Mechanistic-empirical design of unpaved and surface ... · pavement samples appear to give...
Transcript of Mechanistic-empirical design of unpaved and surface ... · pavement samples appear to give...
Mechanistic-empirical design of unpaved and surface-treated roads
1
Mickael Le Vern, M.Sc. Guy Doré, Ph.D.
Jean-Pascal Bilodeau, Ph.D.
9th Annual Western Canada Pavement Workshop (2018)
Edmonton, AB
Background
2
More than 600 000 km of unpaved and surface-treated roads in Canada:
Vital links for remote communities and access to natural resources
Typically low-volume of traffic but often have to support very heavy loads
Background
Pavement structural design is mostly experience-based or empirical (AASHTO,…)
Mechanistic-empirical design is now the recommended practice in North-America
Unpaved and surface-treated roads get very little research attention
3
4
Design objectives
Allowable strains
Required thickness
Site specific conditions
M-E design approach
eadm.
Mechanistic calculation model
Log
ead
m
Log N
Empirical damage
function
5
Project objectives
To develop a mechanistic-empirical method for the design of unpaved and surface treated roads
To develop a calculation model for the determination of strains
To develop a damage function (Strain vs Number-of-cycles-to-failure relationship)
To integrate the method in a practical design tool
Methodology
1. Development of a simple calculation model (existing models adapted for unpaved roads)
2. Development of an empirical damage (rutting) function Establishment of allowable rut depths criterion (25 or 50mm
allowable rut depth at the top of the subgrade)
Measurement of the resilient and permanent vertical strain at the top of the subgrade during loading cycles
Development of a «Strain vs Number of cycles to failure» relationship
3. Preliminary stability analysis
6
7
Performance of unpaved roads
RUTTING
Post-compaction
Local shear close to the
tire
Structural rutting
(Dawson, 2007)
8
1. Mechanistic analysis
Boussinesq equations
Homogenous and isotropic soil Linear elastic behavior
Odemark transformation
ℎ𝑒 = 𝑓. ℎ1
𝐸1
𝐸2×
1 − 𝜇22
1 − 𝜇12
13
2. Development of an empirical damage function
• Very few information available in literature
• Development of damage function using accelerated loading
• Validation using field data from study in UK
9
10
Development of a failure criterion RUTTING AND WATER RETENTION
Rut Depth : 25mm Width : 1.5m
Cross slope Water retention
2% 10mm
3% 2.5mm
4% -
Rut Depth : 50mm Width : 1,5m
Cross slope Water retention
2% 35mm
3% 27.5mm
4% 20mm
(Glennon, 2015)
11
Development of damage function
Laboratory small-scale heavy vehicle simulator
12
Experimental setup and instrumentation
Four experimental test pavements using four subgrade soils - A silty sand (SM) - Silty clay (ML) - Low plasticity clay (CL) - Poorly graded sand (SP)
• Multistage loading approach • Deflectometer and LVDT for
displacement measurements • Variation of water table position • Thin asphalt concrete layer
_
13
Experimental setup and instrumentation
Internal pressure and moisture content of layers
14
Results
0
0.005
0.01
0.015
0.02
0.025
0.03
0 10 20 30 40 50
Ve
rtic
al s
trai
n (
mm
/mm
)
Cycle
Vertical strain in the subgarde soil, Saturated silty clay
Sensor measurement
Permanent deformation
15
Results
Example of permanent deformation accumulation in the subgrade (SP subgrade soil)
16
Results
Damage curve (ML subgrade)
17
Results
Resilient Strain
25mm or 50mm rutting prediction
(extrapolation)
y = 0.4793x-0.462 R² = 0.9971
0.0001
0.001
0.01
0.1
10 100 1000 10000 100000 1000000 10000000 100000000
Re
silie
nt
def
orm
atio
n in
du
ced
by
eac
h c
ycle
(m
m/m
m)
Allowable number of cycles
18
Results validation with field data
𝜖 = 0,07. 𝑁−0,346
𝜖 = 0,25. 𝑁−0,346
Field data: The Design of Unsurfaced Roads Using Geosynthetics (Little, 1992)
19
Example of application
𝜖 = 0,07. 𝑁−0,346
Allowable vertical strain: 0,002 mm/mm
N design = 20 000 ESAL
20
Design comparison between this method and the AASHTO method
21
Comparison with other damage functions for LVR
22
Development and comparison of design charts
Design chart developed in this project Design chart developed by Gupta (2014)
Comparison for a 25mm rut depth criterion and a 150MPa base resilient modulus
0
100
200
300
400
500
600
700
800
900
1000
10000 100000 1000000
Ba
se t
hic
kn
ess
(mm
)
Traffic (ESALs)
MR sol = 20MPa
MR sol = 30MPa
MR sol = 40MPa
MR sol = 50MPa
MR sol = 70MPa
MR sol = 90MPa
MR sol = 100MPa
MR sol = 150MPa
3. Preliminary stability analysis
• Preliminary assessment of failure risk under heavy loads
• Finite element analysis for different soils, pavement thickness and embankment geometry
• Development of an empirical model relating site conditions to safety factor
23
24
Preliminary safety analysis
Analysis with GeoSlope FE software
𝐹. 𝑆. =𝑅𝑒𝑠𝑖𝑠𝑡𝑖𝑛𝑔 𝑓𝑜𝑟𝑐𝑒𝑠
𝐺𝑟𝑎𝑣𝑖𝑡𝑦 𝑎𝑛𝑑 𝑙𝑜𝑎𝑑 𝑖𝑛𝑑𝑢𝑐𝑒𝑑 𝑠𝑡𝑟𝑒𝑠𝑠𝑒𝑠
MG-20
ARGILE
25
Stability analysis model
80 CASES ANALYZED
𝑭. 𝑺. = 𝟎, 𝟎𝟎𝟖𝟓𝟔. 𝑫 + −𝟎, 𝟎𝟎𝟎𝟑. 𝒍𝒏 𝑵 + 𝟎, 𝟎𝟎𝟎𝟑 . 𝒉2 +(𝟎, 𝟎𝟐𝟗𝟒. 𝑵 − 𝟎, 𝟎𝟕𝟖𝟕). 𝒉 − 𝟎, 𝟕𝟐𝟒𝟔. 𝑵 + 𝟑, 𝟓𝟎𝟏𝟗
Standard error between model and GeoSLOPE : 12%
26
Discussion
• The use of accelerated loading tests on laboratory scale pavement samples appear to give realistic results
• The mechanistic-empirical design approach proposed in this study can lead to cost savings regarding materials, in comparison with empirical design methods
• Seasonal damage analysis needs to be done in order to consider rapid deterioration in saturated conditions (spring)
27
Conclusion
A M-E design method for unpaved or sealed pavements has been developed
The M-E based approach makes the design procedure more versatile. It can easily be adapted to any design environment
The strain and performance results are in good agreement with field data and other proposed damage functions
When compared with other design methods, the design results are consistent
An empirical model has also been developed for the preliminary assessment of embankment stability under heavy loads
28
Next steps?
o Design charts have been developed
o The method could be integrated in a low-volume pavement design module in the mechanistic-empirical pavement design software I3C-ME
o The method could be developed as a stand-alone software
29