You are on page 1of 9

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/249657141

Modified Equation for the Assessment of Long Corrosion Defects

Conference Paper · June 2001

CITATION READS

1 593

4 authors:

Adilson Benjamin Ronaldo Vieira


Petróleo Brasileiro S.A. Pontifícia Universidade Católica do Rio de Janeiro
69 PUBLICATIONS   324 CITATIONS    51 PUBLICATIONS   285 CITATIONS   

SEE PROFILE SEE PROFILE

Jose Freire Jaime Castro


Pontifícia Universidade Católica do Rio de Janeiro Pontifícia Universidade Católica do Rio de Janeiro
91 PUBLICATIONS   392 CITATIONS    166 PUBLICATIONS   1,064 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Global Buckling of Heated Pipelines View project

FATIGUE LIMIT OF A COLD DRAWN STEEL SAE 1020 BY THERMOGRAPHY APPROACH View project

All content following this page was uploaded by Jaime Castro on 23 August 2017.

The user has requested enhancement of the downloaded file.


Proceedings of OMAE’01
20th International Conference on Offshore Mechanics and Arctic Engineering
June 3-8, 2001, Rio de Janeiro, Brazil

OMAE01/PIPE-4111

MODIFIED EQUATION FOR THE ASSESSMENT OF LONG CORROSION DEFECTS

Adilson C. Benjamin / PETROBRAS Research Center

Ronaldo D. Vieira / PUC-Rio – Department of Mechanical Engineering

Jose Luiz F. Freire / PUC-Rio – Department of Mechanical Engineering

Jaime T. P. de Castro / PUC-Rio – Department of Mechanical Engineering

ABSTRACT  - constant which value depends of the geometric shape


PETROBRAS is conducting a research project with the adopted to represent the area of metal loss
purpose of investigating the behavior of pipeline with long Aeff – subset of the projected area of metal loss calculated using
corrosion defects. In the first phase of this project a database of an iterative algorithm
corroded pipe tests was generated. In this paper the failure Leff - axial extent of the effective area
pressures of the PETROBRAS database are compared with (ptest)i – failure pressure of laboratory test “i”
those predicted by the ASME B31G, the 085dL, the Effective (pequation)i – predicted failure pressure for the laboratory test “i”
Area and the DNV RP-F101 (Part B) methods. The best fit to
the tests results is achieved by the equation of the DNV RP-
INTRODUCTION
F101 method. However as the length of the defect increases
It is generally accepted that the traditional ASME B31G
this equation have a trend to give results that depart from the
method [1] gives overly conservative results for long defects,
tests results. Based on the equation of the DNV RP-F101
i.e., the values of failure pressures predicted by the B31G
method, a new equation is developed. This equation fits well to
equation are excessively low. Higher values of failure pressures
the PETROBRAS laboratory tests and does not have a trend to
for this type of defect are predicted by the well known
give results that depart from the tests results as the length of the
Effective Area method [2] and 085dL method [2] and the
defect increases.
recently developed DNV RP-F101 (Part B) method [3].
PETROBRAS is conducting a research project with the
NOMENCLATURE purpose of investigating the behavior of pipeline with long
De –outside diameter of the pipe corrosion defects. In the first phase of this project a database of
tn – nominal wall thickness of the pipe corroded pipe tests was generated. This database contains the
t - wall thickness of the pipe results of 9 burst tests of tubular specimens containing a single
t* - remaining ligament wall thickness in the defect external simulated corrosion defect [4]. In all these tests the
d – maximum depth of the corrosion defect tubular specimens were loaded with internal pressure only.
L – axial extent of the corrosion defect A distinguished characteristic of the PETROBRAS
y – yield stress of the pipe material database of corroded pipe tests is that 100% of the tests
u – ultimate stress of the pipe material contained in the database are pressure tests of long defects.
pf – failure pressure of the corroded pipe This high percentage of long defects assures that the specific
po – reference pressure characteristics of the failure behavior of this type of defect will
be detected more easily during the ongoing research work.
M – Folias bulging factor
Databases generated in other research projects [5,6] contain
A – area of metal loss projected in the longitudinal plane
more pressure tests but the percentage of long defects is
through the wall thickness
significantly smaller than in the PETROBRAS database. As the
Ao – original uncorroded area of length L and thickness t
objective of these projects was to develop assessment criteria

