Академический Документы
Профессиональный Документы
Культура Документы
Azmi HASSAN
Director
Kelantan Public Work Department
15990 Kota Bharu, Kelantan.
Malaysia
E-mail: azmihassan@jkr.gov.my
Abstract: This paper looks into the development of a prototype Road Construction Material
Selection System (RC-MSS) which based on the outcome through interviewed to the
pavement experts to obtain the appropriate material for each pavement layer under particular
design conditions. Data collected from the experts would be analysis using nonparametric
statistical method to determine the ranking of each pavement material alternative prior to the
development of RC-MSS. RC-MSS is developed with a prototype of an expert system using
written using programming tool Microsoft Visual Basic 6.0 according the framework from the
system. With this tool, better road design can also be achieved and save more time in the
initial stage of on material selection, before scrutiny to laboratory test. This software would be
the pioneer and provide guidance to engineers and consultants.
1. INTRODUCTION
1313
Journal of the Eastern Asia Society for Transportation Studies, Vol. 6, pp. 1313 - 1328, 2005
pavement materials are severely over stressed. The type and soil conditions, and expected
load repetitions are other parameters found to be indispensable in the formulation of material
selection system for road construction. The material performance and the selection of the
appropriate construction materials for the various layers under certain traffic-loading under
the type of terrains such as flat, rolling and mountainous was incorporated as important
parameters. Such conditions govern the duration of axle loading on the pavement.
Pavement material selection should be based on road user-oriented, not only based on
industry-oriented process. There is a drive towards the development of more appropriate
material selection approaches as these will allow alternative solutions to be readily compared
and, thus, evaluated. A prerequisite for any successful analytical design methodology is the
reliable measurement of representative material properties.
The objective of this study is to develop a first expert system that could be used as a tool in
pavement material selection to help engineers in better decision making.
The system has look into the anticipated traffic loading, soil condition, type of terrain and
geometric features as aspects which should not be neglected in pavement material selection.
Commercial traffic loading is the main factor to any pavement design problem. It heavy axle
loads which cause pavement to damage and fail if poor pavement material is chosen.
Typically, the traffic magnitudes and configurations are expressed in terms of the number of
equivalent standard axles load (ESAL) (Lay 1990; JKR 1985) and equal to the number of
passes of the local standard axle load carried on a pair of dual tires which would cause the
same pavement damage as a single pass of axle load equivalent to 8.16 Tonne (Harold, 2003;
JKR, 1985). The system will incorporated the anticipated ESAL as another input factor which
classified it into three categories (high, moderate and low anticipated ESAL) as illustrated in
Table 1.
Soil condition is also another main factor to be considered in pavement design and material
selection. The properties of the soil are important not only as the foundation of pavement
structure but also crucial factor of cost in construction project. Poor material such as peat, clay
soil, highly organic soils and soils contaminated with hazardous chemicals would caused
serious damage to the pavement in the initial stage and extremely costly to reconstruction
(Harold, 2000). The strength of road subgrade is commonly assessed in terms of the
1314
Journal of the Eastern Asia Society for Transportation Studies, Vol. 6, pp. 1313 - 1328, 2005
California Bearing Ratio (CBR) and this is dependent on the type of soil, its density, and its
moisture content (Huang 1993; Lay 1990). Higher value of CBR would be suitable to use as
subgrade, CBR value which is less than 5 should considered as poor subgrade soil, due to
poor sustainable to heavy load. The system will use the CBR value input data to the system.
The CBR value is classified into three categories (good, moderate and poor) and each
category range is shows in Table 2.
Climbing lane and sharp horizontal curve where multiple heavily loaded vehicles firmly brake,
travel at slow speeds or remain standing for periods of time have caused the pavement to
deteriorate due to highly stresses on the pavement. Consequently, pavement exhibited rutting
and premature distresses due to the combination of truck traffic patterns, counts and weights
before the reaching the predicted design period (Harun, 1992). However, these factors are
only considered in safety purpose on geometric design and less concern in pavement material
by road designer. Hence, geometric feature on vertical gradient and degree of horizontal
curvature the system had been considered in the system and classified into two categories of
critical area and non-critical area as presented in Table 3.
Topography profile of road such as type of terrain is crucial in pavement design and it is
generally divided into three group, that are flat, rolling and mountainous which determine by
the percentage of cross slope as shown in Table 4 (JKR, 1986). The terrain types incorporated
in the system as determine type of material selection.
