Sei sulla pagina 1di 7

International Journal of Pressure Vessels and Piping 85 (2008) 782788

Contents lists available at ScienceDirect

International Journal of Pressure Vessels and Piping


journal homepage: www.elsevier.com/locate/ijpvp

Analysis of a carbon composite overwrap pipeline repair system


J.M. Duell a, J.M. Wilson a, M.R. Kessler b, *
a
Department of Mechanical Engineering, University of Tulsa, Tulsa, OK 74104, USA
b
Department of Materials Science and Engineering, Iowa State University, Ames, IA 50011, USA

a r t i c l e i n f o a b s t r a c t

Article history: A relatively new method has been developed to stop external corrosion and structurally reinforce steel
Received 14 June 2007 pipes by external wrapping of damaged sections using bre reinforced polymer (FRP) materials. Several
Received in revised form 28 June 2008 different defect geometries representing corrosion patches on steel pipe were characterized using nite-
Accepted 4 August 2008
element analysis, by changing the circumferential length of the defect. Pipe vessels containing these
defects along with the composite structural repairs were modeled and the results were compared to eld
Keywords:
tests to determine the effectiveness of the repairs. It was found that the defect width around the
Composite materials
circumference had little impact on the ultimate rupture pressure of the repaired vessel, but inuenced
Finite-element method
Pipelines the stress state in the underlying pipe substrate.
Rehabilitation 2008 Elsevier Ltd. All rights reserved.

1. Introduction completely [3]. FRP repair systems also slow the external corrosion
growth by shielding the damage from the environment while the
Over 1.7 million km of gas, crude oil, and petroleum product pipeline stays in service [4]. Additionally, FRPs have become a cost
pipeline exists throughout the world [1]. Many of these pipelines effective option in the repair of other decient piping such as
have been in service since the 1940s and 1950s [2]. Consequently, corroded sewer lines [5], municipal water supplies, and process
every year between $2 and $3.3 billion dollars in the United States piping and pipework. Common, FRP systems used for reinforcing
alone is lost to corrosion in gas and petroleum pipelines that need pipe walls include glass bre, aramid bre, and carbon bre
to be repaired or replaced [3]. The large amount of external pipeline composites. Toutanji and Dempsey [6] showed that carbon bre
corrosion ultimately increases the cost of energy. Traditionally reinforced polymers (CFRPs) perform better than those reinforced
pipelines with severe corrosion problems have to either be taken with glass or aramid bres in that they improve the ultimate
ofine while the damaged pipe segment is removed and internal pressure capacity of pipes, signicantly enhancing the
a replacement joint is welded in place or reinforced with a sleeve strength, durability, and corrosive properties of the system.
repair where steel sleeves are manufactured in two parts and are In order to evaluate the effectiveness of a given repair system it is
placed over the pipeline and subsequently welded, or bolted to common to rst manufacture a defect into a pristine pipe specimen,
each other along the longitudinal seam [1]. then secondly repair the damaged region, and nally pressurize the
Recently, bre reinforced polymer (FRP) matrix composite pipe monotonically or under cyclic conditions until failure occurs.
overwrap systems have emerged as an alternative repair system. The size and shape of the defect region can have a signicant effect
Repairs made with FRP materials offer distinct advantages over on the level of repair that can be achieved. In this paper we therefore
traditional, welded repairs in addition to reduced cost. Primary evaluate by nite-element analysis four different defect geometries
benets associated with FRP piping repair systems include: 1) the representing corrosion patches on steel pipe given by changes to the
short amount of time needed to complete a repair, 2) the undis- circumferential length of the defect. Pipe vessels containing these
rupted uid transmission in the piping system while the repair is defects along with the FRP repairs were modeled and the computed
made, and 3) explosion potential is eliminated since no welding or failure pressures were compared to eld tests to determine
cutting of the pipeline is required. the effectiveness of the repairs, and the inuence that circumfer-
Industry analysis shows that such repair systems are, on ential defect length has on the efciency of the repair system.
average, 24% cheaper than welded steel sleeve repairs and 73%
cheaper than replacing the damaged section of the steel pipe
2. Experimental repair and pressure testing methods