1
for a wide range of defect geometries it was necessary to have a 2t  1  ( 2 / 3)( d / t ) 
database with as many types of defects as possible. pf = 1.1 y  1 
(7)
Another unique characteristic of the PETROBRAS
De  1  ( 2 / 3)( d / t ) M 
database is that the only varying parameter from one defect to
another is the defect length. As the other geometric and where,
material parameters are not subjected to significant variations
1/ 2
the changes detected in the failure behavior of the tubular  L2 
specimens are due predominantly to the changes in the length M = 1  0.8  (8)
of the defects.  De t 
In this paper the failure pressures measured in the
PETROBRAS laboratory tests are compared with those - For L > 20 De t
predicted by the ASME B31G, the 085dL, the Effective Area
and the DNV RP-F101 (Part B) methods. The best fit to the
M  (infinitely long defect)
tests results is achieved by the equation of the DNV RP-F101
method. However as the length of the defect increases this
2t  d 
equation have a trend to give results that depart from the tests pf = 1.1 y 1  (9)
results. Based on the equation of the DNV RP-F101 method, a De  t
new equation is developed. This equation fits well to the
PETROBRAS laboratory tests and does not have a trend to
give results that depart from the tests results as the length of the Failure Equation of the 085dL Method or Modified
defect increases. B31G Method

2t  1  (0.85)( d / t ) 
EQUATIONS FOR THE ASSESSMENT OF THE pf = (y + 69 MPa)  1 
(10)
PRESSURE STRENGTH OF CORRODED PIPELINES
De 1  (0.85) ( d / t ) M 
As became evident in a recently published review paper
[7], at least ten methods for the assessment of corroded where,
pipelines under internal pressure loading are available. In the
present paper only four of these methods will be addressed, for L  50 De t
namely, the traditional ASME B31G method [1], the 085dL 1/ 2
method or Modified B31G method [2], the Effective Area  L2  L2  
2

method or RSTRENG software [2] and the DNV RP- M = 1  0.6275  0.003375   (11)
 De t  De t  
F101 (Part B) method [3]. These four methods apply failure
equations that have the following general format:
L2
for L > 50 De t M = 3.3  0.032 (12)
 1 (A / A )  Det
pf = po  o
1 
(1)
 1  ( A / Ao M
) 

where, Failure Equation of the Effective Area Method or


RSTRENG Software
A=Ld (2)
Ao = L t (3) 2t  1  ( Aeff / Ao ) 
=1 (rectangular shape) (4) pf = (y + 69 MPa)  1 
(13)
=2/3 (parabolic shape) (5)
De  1  ( Aeff / Ao ) M 
 = 0.85 (6)
where,
for Leff  50 De t
Failure Equations of the ASME B31G Method 1/ 2
 L2  L2  
2

M = 1  0.6275 eff  0.003375 eff   (14)


