INTRODUCTION
Pipeline systems are generally a convenient means for transferring oil and
gas onshore or offshore due to the economic and safety reasons. However, with
increasing age, the pipeline integrity can be affected by a range of corrosion
mechanisms. Corrosion is defined as the destruction or deterioration of a material
because of its reaction with the environment (Fontana and
Greene, 1987). Pipeline is exposed to both internal and external corrosion.
The internal corrosion of pipeline is due to the harsh condition of hydrocarbon
fluid which includes the presence of CO_{2}, H_{2}S and organic
acid (Kusha et al., 2011). External corrosion
occurs due to the extreme conditions of the surrounding environment when the
preventive measures failed; such as older/degraded coating or poorly coated
pipeline (Shafiq et al., 2010; Chouchaoui
and Pick, 1996). This will result in metal loss at the corroded location
in the pipeline and may eventually lead to its failure. The impact of corroded
pipeline problems cause economic consequences; such as reduced operating pressure,
loss of production due to downtime, repairs or replacement and consequently
increase of costs.
Thus, several pipelines systems are kept in operation even though they have
shown signs of corrosion based on the data obtain from the corrosion management,
inspection and monitoring systems, i.e., intelligent pig. The continued operations
of these pipelines are basically done after the FFS assessment to determine
their residual strength and recalculation of the maximum allowable internal
pressure of the product being transferred (Netto et
al., 2005). The structural integrity assessment of corroded pipeline
has become vital to assist engineers to make wise decision toward replacing
or repairing a pipeline. It is essential to ensure the continued safe operation
and non hazardous incidences which might affect the life and the environment.
Various methods for assessing pipeline corrosion are available and commercially
have been practiced by the industry, such as ANSI/ASME (1991)
For a more realistic way of pipeline corrosion representation (Silva
et al., 2007), new criteria were developed, such as RSTRENG Effective
Area (Kiefner and Vieth, 1989) and DNV RPF101 (Netto
et al., 2005). These methods include the specification dealing with
the effects of the interacting defects. Even though these codes have been used
widely for assessing the integrity of inservice pipelines, they are known to
be conservative (Belachew et al., 2009). In other
words, pipelines which have been assessed by these codes for the purpose of
FFS analysis probably would lead to either unnecessary maintenance or premature
replacement.
The occurrence of corrosion is divided into several categories, namely individual
pits, colonies of pits, general wallthickness reduction, or a combination of
these (Lee et al., 2005). An interacting defect
is defined as the one that interacts with neighbouring defects in an axial or
circumferential direction (DNV, 2004). For colonies of
corrosion defects, as the distance between the defects decreases, the defects
will begin to interact, resulting in reduced burst strength of the pipeline.
A more reliable defect assessment method is needed due to the conservatism involved
in the available assessment method (Belachew et al.,
2011). This is an approach to understand the effect of interacting defects
toward the pipeline burst strength in a more reliable and convenient method
other than performing experimental testing.
Therefore, the modelling of the problem using FEA method can assist engineers
to assess the burst strength of pipeline with interacting defects. The objective
of this study is to estimate the burst pressure of corroded pipeline due to
interacting corrosion defects by the means of FEA. The analysis is performed
by nonlinear FEA simulation using ANSYS Software. The effect of different spacing
between two defects aligned in longitudinal direction to their failure pressure
will be studied. Then the FEA results will be compared to the numerical values
calculated from the DNV RPF101 code.
DNVRPF101 CODE
DNVRPF101 method for interacting defects (part A) was used in this study
to estimate the burst pressure of the corroded pipeline. In DNV procedure, all
the defects that are supposed to interact are projected onto a longitudinal
line. The metal loss is represented by the maximum defects depth and the projected
defects length.
In the case of overlapped defects, they are combined to form composite defects.
The formation of combined defects is estimated by taking the combined length
and the depth of the deepest defect. For combination of overlapping internal
and external defects, the depth of the composite defect is the sum of maximum
depth of those two defects. Each defect or composite defects (i) with length
(l_{i}) and depth (d_{i}) is treated as a single defect
and failure pressure (p_{i}) is defined based on the expression below:
where, N is the is number of projected defects, D is the nominal outside diameter
(mm), t is uncorroded measured pipe wall thickness, f_{u} is the ultimate
tensile strength, γ_{m} is the partial safety factor for model
prediction, γ_{d} is the partial safety factor for corrosion depth
and Q_{i} is the length correction factor for individual defect, given
by:
The correction depth over thickness ratio is determined by the following expression:
where, εd is the factor for defining a fractal value for the corrosion
depth and StD[d_{i}/t] is the standard deviation of the measured (d_{i}/t)
ratio.
Next, the combinations of adjacent defects were investigated. Take note that
for combined defects, the effective length (l_{nm}) is the total
length of the projected defects and the spacing between the defects (Fig.
1). For defects from n to m, the effective length is given by:
where, l_{i} is the longitudinal length of an individual defect,
s_{i} is the projected distance between the two adjacent defects. Meanwhile,
the effective depth (d_{nm}) of combined defects formed from all of
interacting defects from n to m (Fig. 1) is calculated as
below:
where, d_{i} is the depth of an individual defect.

Fig. 1: 
Combined length of all combination of adjacent defects (DNV,
2004) 
The failure pressure (p_{nm}) of the combined defects from n to m is
calculated by replacing (l_{i}) and (d_{i}) with (l_{nm})
and (d_{nm}) in Eq. 1 and 2. The
minimum value, calculated for all single and combined defects, is taken as the
failure pressure for the current projection line.
According to the DNV interaction rules, there is no interaction if the longitudinal
(s_{l}) and circumferential (s_{c}) distances between the defects
satisfy the following conditions:
FINITE ELEMENT METHOD
The models were developed from API 5L X65 steel alloy with the nominal dimension
of 300 mm diameter, 20 mm wall thickness and 1000 mm section length. Meanwhile,
the defects geometry was selected to cover the following basic parameters: normalized
defect depth (d/t of 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7 and 0.8), normalized
defects spacing of 0.5, 1.0, 2.0, 4.0 and 8.0). A total of 40 models with internal
defect geometry characterized by a length of l = 100 mm and width of c = 90
mm were generated.
Since the problem involved two equally shaped defects, only a quarter of a
full pipe was generated based on the symmetric condition. This was to reduce
the size of the model and hence to reduce the processing time of the simulation.
These models were meshed with SOLID95 elements which are defined by 20 nodes
having three degrees of freedom at each node, which are translations in the
nodal x, y and z directions. SOLID95 brick elements were chosen because they
can tolerate irregular shapes without as much loss of accuracy and have compatible
displacement shapes and are well suited to model curved boundaries. These elements
also have plasticity, stress stiffening, large deflection and large strain capabilities.
Two layers of elements were used at the defect region through its ligament.
Fine meshes were utilized at the defect region while coarser meshes were utilized
farther from the corrosion defect region, as shown in Fig. 2.
One end of the pipe was fixed by constraining all degrees of freedom as to
simulate the enclosed end of the pipe. The symmetric boundary condition was
imposed to the other end of the pipe (which is closer to the defect) and also
to the sides of the pipe.

Fig. 2: 
Finite element model of the defect, mesh and boundary conditions 
The inner wall of the model is then subjected to incremental internal pressure
loading during the simulation. Figure 2 shows the boundary
conditions imposed on the model in this study.
In this study, nonlinear analyses were carried out. In order to correctly evaluate
the corroded pipe, appropriate failure criterion should be established to decide
the failure point during the simulation (ANSI/ASME, 1991).
The failure/stopping criterion of this simulation was when the VonMises stress
distribution across the entire ligament of the pipe reaches the ultimate tensile
stress, 530.9 MPa. The pipe is considered to fail when this condition is achieved.
As the pressure applied on the internal surface of the pipe increased, the critical
stress starts to propagate along the edge of the defect and spread around the
defect area. The assessment of the critical stress through the entire ligament
was carried out by considering several points at the critical defect area.
RESULTS AND DISCUSSION
Figure 3 shows the results for an example case of d/t = 0.2
and .
The critical section was observed along the edge of the defect. The critical
stress starts from this edge and spreads around the defect area. The simulation
is terminated when the Von Mises stress value through the thickness of this
critical section reaches ultimate tensile stress value of 530.9 MPa.
The failure pressure obtained in this study was normalized with the failure
pressure of an intact pipe obtained using the FEA method. The failure pressure
of an intact pipe found by FEA method was 90 MPa. Figure 4
shows the results of normalized corroded pressure versus normalized defect spacing,
obtained by FEA and DNV code, for an example case of d/t = 0.4.

Fig. 3: 
Von Misses stress distribution for d/t = 0.2 and = 0.5 (internal
view) 

Fig. 4: 
Effect of defect spacing on burst strength for d/t = 0.4 
Generally, the results obtained from FEA are higher compared to the values
obtained using the DNV Code. The non linear finite element analyses yielded
results which have similar trend with the empirical solution using the DNV Code.
These observations apply for other cases as well.
Figure 5 shows the effect of defect depth and defect spacing
on the failure pressure of pipes with defects using FEA. Generally, as the distance
between the defects decrease, the maximum allowable corroded pressure will also
decrease. Furthermore, as the depth of the defects increase, the failure pressure
will decrease.

Fig. 5: 
Effect of defect depth and defect spacing on the failure pressure
using FEA 
The effect of spacing between the defects is not significant for the shallow
defects as was observed for d/t = 0.1 and d/t = 0.2. In other word, the effect
of interaction on the pipe failure pressure for defects with depth of less than
20% of pipe wall thickness is minimal. As the defect depth increase, the effect
of interaction becomes obvious. For d/t ratios of 0.3 to 0.8, the gradual drop
in failure pressure was observed when the normalized spacing is less than 4.
This shows that, when the spacing between the defects is small, they start to
interact with each other and reduce the maximum allowable corroded pipe pressure.

Fig. 6: 
Effect of defect depth on the ratio of FEA results over DNV
Code results 
The drastic drop in failure pressure was observed for d/t = 0.8. The effect
of interaction is more critical for deeper defects.
The results from FEA is consistently higher than the DNV Code results. This
was due to the safety factor applied in the empirical calculation which yields
a lower corroded pipe failure pressure as to avoid reaching the exact failure
pressure during standard operating condition. Figure 6 shows
the effect of the defect depth on the ratio of FEA results over the DNV Code
results. This factor is not a fixed value since the multiplication factor in
FEA was not the same for every cases of defects depth. This was due to stress
concentration at the edges of the defects during the simulation process. The
deeper the defect depth, the higher the stress concentration at the edge of
the defect. Since the remaining pipe wall thickness is small, the stress distribution
on the entire ligament of the edge easily reaches the failure criterion. Figure
6 shows a decreasing trend of the factor as the depth of the defects is
increasing. The best plot on the graph is based on the average value of the
ratio of the FEA results over the DNV Code results. The ratio can be expressed
as a function of:
Ratio = 0.481(d/t)^{2} = 0.805(d/t)+1.583 
CONCLUSION
Consideration for maintenance or replacement of corroded pipeline is crucial
because it affects directly to the cost of the firm. The existing codes that
have been widely practiced by the industry are too conservative. This may lead
to unnecessary maintenance or premature replacement of corroded pipeline. Therefore,
the FEA is one of the reliable methods to assess the burst strength of the corroded
pipeline. Burst strength of pipeline with interacting corrosion defects can
be accurately predicted by FEA using ANSYS software. The application of FEA
can reduce the conservatism involved in the conventional methods. From this
study, the ratio of FEA results over the DNV Code results is a multiplication
factor expressed by a function below:
Ratio = 0.481(d/t)^{2}0.805(d/t)+1.583 