Академический Документы
Профессиональный Документы
Культура Документы
One of the main tasks of the Chilean Direccin Nacional de Vialidad and
its regional and provincial secretariats is to determine the maintenance
requirements of its roads. In developing countries, most roads managed under this process are unpaved. The traditional way of determining maintenance requirements relies on the criteria applied by whoever
inspects the roads. To improve this process, the present work proposes
an easy, objective method for determining the conservation requirements of unpaved roads. Roughness thresholds are proposed because
they adequately summarize the functional characteristic of this kind of
road. Roughness thresholds were determined for different road types
after consulting public-, private-, and academic-sector experts who
proposed international roughness index (IRI) values based on the characteristics of each road. To measure the roughness of unpaved roads,
the use of a response-type roughmeter is recommended. When this type
of equipment is not available, it is proposed that IRI be estimated with
data obtained from visual inspection. The main distresses of unpaved
roads and IRIs were correlated by using data collected from 61 test
sections.
General planning,
Financing,
Land planning,
Operation execution, and
Control.
In the general planning stage, the specific maintenance requirements of roads must be determined at the regional level. For this
purpose, each road in the network is evaluated by field inspectors,
who assign a score (on a scale from 1 to 5) on the basis of the criteria and their experience. The traffic, type of road, and annual
conservation needs are determined from this score. Therefore, the
score assigned by the evaluator as well as the criteria that determine the score affect the budget to be presented in the nancing and
later stages (3).
In this paper, an alternative method is proposed that determines
the conservation needs of an unpaved road from estimated road
roughness.
BACKGROUND
Some authors believe that all types of external pressures result in
inadequate standards for unpaved low-volume roads, thus increasing
total costs for both the administration and users (1, 2). Therefore, an
unpaved road must meet a clear minimum standard.
10
11
RESEARCH METHOD
Roughness Estimation
Despite the importance of roughness, the IRI of unpaved roads rarely
is measured in developing countries, mainly because the equipment
necessary for measuring the extensive network is not available. Two
methods to measure roughness are proposed in the literature:
Gross evaluation of the distresses (6) and
Estimated circulation speed of vehicles using the road (13).
Research method.
12
2
Radj
= 69%; SE = 2.1
Gravel roads:
IRI estimated = 6.970 + 0.60 Matg_tama
2
Radj
= 58%; SE = 2.2
1
7.429
6.7
6.04
0
2.83
47.69
46.38
32.4
2.4
63.17
61.28
13.6
Step
4.144
6.6
5.99
0
1.51
4.38
0
0.92
3.33
0.002
2.14
71.49
69.24
4.4
Constant
Matg_tama
t-value
p-value
S
R2
R2 (adj.)
Mallows C-p
1
6.97
0.6
5.13
0
2.23
60.75
58.44
4.1
13
Others
Steps 3, 4, and 5 were repeated with measurements for each complete section (1,000 m), with the objective of obtaining the r&R for
each 1,000 m. The r&R measurements and the combined effect of
both are evident in Figure 2. Considering measurement sections of
1,000 m instead of 100 m has a smoothing effect that can be observed
on the r&R error.
In this experiment, IRI measurements corresponded to 1,000-mlong sections, in which case the joint r&R error is approximately
0.6 m/km. Hence, the internal measurement error contributes less
to the equation determination error (approximately 30% of the
total error of both equations).
To obtain answers from the expert panel, the Delphi method was used.
The general idea of this method is to obtain the opinion of a group
of experts, using independent surveys or questionnaires that are
answered individually and independently. The goal of the Delphi
method is to obtain an consensus expert opinion about a particular
problem, to form a basis for decision making on matters that have
been little studied or for which scientic ways of measurement are not
available (19). The method is applied to a case in the following way:
1. A survey about the study subject is designed, with questions
intended to be answered quantitatively.
2. The survey is administered to the expert panel.
3. Consistency and homogeneity of the answers are veried. In
the absence of verication, a second round of surveys is administered to the expert panel, showing them the results obtained in the
rst round.