- For L   De t  De t  
20 De t
   

2
L2eff (defects longer than 20 De t ), this equation is to be applied
for Leff  50 De t M = 3.3  0.032 (15)
De t only on the assessment of this type of defect.
The general format of the PB equation is identical to the
format of the equation of the DNV RP-F101 method. The only
Failure Equation of the DNV RP-F101 (Part B) Method difference is that the mathematical expression used to calculate
the Folias factor M was modified to approximate the results of
2t  1  (d / t )  the PB equation of the failure pressures measured in
pf = u  1 
(16) PETROBRAS laboratory tests.
( De  t ) 1  (d / t ) M  The adjustment of the Folias factor expression was
obtained by minimizing the following expressions:
where,
min ABS (  ( ptest ) i  ( pequation ) i 
9
1/ 2 ) (18)
 L  2
i 1
M = 1  0.31  (17)
 De t 
9
 ( ptest ) i  ( pequation ) i 
min  ABS   (19)
i 1  ( ptest ) i 
PETROBRAS LABORATORY TESTS
A detailed description of the laboratory tests and its results The PB equation for the assessment of long defects is
was published in a previous paper [4]. Herein only a brief defined as:
description will be presented.
Burst tests of 9 tubular specimens containing a single
2t  1  (d / t ) 
external simulated corrosion defect were carried out. These pf = u  1 
(20)
specimens were cut from longitudinal welded tubes made of ( De  t ) 1  (d / t ) M 
API 5L X60 steel with an outside diameter of 323.9 mm and a
nominal wall thickness of 9.53 mm. Tensile specimens were where,
tested to determine material properties. A summary of the
tension test results for the transverse tensile specimens is 1/ 2
 L2 1  L2 
4

presented in Table 1. M = 1  0.217     (21)
Each of the 9 tubular specimens had one external corrosion  De t 1.15 x 10 6  De t  
defect. These defects are corrosion patches, longitudinally
oriented, with uniform depth d and uniform width w
(circumferential length). All the defects had the same nominal It is generally accepted that the failure pressure of an
depth of 6.67 mm (70% of the nominal wall thickness) and the infinitely long and uniform depth defect is the lower bound
same nominal width of 95.3 mm (ten times the nominal wall pressure for a set of uniform depth defects which are shorter
thickness). They are smooth rectangular defects, i.e., the shape but have the same depth. In this way, the failure pressures
of the area of metal loss is rectangular with smooth edges. predicted by the PB equation and by the DNV RP-
These edges were made with small radius. The defect lengths F101 equation have the same lower bound because for
and the ratios of the square length to the product of the outside infinitely long defects the differences due to the different Folias
diameter versus the nominal wall thickness are presented in factor equations disappear (M  ) and both equations assume
Table 2. By the values of these ratios it is seen that all the the following simplified form:
defects are long according to the criterion of the ASME B31G
2t  d
method (long defects are the ones in which L > 20 De tn ). plowerb = u 1  (22)
( De  t )  t
The mean of the measured wall thickness and the mean of
the measured remaining ligament thickness for the tubular
specimens are presented in Table 3. The ratios (De / t) and
(d / t) are also presented in Table 3.
COMPARISON BETWEEN TESTS RESULTS AND
ASSESSMENT METHODS RESULTS
PB EQUATION Table 4 presents the actual failure pressures of the 9
Based on the equation of the DNV RP-F101 (Part B) tubular specimens along with those predicted by the PB
method, a new equation, called the PB equation (PB is equation and by the equations of the ASME B31G, the 085dL,
PETROBRAS in an abbreviated form), was developed [9]. As the Effective Area and the DNV RP-F101 (Part B) methods. In
the PETROBRAS laboratory tests include only long defects the calculations of the predicted failure pressures the mean of

3
the measured wall thickness of the specimen and the mean of to a pressure of 10.594 MPa, which is the lower bound pressure
the measured remaining ligament wall thickness in the defect calculated using equation 22. The calculations of the predicted
were used. The ligament thickness value was assumed to be the failure pressures were performed assuming the value of
same all over the defect, even for the Effective Area method 9.76 mm for the pipe wall thickness and the value of 3.07 mm
that can take into account variations in the defect depth. Where for the remaining ligament thickness. These values are,
necessary, the mean yield stress (452 MPa) and the mean respectively, the mean of the mean wall thickness of the
ultimate stress (542 MPa) were used. The Effective Area specimens and the mean of the mean remaining ligament
method calculations were performed using the computer thickness in the defects. For these data the values of the ratios
program RSTRENG2 [8]. (De / t) and (d / t) are 33.2 and 0.69, respectively.
In a previous paper [4], the failure pressures predicted by In figure 4 a detail of the plot shown in figure 3 is
these four methods were calculated using the minimum presented in order to show in an amplified scale the region of
remaining ligament thickness, instead of the mean of the the plot situated between the abscissa values of 4 and 10
measured remaining ligament thickness in the defect. As one of (4  L / De t  10).
the purposes of that paper was the evaluation of the inherent
conservatism of each method the use of the minimum In figure 3 the failure pressures predicted by the
remaining ligament thickness was adequate. However to search PB equation are greater than the ones predicted by the DNV
for the best fit equation the use of the mean of the measured RP-F101 equation for values of L smaller than 47.5 De t .
remaining ligament thickness is more adequate.
The results given by the equations of the four methods For values of L greater than 47.5 De t the failure pressures
investigated are presented in Figure 1. This figure shows that predicted by the PB equation decay more rapidly than the ones
the best fit to the tests results is achieved by the equation of the predicted by the DNV RP-F101 equation. Beyond L equal to
DNV RP-F101 method. However as the length of the defect 256 De t the distance between the PB equation curve and the
increases this equation have a trend to give results that depart
from the tests results. Examining figure 1, and being aware of lower bound horizontal line is almost imperceptible while the
the fact that the value of the failure pressure decreases with the difference between the DNV RP-F101 curve and the lower
increase of the defect length, this trend can be easily detected. bound line is still noticeable. Only beyond a value of L that is
For the four shortest defects the actual failure pressures are excessively high to be shown in the figure (L = 8100 De t ),
greater than 12.5 MPa and for the five longest defects the the distance between the DNV RP-F101 curve and the lower
actual failure pressures are smaller than 12.5 MPa. In Figure 1, bound horizontal line becomes sufficiently small to be
the points predicted by the RP-F101 equation for the four imperceptible.
shortest defects are all located under the 1:1 line, i. e., they are The failure pressures measured in PETROBRAS
on the safe region of the plot. On the other hand, the points laboratory tests were plotted in figures 3 and 4 with the
predicted by the RP-F101 equation for the five longest defects objective of showing which of the two curves (the curve of the
are all, except one, located above the 1:1 line, i. e., they are on DNV RP-F101 equation or the curve of the PB equation)
the unsafe region of the plot. represents better the failure pressure of long defects as a
function of the defect length. However, in doing so it is
The failure pressures of the 9 tubular specimens predicted necessary to be aware to the fact that the ratios (De / t) and
by the equation of the DNV RP-F101 method and by the PB (d / t) of the tubular specimen are different from one specimen
equation are presented in Figure 2. This figure shows that the to another (see Table 3) and from the ratios (De / t) and (d / t)
best fit to the tests results is achieved by the PB equation. calculated with the data used in the generation of the two
Different from the points predicted by the DNV RP- curves. It is also worth to remember that in any set of
F101 equation, the points predicted by the PB equation are experimental results there is always some scatter.
symmetrically distributed around the 1:1 line. Two of the four Figure 4 shows that the decay of the failure pressures
shortest defects (defects with failure pressures greater than calculated by the PB equation is similar to the decay shown by
12.5 MPa ) are located above the 1:1 line and two are under the laboratory tests points. In a different way, the failure
this line. Two of the five longest defects (defects with failure pressures calculated by the DNV RP-F101 equation for values
pressures smaller than 12.5 MPa ) are located above the
1:1 line, two are under this line and one is approximately on the of L greater than 47.5 De t show a trend to depart from the
line. laboratory tests points.
Figure 3 presents two curves of the failure pressure versus Table 5 presents the ratios of the predicted to the actual
the dimensionless length of the defect (L / De t ) generated failure pressure (pequation / ptest ).The predicted failure pressures
were calculated using the PB equation and the equations of the
using the DNV RP-F101 equation and the PB equation. The four methods that are being investigated. These results were
failure pressures measured in PETROBRAS laboratory tests are used to generate the curves of the ratio (pequation / ptest ) versus
also shown. It is also plotted an horizontal line correspondent