Type of Terrain
Flat Rolling Mountainous
Cross Slope ≤3% 3 – 25 % ≥ 25 %
Denoted Symbol T1 T2 T3
1315
Journal of the Eastern Asia Society for Transportation Studies, Vol. 6, pp. 1313 - 1328, 2005
Two stages of questionnaires had been carried out to obtain the opinions from the pavement
experts who specialize in road design or evaluation and familiar in pavement materials’
characteristics. Material suitable to use for nine site conditions (presented in Table 5) which
were the combination of anticipated traffic loading, soil, geometric feature and terrain type
were obtained. These factors were verified again from the experts. While, for poor soil CBR
value (2.5 – 5%) and very poor soil CBR (less than 2.5%), adopted measure to the condition
were selected and suggested by the experts. The suggested materials by the experts were then
rearranged and gather together prior to the second interview conducted. Second stage of
questionnaire, experts were required to rate the materials suggested for each site conditions
based on the material durability, performance and ease of construction using Likert scale
(very suitable=10 to less suitable=1). Adopted measure to poor soil condition was designated
as site condition 10th.
The non-numeric data were ranked as in ordinal type data using Friedman statistical test due
to several related materials were suitable to use (Conover 1999; Daniel 1978). This method
was used to determine the hypothesis test of identical performance of the materials under
particular design condition. was used as the hypothesis analysis tool in second stage. The
knowledge derived from outcome of the statistical method was then stored in the system
database. The material selected and ranking will be used as “knowledge base” and particular
design site condition will be use as the “inference engine” in the system. If-then will
incorporated in the rule-bases of the expert system, the materials and ranking database could
be determine from the design conditions matrix. The results generated by these rule-bases are
used to assist in decision making.
1316
Journal of the Eastern Asia Society for Transportation Studies, Vol. 6, pp. 1313 - 1328, 2005
No. Statement A B C D E
Pavement material selection should
1 considered geometric factor of vertical √ √ √ √ √
gradient.
Pavement material selection should
2 considered geometric factor of horizontal √ √ √ √ √
curvature.
Pavement material selection should
3 √ √ √ √ √
considered soil condition of CBR.
Pavement material selection should
4 considered traffic loading of designed √ √ √ √ √
ESAL.
Pavement material selection should
5 √ √ √ √ √
considered type of terrain.
Vertical gradient considered as critical area
6 √ √ √ √ √
when ≥ 5 %.
Horizontal curvature considered as critical
7 √ √ √ √ √
area when > 7°.
8 Low if designed ESAL when < 300,000. √ √ √ √ √
Moderate if designed ESAL when 300,000
9 √ √ √ √ √
– 10,000,000.
High if designed ESAL when >
10 √ √ √ √ √
10,000,000.
11 Good if soil CBR when > 10 %. √ √ √ √ √
12 Fair if soil CBR when 5 % – 10 %. √ √ √ √ √
13 Poor if soil CBR when 2 % – 5 %. X X X X √
14 Very poor if soil CBR when < 2 %. X X X X √
Note: √ : Pavement experts agree with the statement.
X: Pavement experts disagree with the statement.
From Table 6, all the five domain experts agree the statements 1 to 12. Only one expert agree
to the statements 13 and 14, however, four experts who did not agree with the statement
mentioned that the soil is considered poor whenever CBR value is less than 5 %, and
treatment must be done to that particular soil, no compromise to let the soil untreated. The
treatment selected and suggested by the experts were as followed.
• Chemically stabilized;
• Geosynthetics stabilized;
• Excavated and replaced stronger soil material;
• Dynamic compaction;
• Preloading or vertical drain cohesive.
