = modulus

V

R

= velocity of surface waves

ρ = mass density

ν = Poisson’s ratio

EFFECTIVE SEISMIC STIFFNESS GPa

8 9 10 11 12 13 14 15

0.05

0.06

0.07

0.08

0.09

0.1

0.11

0.01

0.02

0.03

0.04

0

0

0.12

0.13

0.14

0.15

0.16

0.17

0.18

1 2 3 4 5 6 7 16 17 18 19 20 21 22

0

15000

50000

100000

0.19

0.2

Figure 7 Typical seismic stiffness vertical profiles with axle load application

The stiffness of the pavement structure and more particularly the upper 200mm was measured at regular intervals during trafficking. Measurements were taken in longitudinal and transverse directions, relative to the wheel path. Figure 7 shows desperion curves with increase in axle loading to illustrate the relative reduction in the shear wave speeds, seismic modulus, with increased pavement distress. In general the change in longitudinal stiffness was more than that in the transverse direction.

LAYERED EXTRACTION OF PSPA RESULTS

From the initial MML3 laboratory testing different mechanisims of failure occured in certain particular pavement zones. Interface distress in the form of shear failure between the surfacing and the CTB occured on the top of the CTB. Horisontal shear planes were found in the middle of the CTB layer. Flexural cracking in the longitudinal and transverse directions starts at the bottomof the layer and progress with traffic loading to the top. The bottom of the CTB crushes into smaller pieces when pavement life nears its end. Diagnostic trenching after completion of traffic loading confirmed this phenomenon (see Figure 12a under synthesis discussion).

835

Subsequently wavespeed – frequency dispersion curves were analysed accordingly. The CTB layer were partitioned into thee zones the Top, Middle and

Bottom comprising of the top 25 mm, 50mm and 75 mm of the 150 mm CTB layer.

The stiffness reduction methodology was developed by the de Vos (2007) for the characterization of specific pavement zones based on their respective structural performances. The dispersion curves were reduced by averaging of the datapoints occuring in or on the boundry of each respective zone, Top, Middle and Bottom.

Direct ratios were calculated between initial untrafficked and measurements taken as traffic accumulated for each zone at each measurement location in each of the two directions of measurement.

PAVEMENT STIFFNESS PERFORMANCE UNDER FULL-SCALE

TRAFFICKING

In general the change in longitudinal stiffness was more than that in the transverse direction. Figure 8a – d illustrates the stiffness performance. Figure 8 (c) illustrates the performance of the bottom zone of CTB at the worst performing position of each tested section in the longitudinal direction. Performance of this zone and in this direction was the worst.

1.3

1.2

1.1

1

0.9

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

R

2

= 0.57

R

2

= 0.68

R

2

= 0.41

R

2

= 0.65

R

2

= 0.56

7 % CEM 70 kN

5 % CEM 70 kN

2.5% CEM 2.5% LIME 60 kN

2.5% CEM 2.5% LIME 60 kN

5 % CEM 60 kN

Expon. (7 % CEM 70 kN )

Expon. (2.5% CEM 2.5% LIME 60

Expon. (5 % CEM 70 kN)

Expon (5 % CEM 60 kN)

0

0 100 200 300 400 500 600 700 800 900 1000 1100

AXLE LOADS APPLIED in 1000's

(a) Comparative longitudinal stiffness performance top of

CTB

1.2

1.1

1

0.9

0.8

7 % CEM 70 kN

5 % CEM 70 kN

2.5% CEM 2.5% LIME 60 kN

2.5% CEM 2.5% LIME 60 kN

5 % CEM 60 kN

Expon. (7 % CEM 70 kN )

Expon. (2.5% CEM 2.5% LIME 60

R

2

= 0.77

0.7

R

2

= 0.78

Expon. (5 % CEM 60 kN)

0.6

R

2

= 0.62

0.5

0.4

0.3

0.2

0.1

R

2

= 0.95

R

2

= 0.82

0

0 100 200 300 400 500 600 700 800 900 1000 1100

AXLE LOADS APPLIED in 1000's

(b) Comparative longitudinal stiffness performance bottom of CTB

836

1.6

1.5

1.4

1.3

1.2

1.1

1

0.9

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

R

2

= 0.95

R

2

= 1.00

R

2

= 0.20

R

2

= 0.75

R

2

= 0.05

7 % CEM 70 kN

5 % CEM 70 kN

2.5% CEM 2.5% LIME 60 kN

2.5% CEM 2.5% LIME 60 kN

5 % CEM 60 kN

Expon. (7 % CEM 70 kN )

Expon. (2.5% CEM 2.5% LIME 60 kN)

Expon. (5 % CEM 70 kN)

Expon (5 % CEM 60 kN)