4
the dimensionless length of the defect (L / De t ) that are give results that depart from the tests results as the length of the
defect increases.
presented in Figure 5. The PB equation give accurate results for a specific set of
Figure 5 shows that, except for the PB equation results, the defects and must be confirmed for a broad range of pipeline
ratios of the predicted to the actual failure pressure exhibit a characteristics (De/t, d/t, type of material, etc.). In the second
clear trend to increase as the defect length increases. For the phase of the research project that PETROBRAS is conducting,
B31G results this trend is expected because the actual failure the failure pressures predicted by this equation will be
pressure decreases with the increase of the defect length while compared with the results contained in other databases.
the predicted failure pressure calculated by the equation for
long defects of this method (equation 9) is a function of the
defect depth only. For the PB equation the ratios of the REFERENCES
predicted to the actual failure pressure fluctuate around the 1. ASME, 1991, "ASME-B31G - Manual for Determining the
unitary value exhibiting no trend to increase or decrease as the Remaining Strength of Corroded Pipelines – A Supplement
defect length increases. to ANSI/ASME B31 Code for Pressure Piping", The
Table 5 presents also the mean of the ratios (pequation / ptest) American Society of Mechanical Engineers, New York.
for each equation. The maximum of accuracy that one equation 2. KIEFNER, J. F. and VIETH, P. H., 1989, "A Modified
could achieve would be a mean of the ratios (pequation / ptest) Criterion for Evaluating the Remaining Strength of
equal to the unity. The mean ratio of the ratios (pequation / ptest) Corroded Pipe", Final Report on Project PR 3-805,
are 0.998, 1.010, 0.911, 1.178 and 0.746 for PB equation, the Pipeline Research Committee, American Gas Association.
RP-F101 equation, the Effective Area equation, the 085dL 3. DNV, 1999, "Corroded Pipelines – Recommended Practice
equation and the B31G equation, respectively. RP-F101", Det Norske Veritas, Norway.
Table 6 presents the failure pressure prediction statistics 4. BENJAMIN, A. C., VIEIRA, R. D., FREIRE, J. L. F. and
for the PB equation and for the equations of the four methods CASTRO, J. T. P., 2000, "Burst Tests on Pipeline with
that are being investigated. The mean error are 1.8%, 2.6%, Long External Corrosion", International Pipeline
8.9%, 17.8% and 25.4% for the PB equation, the RP-F101 Conference, ASME IPC 2000, Vol. 2, pp. 793-799.
equation, the Effective Area equation, the 085dL equation and 5. VIETH, P. H. and KIEFNER, J. F., 1994, “Database of
the B31G equation, respectively. Corroded Pipe Tests”, PRCInternational, AGA Catalog
Number L51689, April 4, 1994.
6. BATTE, A. D., FU, B., KIRKWOOD, M. G. and VU, D.,
CONCLUSION 1997, “New methods for determining the remaining
A research project conducted by PETROBRAS, as part of strength of corroded pipelines”, Conference on Offshore
a Pipeline Technological Program called PRODUT [10], have Mechanics and Arctic Engineering (OMAE 97), Vol. 5,
generated a database of corroded pipe tests. This database Pipeline Technology, pp. 221-228.
contains the results of 9 pressure tests of pipes with long 7. STEPHENS, D. R. and FRANCINI, R. B., 2000, “A Review
corrosion defects (defects longer than 20 De t ). and Evaluation of Remaining Strength Criteria for
Comparisons of the failure pressures contained in this Corrosion Defects in Transmission Pipelines”,
database with the pressures predicted by the ASME B31G, the ETCE/OMAE 2000 Joint Conference, Paper
085dL, the Effective Area and the DNV RP-F101 (Part B) ETCE2000/OGPT-10255.
methods were performed. The best fit of the tests results was 8. VIETH, P. H. and KIEFNER, J. F., 1993, “RSTRENG2
achieved by the equation of the DNV RP-F101 (Part B) USER’S MANUAL”, Project PR-218-9205, Final Report,
method. However as the length of the defect increases this Pipeline Research Committee, American Gas Association,
equation have a trend to give results that depart from the tests March 31, 1993.
results. 9. BENJAMIN, A. C., 2000, “Empirical Methods for the
For the four methods investigated, the ratios of the Structural Evaluation of Corroded Pipelines with Long
predicted to the actual failure pressure exhibit a trend to Corrosion Defects”, 7th Report of Project 317900,
increase as the defect length increases. Only for the B31G PETROBRAS/CENPES/DIPREX/SEDEM, September,
results this trend is expected because the actual failure pressure 2000 (in Portuguese).
decreases with the increase of the defect length while the 10. GOMES, M. G. F. M. and BAPTISTA, R. M., 2000,
predicted failure pressure calculated by the equation for long "PRODUT: A Pipeline Technological Program to Face the
defects of this method is a function of the defect depth only. Challenges in Brazil", International Pipeline Conference,
Based on the equation of the DNV RP-F101 (Part B) ASME IPC 2000, Vol. 1.
method, a new equation, called the PB equation was developed.
The PB equation is to be applied only on the assessment of
long corrosion defects. This equation fits well to the
PETROBRAS laboratory tests and does not have a trend to

5
Table 1 – Tension test results for the transverse tensile specimens
(y)0.2% (MPa) u (MPa)
mean value 452 542

Table 2 – Length of the defects of the tubular specimens


TS5.1 TS1.2 TS2.2 TS2.1 TS3.1 TS1.1 TS3.2 TS4.1 TS4.2
L (mm) 255.6 305.6 350.0 394.5 433.4 466.7 488.7 500.0 527.8
L2 / ( De t n ) 21.2 30.3 39.7 50.4 60.9 70.6 77.4 81.0 90.2

Table 3 – Thickness measurements of the tubular specimens


TS5.1 TS1.2 TS2.2 TS2.1 TS3.1 TS1.1 TS3.2 TS4.1 TS4.2
t (mm) 9.80 9.66 9.71 9.71 9.91 9.74 9.79 9.79 9.74
t* (mm) 3.09 3.08 3.18 3.05 3.02 3.09 2.97 3.07 3.04
d (mm) 6.71 5.86 6.53 6.66 6.89 6.65 6.82 6.72 6.70
De / t 33.05 33.53 33.36 33.36 32.68 33.25 33.08 33.08 33.25
d/t 0.68 0.61 0.67 0.69 0.70 0.68 0.70 0.69 0.69
Note: d = t – t*

Table 4 – Actual and predicted failure pressures


Failure Pressure (MPa)
Specimen laboratory 085dL Eff. Area B31G RP-F101 PB
test equation equation equation equation equation
TS 5.1 14.401 15.776 12.466 9.487 14.257 14.994
TS 1.2 14.068 15.240 11.928 9.456 13.496 13.950
TS 2.2 13.577 15.323 12.052 9.763 13.478 13.692
TS 2.1 12.841 14.793 11.439 9.364 12.673 12.626
TS 3.1 12.125 14.721 11.266 9.272 12.396 12.155
TS 1.1 11.919 14.697 11.315 9.487 12.459 12.037
TS 3.2 11.909 14.337 10.915 9.118 11.937 11.433
TS 4.1 11.988 14.579 11.232 9.425 12.264 11.713
TS 4.2 11.301 14.385 10.984 9.333 12.045 11.408