The second stage of questionnaire results adopted Friedman statistical test method was
tabulated as shows in Table 7. The Table shows that null hypothesis at 95% significant level
for site condition 1st (rolling and mountainous terrain), 2nd (mountainous terrain), 6th (rolling
and mountainous terrain), 8th (mountainous terrain), 9th (rolling and mountainous terrain) and
10th (any terrain) were rejected. It may be concluded that there were a tendency for some
1317
Journal of the Eastern Asia Society for Transportation Studies, Vol. 6, pp. 1313 - 1328, 2005
Referred to F-
Site distribution Friedman
Terrain Type K k1 k2 Result
Condition Table at 0.95 Test
Quantiles
Flat 14 13 52 1.92 1.72 Accept Ho
1 Rolling 14 13 52 1.92 2.20 Reject Ho
Mountainous 14 13 52 1.92 2.20 Reject Ho
Flat 7 6 24 2.51 0.73 Accept Ho
2 Rolling 8 7 28 2.36 1.32 Accept Ho
Mountainous 8 7 28 2.36 3.02 Reject Ho
Flat 14 13 52 1.92 1.91 Accept Ho
3 Rolling 14 13 52 1.92 1.85 Accept Ho
Mountainous 14 13 52 1.92 1.06 Accept Ho
Flat 6 5 20 2.71 1.89 Accept Ho
4 Rolling 6 5 20 2.71 1.08 Accept Ho
Mountainous 7 6 24 2.51 0.73 Accept Ho
Flat 6 5 20 2.71 2.04 Accept Ho
5 Rolling 7 6 24 2.51 1.22 Accept Ho
Mountainous 7 6 24 2.51 0.73 Accept Ho
Flat 7 6 24 2.51 1.65 Accept Ho
6 Rolling 7 6 24 2.51 6.02 Reject Ho
Mountainous 7 6 24 2.51 4.51 Reject Ho
Flat 8 7 28 2.36 0.89 Accept Ho
7 Rolling 7 6 24 2.51 1.11 Accept Ho
Mountainous 7 6 24 2.51 0.86 Accept Ho
Flat 7 6 24 2.51 1.08 Accept Ho
8 Rolling 7 6 24 2.51 1.30 Accept Ho
Mountainous 7 6 24 2.51 8.67 Reject Ho
Flat 7 6 24 2.51 1.14 Accept Ho
9 Rolling 7 6 24 2.51 8.62 Reject Ho
Mountainous 7 6 24 2.51 0.82 Reject Ho
Flat 5 4 16 3.01 8.02 Reject Ho
10 Rolling 5 4 16 3.01 4.82 Reject Ho
Mountainous 5 4 16 3.01 5.54 Reject Ho
Tables 8 and 9 represented the ranking for each material at each pavement layer under site
condition 1st to 9th and site condition 10 respectively.
1318
Journal of the Eastern Asia Society for Transportation Studies, Vol. 6, pp. 1313 - 1328, 2005
M10
M11
M12
M4
M5
M6
M7
M8
M9
1 2 3 4 5 6 7 8 9
1319
Journal of the Eastern Asia Society for Transportation Studies, Vol. 6, pp. 1313 - 1328, 2005
Table 9. Material Ranked for Site Condition 10th (denoted symbol for ‘TSG’)
Terrain Type/
No Adopted measured to the existing soil Ranked
Denoted Symbol
1 Chemically stabilized 3
2 Geosynthetics stabilized 4
Flat, T1 3 Excavated and replaced stronger soil material 1
4 Dynamic compaction 5
5 Preloading/ vertical drain 2
1 Chemically stabilized 5
2 Geosynthetics stabilized 3
Rolling, T2 3 Excavated and replaced stronger soil material 1
4 Dynamic compaction 4
5 Preloading/ vertical drain 2
1320
Journal of the Eastern Asia Society for Transportation Studies, Vol. 6, pp. 1313 - 1328, 2005
1 Chemically stabilized 5
2 Geosynthetics stabilized 3
Mountainous, T3 3 Excavated and replaced stronger soil material 1
4 Dynamic compaction 4
5 Preloading/ vertical drain 2
5. RC-MSS FRAMEWORK
In the pavement material selection system (RC-MSS) the factors that considered in
developing RC-MSS is shown previous from Tables 1 to 4. In order to adopt RC-MSS, user
should provide the information of existing soil CBR value done by surveyor, anticipated
ESAL, geometric design feature for vertical gradient, degree of curvature and type of terrain.
Geometric features of horizontal curvature group and vertical gradient group were regrouped
to divide to non-critical area (assigned symbol as C1) and critical area (assigned symbol as C2)
as presented in Figure 1.
H1 H2 V1 V2
C1 C2
C
Note: Symbol ‘C’ Denoted as Critical Condition
Figure 1. Regrouping of Horizontal Curvature and Vertical Gradient Groups to Critical and
Non-Critical Group
Design conditions consisted of anticipated ESAL group (denoted as ‘E’), soil CBR group
(denoted as ‘S’) and critical area group (denoted as ‘C’) as shows in Figure 2. These design
conditions would then followed the path sorted from denoted symbol ‘E’, ‘S’, ‘C’ to ‘M’.
However, if design condition consisted of symbol ‘S3’ that is poor soil group. Then, the block
‘TSG’ that is adopted measure to poor soil would be the next station. The ‘modify’ design
condition after ‘TSG’ would considered equivalent to good soil group as denoted by symbol
‘S1’ before proceed to the ‘C’ group to reach Materials Suggested Database denoted as ‘M’, if
the previous design condition of symbol enter block ‘TSG’, the database to reach is Material
Suggested Database with asterisk (denoted symbol ‘M*’). Block symbol ‘TSG’ was extracted
from the result from Table 9. Since it was not significant to conclude that the adopted
measured for poor soil are more likely to perform better than others, hence the adopted
measured were not ranked. The database of ‘M*’ have the same suggested pavement material
1321
Journal of the Eastern Asia Society for Transportation Studies, Vol. 6, pp. 1313 - 1328, 2005
for subbase, base course and surface course as database of ‘M’ except the subgrade where
‘M*’ adopted the subgrade from block symbol ‘TSG’.