0

0 100 200 300 400 500 600 700 800 900 1000 1100

AXLE LOADS APPLIED in 1000's

(c) Comparative transverse stiffness performance for top of CTB

1.4

1.3

1.2

1.1

1

R

2

= 0.26

7 % CEM 70 kN

5 % CEM 70 kN

2.5% CEM 2.5% LIME 60 kN

2.5% CEM 2.5% LIME 60 kN

5 % CEM 60 kN

Expon. (7 % CEM 70 kN )

Expon. (2.5% CEM 2.5% LIME 60

0.9

0.8

R

2

= 0.00

Expon (5 % CEM 60 kN)

0.7

0.6

0.5

R

2

= 0.83

0.4

0.3

R

2

= 0.69

0.2

0.1

R

2

= 0.85

0

0 100 200 300 400 500 600 700 800 900 1000 1100

AXLE LOADS APPLIED in 1000's

(d) Comparative transverse stiffness performance bottom of CTB

Figure 8: Extracts from longitudinal and transverse PSPA stiffness results

Surface deformation

Rutting of the surface was limited and attributed to the interface distress and pulverizing of the CTB as well as the removal of the loose material pumped out through the surface cracks. Where failure occurred in the layer, the surface naturally deformed much more.

Surface crack formation

Cracking was monitored intermittently by visual inspection and photographic recording. Typical isometric views to illustrate crack growth phenomenon of

Section 4A are shown in Figure 9. Note pumping and longitudinal cracks in left wheel track at 320k load applications.

Cracks diagonal to and on the outer edges of the wheel track and transverse cracks in-between the wheels were first to appear. Diagonal cracks resulted from longitudinal in-plane shoving of the HMA through in-plane horizontal shear forces and subsequent interface distress between the HMA and CTB. With increased traffic loading these cracks propagate across the wheel track in a direction perpendicular to it connecting with the transverse cracks formed in-between the wheels. This resulted in a complete transverse surface crack. Longitudinal cracks would start to form under the tyres interconnecting with the transverse cracks to form a crocodile cracking pattern.

837

(a)Transverse shear crack initiation at

270 000 60 kN load applications -

Section 4

(b)Transverse propagation of shear cracks at 285 000 60 kN load applications - Section 4

(c) Continuous transverse cracks and longitudinal cracks after 320 000 60 kN

(d) Plan view of cracking pattern after

1 million 70 kN load applications-

Section 4 Section 7

Figure 9 Extracts from photographic records of crack formation under trafficking

Diagnostic evaluation through trenching

Diagnostic trenches 1000mm x 500 mm were made across selected wheel paths after completion of trafficking to investigate distress and failure mechanisms.

Interface shear between the HMA and CTB layers took place under loading. Shear planes generally formed under the prime impregnated CTB surface. In-plane longitudinal displacement of the HMA was also recorded. Shear marks and smoothed shear surfaces were observed on both the HMA and CTB contact planes.

Longitudinal and transverse flexural cracks were observed. The pavement structure was fractured when the stiffness had reduced by 50 percent. It was also apparent that the fractured CTB blocks had experienced horizontal shoving of the

HMA as trafficking progressed. From the test conducted on Sections 8 and 7 it was evident that the structure was more fractured when the stiffness ratio had significantly dropped to below 0.5 e.g. 0.2 – 0.3. With increased traffic loading the fracturing of the CTB progressively increased, reducing the particle size to smaller fractions with the smallest particles at the bottom.

838

Horizontal shear planes were found about one third from the top of the CTB. In the lower cement stabilized bases trafficked up to a stiffness loss of 50 percent, these shear planes were found to connect the bottom up longitudinal cracks at this depth.

In the bases trafficked to a more severe state of distress it was found that the shear plane had propagated laterally in the CTB structure.

Section 8 was trafficked to a stiffness ratio as low as 0.2 when the pavement had deteriorated dramatically. Deep rutting occurred, severe pumping took place. The particle fractionation and loss of material through pumping lead to the deep rutting as well as shear failure of the asphalt on the edges of the wheel track.

It was found that two types of distress could develop simultaneously on the same pavement section. This was the case in Section 8. At the far end of the section the distress was as three types discussed above. In contrast, at the start of the section the CTB had not shattered, but the HMA had deformed due to shoving of the asphalt material longitudinally as well as transverse shear adjacent to the wheel.

In all cases the distress in the bottom of the CTB was similar in relation to the relative reduction in stiffness in both extent and orientation. It is of course important to remember that trafficking had been done without lateral wander. The similarity of the distress between the full-scale and the laboratory scaled pavements is apparent from the performance of the sections. Section 7 proved to be the best in terms of performance against initial expectations. The findings are also in accordance with the laboratory S/V results.