Table 5 – Ratios of the predicted to the actual failure pressure


Predicted / Actual Failure Pressure
Specimen 085dL Eff. Area B31G RP-F101 PB
equation equation equation equation equation
TS 5.1 1.095 0.866 0.659 0.990 1.041
TS 1.2 1.083 0.848 0.672 0.959 0.992
TS 2.2 1.129 0.888 0.719 0.993 1.008
TS 2.1 1.152 0.891 0.729 0.987 0.983
TS 3.1 1.214 0.929 0.765 1.022 1.002
TS 1.1 1.233 0.949 0.796 1.045 1.010
TS 3.2 1.204 0.916 0.766 1.002 0.960
TS 4.1 1.216 0.937 0.786 1.023 0.977
TS 4.2 1.273 0.972 0.826 1.066 1.009
mean 1.178 0.911 0.746 1.010 0.998

6
Table 6 – Failure pressure prediction statistics

ERROR (%)
Specimen 085dL Eff. Area B31G RP-F101 PB
equation equation equation equation equation
TS 5.1 9.5 13.4 34.1 1.0 4.1
TS 1.2 8.3 15.2 32.8 4.1 0.8
TS 2.2 12.9 11.2 28.1 0.7 0.8
TS 2.1 15.2 10.9 27.1 1.3 1.7
TS 3.1 21.4 7.1 23.5 2.2 0.2
TS 1.1 23.3 5.1 20.4 4.5 1.0
TS 3.2 20.4 8.4 23.4 0.2 4.0
TS 4.1 21.6 6.3 21.4 2.3 2.3
TS 4.2 27.3 2.8 17.4 6.6 0.9
mean 17.8 8.9 25.4 2.6 1.8
std dev 6.6 4.1 5.6 2.1 1.4
minimum 8.3 2.8 17.4 0.2 0.2
maximum 27.3 15.2 34.1 6.6 4.1

16
17 UNSAFE

UNSAFE
16
15
Predicted Failure Pressure (MPa)

15
Predicted Failure Pressure (MPa)

14 14 1:1 line
DNV RP-F101 equation
13 1:1 line PB equation
085dL equation
12 Effective Area equation
13
ASME B31G equation
DNV RP-F101 equation
11
12
10

9
11
8

SAFE
7
SAFE 10
6 10 11 12 13 14 15 16
6 7 8 9 10 11 12 13 14 15 16 17 Actual Failure Pressure (MPa)
Actual Failure Pressure (MPa)
Figure 2 – Comparison of the actual failure pressures with the
Figure 1 – Comparison of the actual failure pressures with the failure pressures predicted by the equation of the
failure pressures predicted by the equations of the DNV RP-F101 method and by the PB equation
four methods investigated

7
34.0 1.40

32.0 1.30

30.0 1.20

28.0 1.10

Predicted / Actual Failure Pressure


Laboratory tests

26.0 DNV RP-F101 equation 1.00


Failure Pressure (MPa)

PB equation
24.0 lower bound
0.90

22.0 0.80

20.0 0.70

18.0 0.60
085dL equation
16.0 0.50 Effective Area equation
ASME B31G equation
14.0 0.40
DNV RP-F101 equation
12.0 0.30 PB equation

10.0 0.20

8.0 0.10

6.0 0.00

0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0 18.0 4.00 5.00 6.00 7.00 8.00 9.00 10.00 11.00
L / sqrt (De t) L / sqrt (De t)

Figure 3 – Curves of the failure pressure as a function of the Figure 5 – Curves of the ratios of the predicted to the actual
dimensionless length failure pressures as a function of the
dimensionless length

17.0

16.0

15.0

14.0
Failure Pressure (MPa)

13.0

12.0

11.0

10.0
Laboratory tests
9.0 DNV RP-F101 equation
PB equation
8.0 lower bound

7.0

4.0 5.0 6.0 7.0 8.0 9.0 10.0


L / sqrt (De t)

Figure 4 – Detail of the curves of the failure pressure as a


function of the dimensionless length

View publication stats

You might also like