RC-MSS Framework
E1 E2 E3
S1 S2 S3 S1 S2 S3 S1 S2 S3
TSG
C1 C2 C1 C2 C1 C2
C1 C2
C1 C2 C1 C2
M8, M8*
M1, M1*
M1, M1*
M7, M7*
M4, M4*
M9, M9*
M10
M11
M12
M1
M5
M6
Note: Symbol ‘M’ Denoted as Materials Suggested Database, and * Represented Material
Suggested Database Directed from the Path from Denoted Symbol ‘TSG’.
6. PROBLEM EXAMPLE
A simple example has been run to illustrate the framework of RC-MSS. The design conditions
are as followed.
Example problem:
• Anticipated ESAL = 500,000
• Soil CBR value, % =3
• Vertical gradient, % =3
• Horizontal curvature, ° =8
• Type of terrain = Flat
Every particular from the design condition has been classified to each group as shows in
Table 10. The geometric feature of vertical gradient was non-critical but horizontal curvature
was critical. As mentioned earlier in Figure 1, once either one of the geometric features of
vertical gradient or horizontal curvature is in critical that is ‘V2’ or ‘C2’, it would be
considered as critical region. So, the example has been classified as symbol ‘C2’ of critical
area. However, the soil condition was classification as poor which denoted as symbol ‘S3’. As
1322
Journal of the Eastern Asia Society for Transportation Studies, Vol. 6, pp. 1313 - 1328, 2005
discussed in the previous section RC-MSS framework, the design would enter the station
denoted symbol ‘TSG’ then exit to station ‘S1’ before reaching the 8th Asterisk Material
Suggested Database denoted as symbol ‘M8*’. The whole flow diagram is illustrated in
Figure 2 which given the ticker plotted line. In the Asterisk Material Suggestion Database,
database of pavement materials suggested for each layer would then referred to the flat terrain
section denoted as symbol ‘T1’ except subgrade material would referred to station denoted
from symbol ‘TSG’. The flow path can be illustrated in Figure 3 and materials suggested is
summarizes as presented in Table 11. No ranked is assigned to the database ‘M8*’ at ‘T1’ as
discussed in the previous section 4 referred to Table 8 except adopted measured to subgrade.
Referred to Table 11 of denoted symbol ‘TSG’ at ‘T1’ shown method of adopted measured
were ranked.
E2 S3 TSG S1 C2
M8*
M8* referred to T1
with sugrade
referred to TSG.
Table 11. Problem Example Summary using Database ‘M8*’ and ‘TSG’ at ‘T1’
1323
Journal of the Eastern Asia Society for Transportation Studies, Vol. 6, pp. 1313 - 1328, 2005
RC-MSS is developed with Microsoft Visual Basic 6.0 (VB6) as the expert system tool due to
the simplicity of this program and save time by providing an easy developing of interface with
capabilities of toolbox facilities (Deitel, et al, 1999). While conventional programming
language may need to type pages of instruction to create the similar visual elements.
The first component is the knowledge base, which contains all relevant, domain-specific,
problem-solving knowledge. It contains rules and facts and it represents this knowledge in a
very simple form that could be understood by human beings. The second component is the
inference engine, which drives the knowledge stored in the knowledge base through reasoning
processes, which are similar to those of a human expert. The third component is user interface
which serves as screen with a friendly means and the keyboard to communicating with the
system (Gonzalez, et al 1993; Omar 2001). It allows the expert system to answer the
questions from user.
The flow diagram RC-MSS’ expert system is shown in Figure 4 and the main diagrams of
expert system applied for RC-MSS are presented from Figures 5 to 8.