(a) Underside of HMA - section 8 (b) Top of CTB – section 8

Figure 10 Extract from photographic records of diagnostic findings through trenching

SYNTHESIS OF SCALED AND FULL-SCALE APT PERFORMANCE FINDINGS

Manifestation of distress followed a pattern that had been identified in MMLS3 laboratory study on scaled pavements ( de Vos et al. 2007). The postulated failure mechanisms are based on the findings presented above.

Interface distress between HMA and CTB contact surfaces

Asphalt surfacing exhibited in-plane horizontal longitudinal displacement at the interface. This resulted in diagonal shear fractures (cracking) on the sides of the wheel path. This phenomenon is clearly related to shear bond between asphalt surfacing and CTB. Continued traffic loading leads to pulverizing and distress

839

aggravates when water infiltrates during load trafficking. Pulverized material migrates through cracks by means of pumping action. Initial surface rutting relates to removed pulverized interface material. Diagnostic trenching exhibited clear evidence of this distress. Some typical details are depicted in Figure 10 a and b. In the later stages of pavement life the HMA shears off completely and the whole pavement experiences a dynamic bow wave under trafficking loads. The bow wave was so active that it was visible to the naked eye.

Longitudinal and transverse flexural cracking

Bottom-up transverse and longitudinal flexural cracking of the pavement was the primary mechanism of CTB structural failure. See Figure 12a and b. Reductions in stiffness ratios were found to be greatest in the bottom pavement zone for all pavements.

Shear plane formation in the CTB

A secondary structural mechanism of failure is the occurrence of a horizontal shear plane in the CTB. This plane was generally found to occur in the middle of the combined HMA-CTB structure (100 mm from the top). This position is also one third of the CTB thickness from the top thereof (50 mm). This plane links the two longitudinal cracks and forms a ‘shear box’. (see Figures 12a and b. A similar mechanism was found with the scaled laboratory testing of this material (de Vos et al. 2007).

A linear elastic shear stress analysis was done with Bisar 3.0 (Shell, 1998) to simulate and investigate this phenomenon. Two pavements model were analyzed for evaluation of the shear stress distribution in the pavement before and after interface distress. The pavement structure representing the pavement before interface distress had a full friction interface between the HMA and CTB. The second pavement model had a 10 mm layer in between the HMA and CTB with the same attributes than the sub base, to simulate the sheared and pulverized interface. Shear stress pavement profiles for the ‘Bonded’ and ‘Unbonded’ scenarios are illustrated in Figure11. The plots illustrate that maximum shear stress is initially at the HMA-CTB interface. After failure thereof, the maximum shear stress is found in the middle of the combined pavement structure, 100 to 125 mm from the top, hence formation of the horizontal in-plane shear crack. Table 9 shows working stress to strength ratios for both the interface and shear plane mechanisms.

Table 9: Stress to strength ratios for the interface and shear plane failure mechanisms

Condition

5% / 60 kN

5% / 70 kN

7% / 70 kN

HMA - CTB Interface

Mechanisms

1.09

1.12

0.82

CTB Mid Layer Shear

Plane Mechanisms

0.90

0.97

0.75

840

50

100

150

0

0 25 50 75 100

SHEAR STRESS IN kPa

125 150 175 200 225 250 275 300 325

BONDED

UNBONDED

200

250

Figure 11: Shear stress plots for 5 percent CTB under 60 kN axle load

(a) Front view – section 4B- note horizontal shear plane and longitudinal cracks

(b) Plan view – section 4B- note longitudinal and transverse cracks

(c) Progressive phases of failure-scaled

3 percent CTB section.

(d) Shear plane on neutral axis of

50mm scaled 3 percent CTB with seal,

Figure 12 Photographic depictions of distress development under trafficking

841

From the illustrated shear stress plots and ratios presented, it was concluded that the interface distress occurs first followed by shear plane formation. The high stress to strength ratios for the 5 percent CTB sections further indicate why the interface distress initiates much earlier relative to the gradual stiffness loss at the bottom part of the base. A comparison between the 5 and 7 percent cement treated sections’ ratios indicate why the shear crack formation for the higher cemented sections start later in the pavement life ( after 50 percent base stiffness loss).

Diagnostic evaluation of Sections 4B, 8 and 7, that were tested with higher loads to reduce stiffness values as low as 30 to 20 percent of the initial, showed that these

CTB blocks were fractionized further into much smaller particles. At that stage of pavement life, severe lateral asphalt movement had taken place resulting in wide transverse and diagonal cracks (1 – 2 mm). Fine crushed material was pumped out; this resulted in vertical shear of asphalt as there was less supporting base material left beneath the wheel tracks. Subsequent significant increases in rutting were also observed under these conditions. Maximum surface deflections of 2.8 to

4 times that of the initial untrafficked deflections were measured. Relative increases in the deflection ratios are smaller from 50 to 80 percent stiffness loss as the pavement loses structural integrity and ability to respond elastically under loading. Figure 13 illustrates this phenomenon.

