Introduction
2
Several techniques are currently used to identify
conditions that are corrosive to buried iron pipes.
The most common method is the 10-point scoring
(10-P) that was introduced by CIPRA (Cast Iron
Pipe Research Association, predecessor of DIPRA,
Ductile Iron Pipe Research Association) in 1964 for
cast iron pipes. The method was subsequently
extended to ductile iron pipes [10]. The 10-P method
uses five soil properties including resistivity, pH,
redox potential, sulfide, and moisture content. In
essence, this method classifies the soil as either
corrosive or non-corrosive, but it does not
provide information on the intensity of corrosivity.
Najjaran et al [11] introduced a fuzzy expert system
that determines the soil corrosiveness using
corrosivity potential (CoP), i.e., a real number in the
[0, 1] interval, for given soil properties. The expert
system used a fuzzy rule base that was developed
based on expert opinion. They also used a linear
regression model to predict the pipe deterioration
rate from the CoP. The model was tested using a set
of field data that were previously published in [12].
Although the CoP seems to be an appropriate
criterion to gauge soil corrosivity, the regression
model did not show a strong correlation between the
CoP and deterioration rate. This can occur due to
imprecise measurements of soil properties and
corrosion rates, or it simply indicates that the
corrosion rate is also influenced by factors other than
those considered in the fuzzy expert system.
In this paper, a new approach is proposed in
which the CoP obtained from the soil properties is
only one body of evidence providing partial
information about the corrosivity of the
environment. Another body of evidence for
corrosivity potential is estimated from corrosion rate
obtained from maximum pit depth and pipe age.
Both bodies of evidence are derived from uncertain
(vague and incomplete) data. Fuzzy logic can
provide an appropriate framework to deal with such
data. In this paper, two fuzzy models are developed
to estimate the corrosivity potential, one (CoP1)
based on soil properties and another (CoP2) based on
corrosion rate. Subsequently, the two CoPs (CoP1
and CoP2) are fused using evidential reasoning,
proposed by Dempster [13] and Shafer [14], to
obtain an updated CoP that is expected to be more
reliable than each of the two individual estimates
obtained from the fuzzy models.
Corrosion
Rate
Inference
Mechanism
CoP
Evidential
Reasoning
DST
Fusion
Inference
Mechanism
CoP2
CoP
SISO Rule
Base Model
Redox
potential (mV)
Sulfide
30
51
-1
pH
Clay fines %
0.020
3000
7.1
0.026
Same as above
0.030
Same as above
MISO Rule
Base Model
Soil
Properties
Determination of Corrosivity
Potential (COP)
R (-cm)
Fuzzy Inference
Corrosion rate
(mm/yr)
Example*
Final Decision
4
The partitions of the CoP are uniformly
distributed in its universe of discourse, i.e., the [0, 1]
interval. The partitions of the corrosion rate are
defined based on the distribution of maximum pit
depths obtained from about 500 pipe samples. The
universes of discourse and partitions of the soil
properties are defined based on their intervals used
in the 10-P method [5].
VH
Redox potential
(mV)
Sulfide
Corrosion Rate
(mm/yr)
CoP
Clay fines %
pH
VL
R (-cm)
n/a
n/a
n/a
n/a
n/a
n/a
1500
n/a
n/a
n/a
2000
n/a
n/a
n/a
0.010
0.25
1500
-50
-1
2000
-50
-1
0.010
0.25
2000
20
-20
-1
0.025
0.25
2500
30
0.030
0.50
2000
20
-20
-1
0.025
0.25
2500
30
25
0.030
0.50
2500
7.5
30
50
0.035
0.50
0.75
3000
45
100
0.040
2500
7.5
30
50
0.035
0.50
3000
45
100
0.040
0.75
5000
14
60
150
0.050
0.75
1.00
5000
14
60
150
0.060
n/a
n/a
n/a
n/a
n/a
0.050
0.75
n/a
n/a
n/a
n/a
n/a
0.060
1.00
n/a
n/a
n/a
n/a
n/a
0.100
1.00
n/a
n/a
n/a
n/a
n/a
0.100
1.00
5
of evidence (i.e., CoP1 obtained using the soil
properties and CoP2 obtained using the corrosion
rate) for all three examples broadly agree with the
large mass assigned to L. The ignorance mass is
relatively small but it steadily increases as the
corrosion rate (second body of evidence) increases.
For instance, ignorance is significantly higher in
example 3 as a consequence of conflicting evidence
from the two bodies of information. Thus, ignorance
mass can be used as measure of reliability for the
final inference of corrosivity potential.
Table 3 BPA for Example1
VL
MOM
0.020
0.980
BPA
0.017
0.833
MOM
BPA2
0
0
1
0.650
M
H
CoP1, 1 = 0.85
0
0
0
CoP2, 2 = 0.65
0
0
0
0
VH
0.150
0
0
0.350
VH
VL
0.011
0.006
0.541
0.292
0
VH
0.098
0.052
0.006
0.931
0.063
CoP
References
CoP1
CoP2
[1]
[2]
[3]
VH
Ex.1
0.006
0.931
0.063
Ex.2
0.006
0.803
0.019
0.172
Ex.3
0.006
0.292
0.097
0.605
Conclusions
[4]
[5]
[6]
[7]
[8]
6
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]