Select Pavement
Material
Section (i+1)
Ultimate output
Selected material
output
Print
Material cost
calculation
1324
Journal of the Eastern Asia Society for Transportation Studies, Vol. 6, pp. 1313 - 1328, 2005
Figure 5. RC-MSS Data Input Diagram (Geometric Design and Traffic Estimation)
1325
Journal of the Eastern Asia Society for Transportation Studies, Vol. 6, pp. 1313 - 1328, 2005
1326
Journal of the Eastern Asia Society for Transportation Studies, Vol. 6, pp. 1313 - 1328, 2005
8. CONCLUSION
With this prototype development of pavement material selection system, the road designer is
more likely to view more proper pavement materials alternatives in the initial stage of
material selection prior to the scrutiny laboratory work is carry out. Better road design can
also be achieved and save more time on material selection process by engineer, before they
carry out the laboratory test on the selected material. The system will also be an excellent
device for novice engineers to make better pavement design decisions. However, further
investigation such as weather influential, material property in the future to make the database
of the system more effective, valid, precise and reliable database are needed to be more
comprehensive and suitable to our climate. Since this is a prototype material selection system,
the materials suggested are more general. Hence, to make the system more efficient, further
development of mechanistic analysis should be incorporate to the system. Furthermore,
thickness design, specific material performance and life cycle cost analysis should integrate to
make a comprehensive pavement material selection system.
ACKNOWLEDGEMENT
REFERENCE
Gonzalez, A.J. and Dankel, D.D. (1993) The Engineering of Knowledge-Based Systems:
Theory and Practice. Prentice Hall, New Jersey.
Deitel, H. M., Deitel, P.J. and Nieto, T.R. (1999) Visual Basic 6: How to Program. Prentice
Hall, New Jersey.
Atkins, H.N. (2003) Highway Materials, Soils, and Concrete, 4th Edition. Prentice-Hall,
New Jersey.
JKR (1985) Manual on Pavement Design. Arahan Teknik (Jalan) 5/85. Public Work
Department, Kuala Lumpur.
JKR (1986) A Guide On Geometric Design of Roads. Arahan Teknik (Jalan) 8/86. Public
Work Department, Kuala Lumpur.
Lay, M.G. (1990) Transportation Studies Volume 8, Handbook of Road Technology, 2nd
Edition, Vol. 1, Planning and Pavements.
1327
Journal of the Eastern Asia Society for Transportation Studies, Vol. 6, pp. 1313 - 1328, 2005
Conover. W.J. (1999) Practical Nonparametric Statistics: Third Edition, Wiley Series in
Probability and Statistics: Applied Probability and Statistics Section, John Wiley & Sons, Inc,
New York, Chichester, Weinheim, Brisbane.
Huang, Y.H. (1993) Pavement Analysis and Design. Prentice-Hall, New Jersey.
APPENDIX
Notations
SG1 Compacted existing soil: Compaction of the natural soil to strengthen the soil
property by increasing dry density without remove or replaced the existing soil.
SG2, TSG2 Geosynthetic stabilize: Prevent migration of weak soil and provide a working
platform. It can be goetextiles, geogrids, geonets or geomembranes.
SG3 Remove and replace existing soil: Substitute with stronger soil and compact.
TSG1 Chemically stabilized: To strengthen the soil, remove soil sensitivity to water.
Stabilizer can be lime, cement or asphalt depend on each criteria and suitability
of particular soil.
TSG Excavated and replaced stronger soil material.
TSG4 Dynamic compaction: Free fall up to 20m of heavy weight in a pattern
designed to remedy poor soil conditions.
TSG5 Preloading/ vertical drain: Suitable applied to areas with compressible water
satured soil such as clay and silty clays.
SB1 Untreated subbase material: Generally consists of lower quality materials than
the base course. Act as filter and drainage system.
SB2 Treated subbase material: Strengthen subbase using lime, cement or aspalt.
BC1 Untreated base course material: Consists of quarried rock crushed or natural
gravel. Provide the structural functions.
BC2 Treated base course material: Strengthen base course using lime, cement or
asphalt.
BC3 Dense bitumen macadam base course: Suitable to use in heavily trafficked road
and less susceptible to environmental moisture effect.
BC4 Dense bitumen macadam base with crusher run underneath: Same as function
BC3 but provide better working platform.
BC5 Untreated but thicker base course: Same as function BC1 but provide more
strength and hence reduce the thickness of surface layer.
SC1 Dense graded asphalt mixture: Asphalt Concrete is of this type. Able to
provides resilient, waterproof, resistance to deformation. But susceptible to
heat and to fuel spillage.
SC2 Dense graded asphalt mixture with modified binder: Same as SC1, but provide
extra durability to asphalt.
SC3 Gap graded asphalt mixture: Stone Mastic/ Matrix Asphalt is of this type.
Suitable to use in highly stress areas, increase skid resistance, lower noise level.
SC4 Open graded asphalt mixture: Porous Asphalt is of this type. Able to drain
water, reduce splash, prevent skidding, lower noise level, but low structural
strength, not suitable to build in steep terrain and shorter life span.
1328