4. The panel members responses are averaged and interpreted as
the answer from the panel as a whole.
Annual average daily traffic (AADT). The 10th, 50th, and 90th
percentiles of the total of unpaved roads in the national road network
(AADT = 7, 40, and 300 vehicles/day, respectively) were used (25).
Heavy traffic (%Heavy AADT). Because of little variability in
the heavy traffic composition, only 10th and 90th percentiles were
used from the percentage of heavy vehicles of the total of unpaved
roads of the national road network (20% and 40% of heavy vehicles
composition) (25).
Error in m/km
2.00
1.80
1.60
1.40
1.20
1.00
0.80
0.60
0.40
0.20
0.00
Sections of 100 m
Sections of 1,000 m
repeatability Reproducibility
Type of error
r&R
14
TABLE 2
AADT
%Heavy
AADT
20
40
20
40
20
40
8.6
8.8
8.2
8.3
7.4
7.1
9.5
9.4
8.9
8.7
8.0
7.8
10.5
10.6
10.0
9.8
9.0
8.9
11.8
11.7
11.1
11.1
10.0
10.2
12.7
12.6
12.1
11.9
11.2
10.9
Class
7
40
300
(R
2
adj
= 97%; SE = 0.26 )
TABLE 3
Term
Coef.
Constant
AADT
Class
S = 0.25746
13.4195
0.1178
0.004722
0.000359
0.99333
0.03324
R 2 = 0.9753
Students value.
SE Coef.
ta
113.93
0.000
13.16
0.000
29.89
0.000
R 2 (adj.) = 0.9735
between 8.0 and 12.5 m/km. These results are consistent with a previous experience (in Chile) in which roughness limits were determined in paved roads (17 ). In this investigation, values of IRIthreshold
between 4.8 and 8.1 m/km were determined, depending on the road
surface material (concrete or asphalt) and on the type of road (urban
or express).
APPLICATION EXAMPLE
As an example, the proposed method was applied to a set of test sections previously mentioned. First, the Class and AADT data were
obtained for each section from the road inventory (25). Both values
15
were used to determine IRIthreshold, using the expressions for earth and
gravel roads. Then, data from the inspection were used to calculate
IRIestimated of each section.
The differences between IRImeasured and IRIthreshold versus the differences between IRIestimated and IRIthreshold are shown in Figure 4, where
each point represents 1 of the 61 mentioned sections. Those points
that exceed 0 on the y-axis are those that require intervention based
on the IRIestimated; sections that exceed 0 on the x-axis are those that
require intervention based on the IRImeasured. The ideal situation
would be the alignment of points in a straight line that passes
through the origin with a 45 slope. Although a certain dispersion
exists, both characteristics are fullled by the straight line formed.
Figure 4 also shows that eight sections require intervention,
because the observed IRI (IRImeasured) in each one exceeds the expertdetermined threshold. Also observed are an important number of
sections in which the difference is small (0 to 1 m/km), even though
the IRIestimated stays below the proposed threshold. For this reason,
they could be added to the list of candidate sections. In the following stage, the operation(s) to be carried out in the selected sections
will be determined, and a cost will be associated with each one.
Roughness is
measured
No
Yes
Visual inspection
(IRIestimated)
Roughness measurement
(IRImeasured)
IRIestimated/measure >
IRIthreshold
No
Yes
Section entry to the
candidate list
CONCLUSIONS
A simple, objective method was developed to determine the maintenance needs of unpaved roads based on the expected level of service
from data obtained by visual inspection. The results of statistical
analyses performed to find the equations that relate IRI and distresses indicate a signicant correlation (approximately 60% of
gravel roads and of 70% earth roads). This nding would be important in the absence of testing equipment; it would be replaced with
12
y = 1.013x + 0.038
R = 0.687
10
8
Difference IRI measured - IRI threshold
FIGURE 3
6
4
2
0
-12
-10
-8
-6
-4
-2
-2
-4
-6
-8
-10
-12
Difference IRI estimated - IRI threshold
FIGURE 4
10
12
16