* Corresponding author. Tel.: 1 515 294 3101; fax: 1 515 294 5444. Three types of materials were used to fabricate and evaluate
E-mail address: mkessler@iastate.edu (M.R. Kessler). repaired pressure vessels for rupture testing: a segment of

0308-0161/$ see front matter 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijpvp.2008.08.001
J.M. Duell et al. / International Journal of Pressure Vessels and Piping 85 (2008) 782788 783

Nomenclature t nominal wall thickness of original pipe (mm)


trepair design repair thickness (mm)
D external pipe diameter (mm) ts minimum remaining wall thickness of the pipe, m (in.)
Eav average modulus of the substrate/composite 3 total strain
combination (MPa) 3c allowable circumferential strain
Ec tensile modulus for the composite laminate in the 3elastic elastic strain in the composite laminate
circumferential direction (MPa) 3plastic extra strain in the composite after yield
Es tensile modulus for steel (or pipe material) (MPa) FRP bre reinforced polymer
P internal design pressure (MPa) CFRP carbon bre reinforced polymer
Plive internal pressure of the pipe at the time of repair (MPa) DGEBA diglycidyl ether of bisphenol-A
Pyield internal pressure when the substrate pipe begins to FE nite-element
yield (MPa) FEA nite-element analysis
s Specied Minimum Yield Strength (SMYS) of the pipe ASME American Society of Mechanical Engineers
(MPa)

pipeline steel, an epoxy putty (to ll defect regions in the pipe), structure to allow the repair system to be applied. First, the two-
and a carbon bre/epoxy composite wrap. The pipeline steel was part high viscosity putty was mixed using a power drill and
ASTM A-106 Grade B, a seamless, plain carbon steel that is a paddle mixer and applied to ll the defect (as shown in Fig. 3a).
commonly found in the pipeline industry. The epoxy putty and The carbon fabric was impregnated with the premixed low
resin, the carbon bre, and application tools to apply the repair viscosity epoxy/amine prepolymer using an adhesive roller. After
are shown in Fig. 1. The putty that is used to ll the defects on the the fabric was properly saturated, it was wrapped around the test
pipe, restoring its undamaged dimensions is diglycidyl ether of vessel and the freshly applied putty using hand tension to pull the
bisphenol-A (DGEBA) based epoxide cured with an alphatic wet fabric, while keeping the centerline of the wrap with the center
amine hardener and a thixotropic fumed silica additive. The of the defect. This application procedure is shown in Fig. 3b. As the
carbon bre/epoxy wrap that provides the structural reinforce- wrap was being applied, air bubbles were forced out of the system
ment is created by impregnating a woven carbon bre fabric with by hand. A total of six layers of carbon/epoxy wrap were used to
a similar unthickened DGEBA epoxy/amine prepolymer resin. The cover the aw, giving the repair (excluding the epoxy putty)
plane weave carbon fabric consists of 12 K tow in the warp a thickness of 3.1 mm (0.122 in). Once the wrap was applied the
direction and 6 K tow in the ll direction. The 12 K indicates that epoxy was allowed to cure for at least 24 h, in a room temperature
there are approximately 12,000 carbon laments (bres) in each environment, before testing began. A picture of a completed vessel
bundle (tow), while 6 K indicates approximately 6000 laments is shown in Fig. 4.
per tow. The rupture testing consisted of rst lling each wrapped
Test vessels were prepared from 6 in. nominal diameter, pressure vessel with water vertically to insure no air was present
Schedule 40, steel pipes cut into 1.52 m (5 ft) lengths with welded within the system. Next, the vessel was moved to a secure area to
end-caps, each having a 9.5 mm (0.375 in.) NPT tting. The prevent injury to personnel. An air actuated, hydraulic power unit
Schedule 40 designation corresponds to a wall thickness (t) of was used to pressurize the test vessel. The hydraulic pump was
7.11 mm (0.280 in.) and an outer diameter (D) of 168.3 mm connected to the vessel and a transducer was used to record the
(6.625 in.). Simulated defects were machined into the pipe wall pressure in the line. The pump forces water into the pressure vessel
using defects with a longitudinal length of 15.24 cm (6 in.), and until rupture occurs. The internal pressure required to fail the
a depth to 50% of the wall thickness (3.56 mm or 0.14 in.). In the vessel was recorded. Pressurization to failure takes 1015 min.
hoop dimension defect lengths are 15.24 (6 in.) or completely
around the circumference. These defects, shown in Fig. 2, are 3. Finite-element modeling methodology
referred to as 6  6 patch and axisymmetric respectively.
Once the defects were created in pipe test vessels, the steel was In order to evaluate the changing stress eld with increasing
sandblasted to a near-white nish and placed on a support pressure levels for the repaired pipe vessels, several nonlinear

Fig. 1. Pipe repair application kit. (a) Reinforcement and application tools. (b) Two-part putty and epoxy used to ll defects and wet out carbon bre wrap.
784 J.M. Duell et al. / International Journal of Pressure Vessels and Piping 85 (2008) 782788

Mises yield theory and the associated ow rule as the classical


metal plasticity model. The putty was modeled with a bilinear
elastic stressstrain behavior, with an experimentally determined
ca. 50% drop in modulus occurring at the yield stress of 33.1 MPa.
The orthotropic woven composite wrap required nine elastic
constants to model the linear elastic behavior of the carbon/epoxy
material. These values were determined by a combination of direct
measurements, literature values, and rule of mixtures estimates as
detailed by Duell (2004), and they are listed in Table 1. The ultimate
strength of the composite overwrap was found experimentally to
be 576 MPa (83.6 ksi)  12% (uncertainty is for 95% condence) for
the 12 K hoop direction [7].
A constant pressure loading condition (slightly higher than the
expected failure pressures) was applied incrementally on all of the
surface areas of the inside wall of the pipe. Symmetry boundary
conditions were applied at the appropriate planes of symmetry
shown in Fig. 5. The model was then solved iteratively using
automatic timestepping (saving the results from every substep).
The maximum stresses at each substep were analyzed to estimate
the pressure level necessary for a critical (ultimate) stress to be
reached assuming a maximum stress failure criterion in the
composite or a von Mises base failure criterion in the steel. The
vessel can be considered as a burst failure when the von Mises
equivalent stress in the steel in the FEA simulation reaches the true
Fig. 2. Pipe test vessels with machined aws with a depth of 50% wall thickness for (a) ultimate tensile stress (UTS) or when the maximum stress in the
axisymmetric and (b) 6  6 patch defects. composite exceed its ultimate strength in one of the principle
material directions (576 MPa in the hoop direction). The applied
pressure was then rened on the model for subsequent analysis
three-dimensional nite-element calculations were performed, for based on these estimates to predict the ultimate failure pressure of
four different defect geometries, a 1  6 patch, a 3  6 patch, a 6  6 the pipe test vessel.
patch, and axisymmetric. In all cases, the defect length (along the
length of the pipe) is 15.24 cm (6 in.) and the defect depth is 50% of 4. Results and discussion
the wall thickness, while the hoop dimension defect lengths are
2.54 cm (1 in.), 7.62 cm (3 in.), 15.24 (6 in.) and completely around Representative results from the nite-element simulation for
the circumference respectively. A portion of the repaired pipe each of the four defect geometries at predicted failure pressures are
specimen was discretized making use of three planes of symmetry shown in Fig. 7. In all cases, the highest failure inducing stresses
of the specimen geometry and loading conditions as shown in occurred at the center of the defect region. A through thickness plot
Fig. 5. The ANSYS v8.0 nite-element software package was used to of the three components of stress (radial, hoop, and axial) at the
create the model, generate the meshes (using ten node tetrahedral center of each defect is plotted along with the corresponding
solid elements), and to perform the FE calculations. The predicted contour plot of the hoop stress in the general defect region. A
burst pressures from two of these models (6  6 patch, and thickness distance of 0 mm corresponds to a point on the inside
axisymmetric) will be compared to experimental burst results. steel surface of the pipe vessel under the center of the defect, while
The material properties of the CFRP, the steel, and the ller putty the maximum distance (11.18 mm) corresponds to the outside
were found through an extensive materials characterization surface of the composite wrap over the center of the defect. There
program in Ref. [7]. This series of tests involved coupons for tensile, are two discontinuities in the stress versus thickness plots as the
shear, and compression. A summary of the material properties used material changes from steel, to putty, to CFRP.
as input in the nite-element modeling is listed in Table 1. The The failure pressure for all four cases was determined as the
elastic/plastic stressstrain behavior of the pipeline steel is shown value that causes a stress based failure criterion to be met. In this
in Fig. 6. Individual data points from the stressstrain curve were case, that failure criterion means that failure of the pressure vessel
converted to true stress and used as input to the nonlinear material occurs at the lowest pressure that causes the von Mises stress to
constitutive model for the steel. The FEA code assumes the von exceed the ultimate strength of the steel (maximum-distortion-

Fig. 3. Pipe repair process (a) lling defect with putty (b) wrapping epoxy wetted carbon fabric around the defect.
J.M. Duell et al. / International Journal of Pressure Vessels and Piping 85 (2008) 782788 785

Table 1
Material properties used in nite-element models

Material Material properties

Linear (below yield) Nonlinear (above yield)


Steel (ASTM Youngs Modulus, E Yield Stress, sY
A106 Grade B) 207 GPa (30 Msi) 300 MPa (43.5 ksi)
Poissons Ratio, n Above yield stress, individual
0.30 values from the experimental
tension tests (shown in Fig. 5)
were used as input for a
nonlinear elastic-plastic
model
Putty Youngs Modulus, E Yield Stress, sY
(Thickened Epoxy) 1.74 GPa (252 ksi) 33 MPa (4.80 ksi)
Fig. 4. Repaired pipe test vessel. Poissons Ratio, n Tangent Modulus, Etan
0.45 0.87 GPa (126 ksi)

Composite (Woven Youngs Modulus,a N/A


energy failure theory), or a principal stress in the composite or
Carbon Reinforced Ex 5.5 GPa (0.80 Msi)
putty exceeds the ultimate tensile or compressive strength of that Epoxy) Ey 23.4 GPa (3.40 Msi)
material (maximum-normal-stress failure theory). Ez 49.0 GPa (7.10 Msi)
In all cases, the failure criterion is reached rst for the composite Poissons Ratioa
wrap, so the maximum stress in the FRP is the same for all four of nxy 0.430
nxz 0.196
the graphs shown in Fig. 7, even though the failure pressures are nyz 0.430
not the same. The maximum stress to cause failure in the composite Shear Modulusa
was the hoop stress (at 576 MPa) for all four simulations, and this Gxy 0.69 GPa (0.10 Msi)
maximum occurs at the inside edge of the composite (adjacent to Gxz 29.6 GPa (4.00 Msi)
Gyz 0.69 GPa (0.10 Msi)
the putty).
The predicted burst pressures for the four different repaired a
The x, y, and z directions correspond to the radial, axial, and hoop directions
defects are listed in Table 2. The predicted burst pressures do not respectively.
vary only slightly for the four different systems, with just 2.2%
variation between the maximum predicted burst pressure, and the
minimum predicted burst pressure. The results from real burst tests region, beyond the yielding point, but below the ultimate strength
using pressure vessel specimens with axisymmetric and 6  6 (see Fig. 6).
defects are given in Table 2 for comparison. Excellent agreement is One consequence of this higher hoop stress in the steel at the
found between the experiments and the nite-element model center of the narrower defect (the 1  6 defect) is that the fatigue
predicted burst pressures. Experimental burst data for repaired life of a pressure cycled vessel is expected to be lower than the
vessels with the 3  6 and 1  6 specimen are not available. broader shaped defects (e.g. 6  6). Recently Wilson et al.[8]
As shown in Fig. 7, the amount of change in the hoop stress conrmed this when they reported the results of fatigue tests in
through the composite wrap (at the center of the defect) is much which the internal pressure was cycled from 1.73 MPa (250 psi) to
higher as the defect width decreases. Thus, the hoop stress for the 16.6 MPa (2400 psi) every 10 s. This extreme pressure variation
1  6 defect decreases by over 26% from the inside of the composite (which ranged from 13% to 113% of the maximum allowable oper-
wrap to the outside, while for the 6  6 and the axisymmetric ating pressure of the piping) resulted in fatigue failures in repaired
defects the decrease in hoop stress through the wrap is only specimens containing both 1  6 and 6  6 style defects (although
about 3%. with a wall loss of 80% compared to the 50% wall loss modeled in
To compensate for the reduced load carried by the composite for this study). The wrapped pressure vessel with the 1  6 defect
the 1  6 defect, a higher load is required in the steel, hence an failed after 900 cycles, while the wrapped pressure vessels with the
initial (inside wall) hoop stress value in the steel of nearly 500 MPa 6  6 defect failed after an average of 3299 cycles. In all cases the
at the defect center for the 1  6 defect compared to initial hoop failure mode in the fatigue testing was a hoop-type failure, sepa-
stress values of 440, 450, and 460 MPa for the axisymmetric, 6  6, ration along the axial direction of the pipe, in the steel substrate in
and 3  6 defects respectively. For all four defect types, the stress the defect. The composite overwrap, however, remained intact,
state in the steel at the center of the defect is in the strain hardening with subsequent leaking occurring at the edge of the overwrap.
In great contrast, when the repaired vessels were pressurized in
monotonic static loading, the vessels burst violently, with the wrap
exploding apart and the steel pressure vessel rupturing along
a longitudinal crack running the length of the defect region as
shown in Fig. 8. Another difference between specimens with the
four different defect types is the increasing degree of non-
symmetry in the stress contours around the defect region as the
defect becomes narrower. This is clearly seen in hoop stress contour
plots on the outside surface of the composite wraps in Fig. 7. One
consequence of this non-symmetry in the composite stress eld is
that a larger portion of the composite is subjected to large near
failure stresses in the axisymmetric and 6  6 defect than the 3  6
and 1  6 systems. This makes the systems with larger defects more
susceptible to aws or material variations within the composite
Fig. 5. The full 3D Model used for the nite-element calculations and a zoomed in
view showing the discretized nite-elements at the defect region (the elements are not wrap than the repaired pipes with smaller defect regions, which
shown in the full 3D model). could result in lower ultimate burst pressures.
786 J.M. Duell et al. / International Journal of Pressure Vessels and Piping 85 (2008) 782788

400
Ec trepair Es ts
Eav  (2)
trepair ts
300
where Es is the elastic modulus of the steel. Then the elastic strain,
Stress (MPa)

3elastic, in the composite laminate is found as follows:


200
1 PD PD
3elastic   (3)
Eav 2 trepair ts 2 Ec trepair Es ts
100
When the substrate pipe begins to yield, the internal pressure,
Pyield, is found by rearranging Eq. (3) with P Pyield:
0
0 0.0025 0.005  
23elastic Ec trepair Es ts 2s Ec trepair Es ts
Strain Pyield (4)
600 D DEs
Assuming the pipe steel is elasticperfectly plastic, as the pressure
rises above Pyield the substrate pipe carries no further load and any
500
further load is carried solely by the composite. Therefore the extra
strain, 3plastic, in the composite after yield is given by:
400  
P  Pyield D
Stress (MPa)

3plastic (5)
300
2Ec trepair

Therefore the total hoop strain, 3, as a function of pressure is given


by the sum of the strains before and after the yield pressure is
200
reached as follows:

100
Pyield D P  Pyield D
3 3elastic 3plastic  (6)
2 Ec trepair Es ts 2Ec trepair

0 Equating the total strain, 3, to the design allowable strain of the


0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 composite, 3c, the minimum thickness of the repair can be derived
Strain from the following equation to yield Eq. (1):
Fig. 6. Elastic/Plastic stressstrain curve of the A106 steel in tension used to model
PD ts
nonlinear pipe behavior. 3c s (7)
2Ec trepair Ec trepair
The effectiveness of the repair can be estimated by comparing
the burst pressure of the repaired vessels (w44 MPa) with the In practice, when repairs are carried out at pressure, Eq. (7) must
predicted burst pressure of an unrepaired vessel. The burst pres- be modied to account for the internal pressure of the pipe at the
sure of an unrepaired vessel can be made using the ASME B31G time of repair (Plive) as follows:
criterion [9] or using the RSTRENG technique [10], both well
PD sts Plive D
established methods for assessing corroded pipelines (with the 3c    (8)
2Ec trepair Ec trepair 2 Ec trepair Es ts
ASME B31G criterion known to be overly conservative). For the
defects analyzed in the study, these methods (which do not take In this case, the design repair laminate thickness, trepair, may be
into account the circumferential length of the defect) result in calculated by iteration using Eq. (8).
a predicted burst pressure of 22.4 MPa (3.25 ksi) and 26.2 MPa If Eq. (1) is used for the pipe system evaluated in this study (i.e.
(3.80 ksi) for the ASME B31G criterion and the RSTRENG technique D 168.3 mm, ts 3.56 mm, Ec3c 576 MPa, s 300 MPa) and if
respectively [7]. a similar failure pressure is assumed (i.e. P 44 MPa) then the
calculated minimum repair thickness, trepair, is found to be
5. Comparison with ASME repair standard 4.57 mm. This is higher than the 3.1 mm for the composite repair
used in the FEA models to obtain the same predicted burst pres-
A recently developed guideline for determining the thickness of sures. Of course, several assumptions are inherent in the derivation
a composite overwrap for pipe defects has been developed as an of Eq. (1) which are not made in the FEA model. First, the
ASME Post Construction Repair Standard [11]. In this standard the assumption that the substrate material is elastic, perfectly plastic
minimum repair thickness, trepair, assuming the repair is applied at (i.e. no strain hardening) is overly simplied as shown by the
zero internal pressure, is determined by: substantial increase in stress in the stress curve beyond yield for the
  steel pipe sample shown in Fig. 6. The omission of strain hardening
1 PD
trepair  sts (1) from the derivation of Eq. (1) is likely the primary cause of the
3c Ec 2
ASME standard over predicting the minimum repair thickness.
where 3c is the allowable circumferential strain, Ec is the tensile Also, the axial stress that is generated by the repair is neglected in
modulus for the composite laminate in the circumferential direc- the closed-form solution. This may be a valid assumption for cases
tion, P is the internal design pressure, D is the external pipe where signicant constraints are placed on the pipe wall in the
diameter, s is the Specied Minimum Yield Strength (SMYS) of the axial direction, such as pipeline components that may limit axial
pipe, and ts is the minimum remaining wall thickness of the pipe. movement or soil friction in buried pipelines. However, it cannot be
Eq. (1) is derived by rst assuming an average modulus of the neglected in the test specimens with welded end-caps that we
substrate/composite combination (Eav): tested and modeled using nite-element analysis. Likewise, Eq. (1)
J.M. Duell et al. / International Journal of Pressure Vessels and Piping 85 (2008) 782788 787

Radial Stress
Hoop Stress
Axial Stress
a b
Axisymmetric Defect 6x6 Defect
700 700
Steel Putty Composite Steel Putty Composite
600 600
500 500
Stress (MPa)

Stress (MPa)
400 400
300 300
200 200
100 100
0 0
-100 -100
0 2 4 6 8 10 0 2 4 6 8 10
Distance (mm) Distance (mm)

14.6 MPa 572 MPa -4.9 MPa 584 MPa

c d
3x6 Defect 1x6 Defect
700
Steel Putty Composite 700
600 Steel Putty Composite
600
500
500
Stress (MPa)

400
Stress (MPa)

400
300
300
200
200
100
100
0
0
-100
0 2 4 6 8 10 -100
0 2 4 6 8 10
Distance (mm)
Distance (mm)

-5.1 MPa 639 MPa 1.1 MPa 618 MPa

Fig. 7. FEA predicted radial, hoop, and axial stress at the center of the 50% wall loss defect from the inside of the pipe wall to the outside of the composite wrap at burst pressure.
Contour plots show corresponding distribution of hoop stress for (a) axisymmetric defect, Pburst 43.8 MPa (b) 6  6 defect, Pburst 44.7 MPa (c) 3  6 defect, Pburst 44.8 MPa (d)
1  6 defect, Pburst 43.8 MPa.

does not account for the defect geometry (i.e. length or width); only
Table 2
the minimum remaining wall thickness (of the substrate) is used to Comparison of repaired vessels
infer the internal pressure at the point of substrate yield. Finally, the
presence of the ller putty is not considered in the closed-form Defect type width  length (inches)  50% thickness Burst pressure, MPa (ksi)

expression, while the minimal reinforcing from the putty is FEA predicted Tested
included in the FEA model. Axisymmetric 43.8 (6.35) 43.8 (6.35)
While these assumptions are very reasonable to obtain 66 44.7 (6.48) 43.1 (6.25)
a conservative closed-form solution to use for the minimum repair 36 44.8 (6.50) N/A
16 43.8 (6.35) N/A
thickness, trepair, in the ASME standard, the FEA results are able to
788 J.M. Duell et al. / International Journal of Pressure Vessels and Piping 85 (2008) 782788

hoop direction had little impact on the failure pressure estimated


using nite-element techniques. Defect length along the axial
direction should be examined to determine what inuence it has
on the failure pressure of vessels repaired with a composite
overwrap. In the future, this work could be extended to charac-
terize more defect sizes, as well as to create a model to determine
where repaired vessels would rupture from excessive internal
pressure.

Acknowledgements

Financial support from the Oklahoma Center for the Advance-


ment of Science and Technology project number AR03(1)-050 is
gratefully acknowledged. The technical advice and helpful discus-
sions with Roger Walker, Will Goertzen, Dixit Kadakia, and Alan
Morton are gratefully acknowledged.

References
Fig. 8. Example of a ruptured vessel with an axisymmetric defect.
[1] Mohitpour M, Golshan H, Murray A. Pipeline design and construction: a prac-
tical approach. 2nd ed. New York, NY: ASME Press; 2003. p. 499518.
[2] Chapetti MD, Otegaui JL, Manfredi C, Martins CF. Full scale experimental
predict the actual burst pressure with greater accuracy (with less
analysis of stress states in sleeve repairs of gas pipelines. International Journal
than 4% variation between the numerical and experimental burst of Pressure Vessels and Piping 2001;78:37987.
pressures, as shown in Table 2). Of course, when designing the [3] Koch GH, Brongers MP, Tompson NG, Virmani YP, Payer JH. Corrosion cost and
repair system, it is important to consider the large uncertainty in preventative strategies in the United States. Federal Highway Administration,
Ofce of Infrastructure Research and Development; 2001. p. 260311.
the strength of the composite material, and material and bond [4] Cuthill J. Advances in materials, methods, help gain new users. Pipeline & Gas
degradation that can occur with time and at elevated temperature Journal Nov 2002;229(11):646.
when subjected to stresses over an extended period of time. [5] Marsh G. Composites renovate deteriorating sewers. Reinforced Plastics
2004;48(6):204.
Goertzen and Kessler [12] recently evaluated the creep behavior for [6] Toutanji H, Dempsey S. Stress modeling of pipelines strengthened with
this type of carbon bre/epoxy matrix composite and estimated, advanced composite materials. Thin-Walled Structures 2001;39:15365.
using time-temperature-superposition, that the stress levels to [7] Duell JM. Characterization and FEA of a carbon composite overwrap repair
system. Masters thesis, University of Tulsa Graduate School; 2004.
induce failure at 50 years ranged from 84% at 30  C to 42% at 50  C [8] Wilson JM, Kessler MR, Walker RH, Duell JM, Kadakia D, Sousa N. Fatigue testing
of the ultimate tensile strength; though elevated stress rupture of steel pipes repaired with carbon/epoxy composites. In: Proceedings of the
experiments are needed to validate the time-temperature-super- 2005 Rio Pipeline Conference and Exposition, Rio de Janeiro, Brazil; Oct. 2005.
Paper IBP1089_05.
position work. For this reason signicant design margins should be
[9] ASME. B31G manual for assessing remaining strength of corroded pipes. New
included in any design. York, NY: American Society of Mechanical Engineers; 1996.
[10] Kiefner JF, Vieth PH. A modied criterion for evaluating the remaining
strength of corroded pipe. Final report on project PR3-805. Columbus, OH:
6. Conclusions
Battelle Memorial Institute; 1989.
[11] ASME. Post construction standard, PCC-2, repair of pressure equipment and
The results from the nite-element analysis compare well with piping, article 4.1. Non-metallic composite repair systems for pipelines and
experimental rupture tests. The full-scale evaluations performed pipework: high risk applications. New York, NY: American Society of
Mechanical Engineers; 2006.
on test vessels with machined defects failed at pressures predicted [12] Goertzen WK, Kessler MR. Creep behavior of carbon ber/epoxy matrix
by the nite-element simulations. Varying the defect length in the composites. Materials Science and Engineering A 2006;421:21725.

Potrebbero piacerti anche