The various distress modes causing the disintegration of the pavement structure were found to occur in different sequences depending on the structure of the pavement and the strength of the materials. For the higher strength cemented bases the interface distress occurred later in pavement life relative to the base stiffness loss than the lower strength cement treated bases. This is ascribed to the stress strength ratio under traffic loading at the pavement interface.

1

0.9

0.8

0.7

1.3

1.2

1.1

0.6

0.5

0.4

0.3

1.6

1.5

1.4

R

2

= 0.99

R

2

= 0.77

1

0.9

0.8

0.7

1.3

1.2

1.1

0.6

0.5

0.4

0.3

1.6

1.5

1.4

STIFFNESS RATIO

DEFLECTION RATIO

Expon. (STIFFNESS RATIO)

Power (DEFLECTION RATIO)

0.2

0.1

0.2

0.1

0 0

1200 0 200 400 600 800 1000

AXLES LOADS IN 1000's

Figure 13 Stiffness – Deflection performance of 2.5 % cement and 2.5 % lime

stabilized red base material under 60 kN axle loading

842

CONCLUSIONS

The following noteworthy aspects of the performance of the CTB were concluded:

♦ Bottom-up transverse and longitudinal flexural cracking of the pavement was the primary mechanism of CTB structural failure.

♦ Horizontal shear planes manifest as secondary CTB structural mechanism of failure in the CTB.

♦ Distress due to delamination between surfacing and the CTB was found to relate to the shear stress and bond strength. Asphalt surfacing exhibited in-plane horizontal longitudinal displacement. This resulted in the observed diagonal shear fractures on the outer edges of the wheel path.

♦ Response and performance of full-scale pavement structures and model pavements in the laboratory were compatible with manifestation of similar failure mechanisms. Difference in surface distress was attributed to the difference in contact stresses.

♦ Pulverized material migrates through cracks by means of pumping action leading to distress of the surfacing.

♦ The axle loads of 60 kN and subsequently 70 kN, provided insight into distress mechanisms and information pertaining to performance.

♦ With the 7% CTB, de-bonding distress occurred after the PSPA stiffness had reached 0.5. This improved performance of the pavement.

ACKNOWLEDGEMENTS

The authors gratefully acknowledge permission to publish the findings from the research reported in this paper. It was done as an integral part of project

206/CON/ES/DEN/2003 for the ANE. Funding was sponsored by the World Bank.

The opinions expressed by the authors do not necessarily reflect those of the sponsors.

REFERENCES

Celayn, M, Nazarian, S 2006. Seismic Testing to Determine Quality of Hot-Mix

Asphalt, CD-Rom Proceedings 85 th

Annual TRB Meeting, Washington, D.C.

Croney, D and Croney, P 1991. The Design and Performance of Road Pavements;

McGraw-Hill International Series in Civil Engineering, 2 nd

Edition. de Vos, E R 2004. Development of a Accelerated Laboratory Testing Protocol for

Cement Stabilized Base Layers using the Third Scale Model Mobile Load

Simulator; B.Ing Thesis, University of Stellenbosch, November de Vos, E R 2007. Performance Characterizaion of Cement Treated Sand Base

Material of Mozambique, M.Ing Thesis, University of Stellenbosch, April

843

de Vos, E R, Hugo, F, Strauss, P J, Prozzi, J A, Fults, K W, Tayob, H 2007.

Comparative Scaled MMLS3 Tests vs. Full-Scale MLS10 Tests in Mozambique;

CD-Rom Proceedings, 86th Annual TRB Meeting; Washington D.C.

Lorio, R,1994. The Asphalt Shear Box Test, B.Eng Thesis, Univ Stellenbosch.

Masondo, PT, 2005. Testing Parameters for Cement Treated Base Courses;

Project submitted for fulfilment of BEng Degree; Faculty of Engineering, University of Stellenbosch; November

Melis, L M, Meyer, A H and Fowler, D W 1985. An Evaluation of Tensile Strength

Testing; Research Report 432-1F, Centre for Transportation Research, The

University of Texas at Austin, November.

Nazarian, S, Baker, M R, and Crain, K 1993. Fabrication and Testing of a Seismic

Pavement Analyzer; SHRP Report H-375. SHRP, National Research Council,

Washington, D.C.

Shell International Oil Products BV 1998. Bisar 3.0 (computer programme).

844

Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement