Design,testing and field deployment of a composite clamp for pipeline repairs

Main components

A composite clamp comprises two half cylindrical shell sections with flanges, which are brought together over a pipe and fastened using bolts to effect the repair, see Fig. 1a. The main components of the clamp are clearly shown in Fig. 1(b–c). Note that the pressure ports used for leak testing after clamp manufacture and installation, and for grouting, are optional. Pressurised fluid leaking from the pipe is contained in the annulus between the pipe and the clamp using hard rubber seals, such as Nitrile Butadiene Rubber (NBR) used in this study, located within grooves machined into the composite at both ends of the clamp and along the flanges. The only contact between the repair system and the pipe is from the NBR seals.

OPEN IN VIEWER

1

a The composite clamp; b main components of the assembly; c internal view showing the ‘picture frame’ seals; and d general dimensions of shell thickness (S), flange thickness (F) and internal diameter (D_i)

The clamp can be fabricated using a number of composite processing techniques; however vacuum-assisted liquid moulding is deemed the most cost-effective to achieve a balance of good quality laminates while still keeping the clamp affordable. Leak containment is achieved through the use of rubber seals that are suitable for the service conditions and environments. The seal is applied in a ‘picture frame’ configuration as shown in Fig. 1 c. Hence, each half-clamp has an associated fully enclosed semi-circular annulus, such that fluid (leaking from a damaged pipe) is contained within each half-clamp. Finally, the two half-clamps are held together using socket head capscrews. The flanges are thickened in order to allow sufficient resistance to localised compressive loads applied by bolts during tightening, as well as to increase the stiffness of the flange in order to resist the tendency to pry open during subsequent pressurisation when in service.

The general dimensions of the clamp are shown in Fig. 1 d: shell thickness (S), flange thickness (F) and internal diameter (D_i).

Design specifications

The 7 MPa (1000 psi) clamp is one of the common industry pipeline metal clamps being offered. 7 MPa refers to the maximum allowable working pressure (MAWP) of the intended pipe to be repaired. Similarly, the composite clamp developed here is for an MAWP of 7 MPa, with a factor of safety of 1.5, hence a design pressure of 10.5 MPa (1500 psi).

The temperature limit of the composite clamp is determined by the resin used. Maximum service temperature is specified based on industrial standards ISO/TS 24817 ¹³ and ASME PCC-2 ² , which specify it to be 30°C less than the glass transition temperature (T_g) of the resin system used to fabricate the clamp for through-wall defects. T_g refers to a temperature range where a thermosetting polymer changes from a hard or ‘glassy’ state (below T_g), to a more compliant or ‘rubbery’ state (above T_g). Note that the rubber seals used for the composite clamp would also have an allowable temperature well above the maximum service temperature.

Other specifications are application dependent, like the length and diameter of the repair.

Clamp thickness

An estimate of the shell thickness is calculated assuming only the hoop strain component. The shell thickness, S, of the clamp that can achieve a hoop strain, ε, which is below the maximum strain limit, is initially defined by the following equation: (1)

where

D_i is the internal diameter of the clamp

P_i is the internal pressure (design pressure)

E is the circumferential modulus of elasticity of the clamp material, 17 GPa.

The 7 MPa clamp under study is designed for repair of an 8-in nominal diameter pipeline. Based on the calculations and assumptions above, it has a 35 mm shell thickness and an inner mould line diameter (D_i) of 223 mm. A 2 mm clearance gap between clamp inner mould line and pipe (i.e. an annulus) is specified for tolerance purposes. These dimensions are further verified through FEA optimisation as detailed in the sub-section ‘FEA optimisation’.

The clamp thickness is one of the main components that influences the radial expansion of the clamp under pressure. Right balance between shell and flange thicknesses is critical to ensure that the deflections of the clamp do not result in excessive strains in the structure (due to radial expansion of the clamp) and/or out-of-plane displacements that would otherwise promote seal extrusion. Some notable trends are observed here. A thicker clamp would generally reduce the likelihood of seal extrusion, as illustrated in Fig. 3 a, which is based on FEA predictions of clamps with varying thicknesses but with a fixed inner diameter and a fixed flange-to-shell thickness ratio. The increase in shell thickness results in decrease in magnitude of radial expansion, but the effect plateaus at higher thickness values. The same trend is observed for clamps with different internal pressures, viz. 10.5 MPa (1.5 MAWP) and 16.5 MPa (2.35 MAWP), but with the higher pressure requiring increasingly thicker clamps to maintain the radial expansion at the same magnitude of the lower pressure clamps.

OPEN IN VIEWER

3

a Evolution of clamp radial expansion with clamp shell thickness; b Evolution of clamp radial expansion with flange-to-shell thickness ratios; c Evolution of clamp radial expansion with bolt location with respect to inner mould line of the clamp

When the flange thickness is increased while maintaining shell thickness, thus increasing the flange-to-shell thickness ratio, the radial expansion shows a downward trend. This is illustrated in Fig. 3 b, for shell thicknesses of 35 mm and 74 mm. While the effect of the flange-to-shell thickness ratio on the radial expansion is more pronounced for thinner clamp shells, it is less significant than the effect of having a thicker shell.

Practically, it is important to understand that simply thickening the clamp will also add weight and bulk that may prevent its application in the field, especially in locations where space is limited. Another significant parameter that affects clamp radial expansion is bolt location. Figure 3 c shows the evolution of clamp radial expansion when the bolt location is varied with respect to the inner mould line of the clamp. Note that the bolt locations are non-dimensionalised such that the values show multiples of a baseline minimum bolt location. When the bolt location is relatively close to the inner mould line of the clamp, the effect of flange thickness is negligible. Its effect only becomes significant as the bolts moved further out, whereby increasingly thicker flanges become necessary in order to reduce clamp radial expansion. At the same time, however, pushing the bolt location further out significantly exacerbates seal extrusion at the groove, and this has a more dominant effect compared with changing the flange thickness. This emphasises the role of the bolt location in the clamp design.

It should be highlighted that having the row of bolts too close to the inner mould line of the clamp necessitates cutting into the outer diameter of the shell, thus weakening of the structure, or requiring additional thickening of the flange, which would then make the clamp more bulky overall.

Hence, the final dimensions of the clamp are an optimised balance between shell and flange thicknesses and also the positioning of the bolts on the flanges. The right balance will not only fulfil the technical aspects of the design but also its practical application in the field in terms of space, weight and cost constraints.

Torque on bolts

The torque applied on each bolt, T, is calculated using the following equation: (2) where

k is the correction factor, 0.177, assuming lubricated untreated steel screw threads ¹⁴

D_bolt is the diameter of the bolt

SF is the safety factor

P_i is the internal fluid pressure

p is the pitch between bolts

R_i is the inner radius of the clamp

For the clamp under study, T is calculated to be approximately 135 Nm, corresponding to a preload (i.e. clamping force) of 60 kN.

FEA optimisation

After the initial clamp sizing calculations, FEA is used to optimise the design. If design limits are exceeded, the thickness of the clamp is incremented accordingly. It has been shown previously ¹² that the bolt hole location, flange thickness and shell thickness can be optimised using Design of Experiment methods coupled with 2D (two-dimensional) FEA.

This provides a ‘design window’ within which the radial expansion of the clamp and the separation between half-clamps at the flange interface are kept as low as possible. The design window is set to limit the clamp expansion to 0.5 ± 0.2 mm, which ensures at least 50% of the designed seal compression is maintained. From the analysis, the abovementioned shell thickness of 35 mm, coupled with a flange thickness of 56 mm, results in a clamp that meets this requirement.

Materials and manufacture

Experiments were performed to validate the design methodology. The clamps under study were made using 1200 gm⁻² E-glass biaxial non-crimp fabric, of which plies are laid up to be parallel to the upper and lower surfaces of the flanges and parallel to the inner mould line, in a 0/90 configuration with respect to the pipe axis. The composite was manufactured via Vacuum Bag Resin Infusion (VBRI) with a commercially available vinyl ester resin at room temperature (RT), followed by an 80°C free-standing postcure, and subsequent machining of grooves for the seals and holes for the screws. Except for the clamp deployed for the field application, none of the other clamps used in this study was fabricated with any pressure ports. The vinyl ester resin used has a glass T_g of 110°C, as measured using Dynamic Mechanical Analysis (DMA). ¹⁵ Hence, the clamps are suitable for use up to a maximum service temperature of 80°C.

In the present work, NBR seals with a Shore A hardness of 65 were used. The seals were waterjet cut from large sheets. It is noteworthy that NBR seals are widely used in the O&G industry because of their durability in service and broad chemical resistance.

Socket head capscrews, manufactured to ASTM A574, ¹⁶ alongside 2H black nuts manufactured in accordance with ASTM A194 ¹⁷ and washers with hardness of 38–48 Rockwell were used to fasten the two half-clamps together. The screws used were fluorocarbon coated to provide the necessary corrosion protection.

The choice of materials and manufacturing method were driven by the need for simple and inexpensive tooling, rapid turn-around on short-runs or one-off products, and low-cost raw materials. Carbon fibre reinforcements were initially considered, but it has the disadvantage of high cost of material and potential for causing galvanic corrosion of the steel pipe, hence the selection of glass fibres. Bi-Axial non-crimp E-glass fabric allows ease of handling and works well with resin infusion. The resin for the clamp was selected based on a combination of suitability for infusion processes, high chemical resistance and a T_g suitable for typical service temperatures of 80°C.

Pressure tests

Two main sets of pressure tests were undertaken; short and long term. The short-term tests were carried out at RT, approximately 20°C, as well as at two elevated temperatures of 65 and 80°C. The long-term test was performed at only one test temperature of 65°C. Throughout this study, an 8-in pipe was considered, and a design pressure of 10.5 MPa was also used as the test pressure.

Short-term pressure tests

The test rig used for short-term pressure testing the clamp was constructed by mounting hydraulic pressure lines to the inside of a 750 mm long, 8-in diameter steel pipe. Two ¼-in BSP holes were drilled and tapped through the surface of the pipe within the defined area to be sealed by the clamp. Hydraulic pressure fittings were assembled to one of the threaded holes on the inside of the pipe to connect it to a hand-operated hydraulic pump (Enerpac P39). A blanking off plug was screwed into the other threaded hole, to be used to bleed out air from the hydraulic system. An independent analogue pressure gauge and a pressure transducer were attached to the assembly. The pressure transducer was connected to a data logger to record the test pressure.

The clamp was assembled over the holes in the pipe, and bolts were tightened to a torque of 135 Nm. The assembled clamp and pipe were placed inside an oven (but not turned on) at RT and the clamp was pressurised to 10.5 MPa, and then held for 1 minute. The temperature was then raised to 65°C with the pressure still held at 10.5 MPa. This was finally repeated at 80°C, before pressure was gradually increased up until failure occurred. Thermocouples were used to verify the test temperatures. Figure 4 a shows the pressure test setup, instrumented with strain gauges.

OPEN IN VIEWER

4

a Pressure test set-up for short-term test. b Strain gauge locations in relation to seal (shaded overlay)

Eight strain gauges were placed on the outside of the pressurised half of the clamp to measure the strains during installation and testing of the clamps. Each strain gauge on the clamp has a corresponding strain gauge, attached to a thermal expansion compensation block, made of the same composite as the clamp, to complete the half bridge. Figure 4 b shows the locations of the gauges. Note that gauges 2 and 4 were oriented in the axial direction, whilst the remainder were in the hoop direction.

Long-term pressure test

A purpose-built pressure test system was used to conduct the long-term survival test for 1000 hours. This long-term test is referenced in ASME PCC-2 ² . The clamp test system used in this work consists of a pipe spool with sealed ends, inlet and outlet ports for filling with water, and a water pump for pressurisation. Valves connected to the inlet and outlet ports served to seal the pressurised water in. Pressurisation of the pipe was achieved by injecting and filling the entire pipe and the clamp annulus with water via the inlet port. An analogue pressure gauge and a digital pressure transducer were connected to the pipe to monitor the pressure in the test system by recording pressure via a data logger. Strain histories in the clamp were monitored via strain gauges attached as shown in Fig. 5.

OPEN IN VIEWER

5

Strain gauge location on the long-term test clamp, on outer surface of top (where defect is situated) and bottom

The pipe spool was soaked in a large water tank with heaters to raise the temperature to the desired level for testing. A temperature control feedback loop was used to regulate the pre-set temperature of 65°C, while the water was circulated by means of a pump throughout the test to ensure proper distribution of heat. A thermocouple was attached to the clamp to log the local temperature.

Figure 6 a shows the pipe spool, clamp assembly and the heated water tank. The mid-span of the pipe spool has a thinned-down area as shown in Fig. 6 b, based on the non-through-wall defect described in ISO/TS 24817 ¹³ . Only 30% remaining wall thickness of the pipe is maintained in this location, with a 5 mm through-wall hole in the middle hence simulating a corroding and, as a result, leaking pipe. The clamp was assembled over this entire defect, whilst ensuring that the seals were firmly seated on non-defective areas of the pipe.

OPEN IN VIEWER

6

Long-term test set-up. a Apparatus for long-term survival test with clamp shown assembled over leaking pipe. b Thinned-down area and hole on the test pipe to simulate a leak within a corroded region

The test was held under the conditions of 65°C and 10.5 MPa internal pressure for a minimum requirement of 1000 hours based on ASME PCC-2, ² and was witnessed by a third-party certification body, Lloyd's Register.

Short-term pressure tests

Strain behaviour

Figure 8 shows the 8 strain readings during installation of the clamp at RT, at 10.5 MPa pressure at RT and at 19.2 MPa pressure at 80°C. Depending on the location, hoop strains were shown to increase slightly during installation, up to a high of 1430 με, at a location of SG1. The influence of the seal on the strain can be observed for SG1, SG3 and SG5 which are located along the same circumferential position but at different axial points of the clamp. Locations of SG1 and SG5, which are close to the opposite ends of the seal, have higher strain levels (i.e. 1430 με and 1308 με, respectively) than location of SG3 (at only 251 με), which is at the mid-point of the clamp and furthest from the seal perimeter. When fully clamped, the unsupported composite material contained within the perimeter of the seal may demonstrate some anticlastic behaviour. Referring to Fig. 9, if the semi-cylindrical section of the clamp is deflected by the seal at points A and B, then there is a tendency for points D and C to contract, thus depressing point X. This would account for the observed dip in strain at the location of SG3.

OPEN IN VIEWER

8

Strain measurements during installation at RT, 10.5 MPa pressure at RT and 19.2 MPa pressure at 80°C

OPEN IN VIEWER

9

Illustration of anticlastic curvature of a flat surface under deflection

Some of the strain readings also suggest that the majority of strain that the clamp experienced has resulted from the process of installation, rather than due to service pressures. Even at the 10.5 MPa internal pressure, the highest strain recorded was 1493 με, which is only marginally higher than the highest strain recorded for the case of installation. Similarly, the SG1 and SG5 hoop strains also did not register any appreciable increase in strain. From this observation, it appears that the applied pressure opposes the anticlastic effect mentioned above, thus increasing the low-magnitude strains recorded during installation, such as at locations SG3 and SG6.

Nonetheless, results from the tests indicate that the strain levels are generally raised when the clamp is pressurised beyond its intended design. This can be seen for the strain readings at 19.2 MPa pressure where the highest strain has increased to 3396 με.

FE Model and comparison with test data

The strain gauge readings at the 8 instrumented points of the tested clamp are compared against the FE-predicted strain levels at the corresponding locations, see Fig. 10 a–c. Simulation predictions show reasonably good correlation to the measured values for the different pressures and temperatures: (i) installation at RT, (ii) 10.5 MPa pressure at RT and (iii) 19.2 MPa pressure at 80°C. For the installation condition, the simulation slightly under-predicts the strain with the highest discrepancy being 910 με versus the actual of 1430 με. However, the prediction is more conservative for the pressurised conditions. At the10.5 MPa, RT condition, the highest predicted strain is 2015 με compared to the corresponding measured strain of 1493 με, while at the 19.2 MPa, 80°C condition, the predicted strain is 3508 με, compared to the measured strain of 3396 με.

OPEN IN VIEWER

10

Comparison of strains at different strain gauge locations between test measurements and simulation predictions: a installation at RT, b 10.5 MPa pressure at RT and c 19.2 MPa pressure at 80°C

Using the HWC material properties, predicted stress contours in the clamp at the design pressure of 10.5 MPa, as illustrated in Fig. 11, show that the hoop and radial stresses within the clamp remain well below the material strength. Radial stresses are as low as 17 MPa for the bulk of the clamp, as represented by the white regions. The vinyl ester resin has a tensile strength more than 5 times that predicted at 86 MPa. In terms of hoop stresses, the bulk of the clamp experiences not more than 50 MPa. Compared to the AM and HWC tensile strengths of the composite laminate, i.e. 396 MPa and 128 MPa, respectively, this results in a margin of 7.9 and 2.5 times, respectively.

OPEN IN VIEWER

11

FE predictions of radial and hoop stresses in the clamp using HWC material properties

Some areas of locally high stresses are also predicted, especially within the unmachined area at the bolt recesses. This is due to the geometry of the bolt recess, and has been shown by testing to stress relax with the formation of small cracks that do not bear any detrimental effect on the structural performance of the clamp.

Fracture and failure behaviour

Testing has demonstrated the clamp to adequately hold 10.5 MPa pressure for 1 minute at RT, 65 and 80°C, with no visible or audible signs of failure. The maximum pressure recorded was 19.2 MPa at 80°C, and it failed by seal extrusion, as illustrated in Fig. 12 a, while no significant damage to the composite clamp was detected. Fractures were observed at the base of the unmachined area around the bolt recesses, as shown in Fig. 12 b. The fractures, however, did not affect the pressure-bearing capability of the clamp, because, with replacement of a harder seal, the clamp was able to withstand pressures all the way up to, and beyond, its design limit. The harder rubber seals resulted in failure of the composite clamp itself, at a pressure of 22.6 MPa, which is more than 3.2 times the MAWP of the clamp ¹² . Hence, the harder seal prevented seal extrusion beyond the maximum capacity of the composite clamp.

OPEN IN VIEWER

12

Damage modes observed during pressure testing of the clamp. a Seal extrusion failure; b fractures occuring at the base of the unmachined areas; c external shear fracture in the flange; d internal shear fracture in the flange

The failure mode of the clamp (with the harder seal) was observed to be interlaminar shear along, and on the external surface of, the flange on the pressurised half-clamp, as shown in Fig. 12 c. The failure was within the seal groove, hence providing a leak path, while the shear plane was along one of the ply interfaces, as shown in Fig. 12 d, from the inside of the clamp.

Long-term pressure test

The pressure in the system held for the duration of the test (1000 hours), as shown in Fig. 13 a, without external introduction of additional pressure. Fluctuations in temperature, however, did result in some small pressure variations.

OPEN IN VIEWER

13

a Pressure and temperature measurements for the duration of the survival test; b Pressure test on the clamp at RT after the survival test

Upon successfully completing the test, the pipe was depressurised. Once stabilised to RT, the clamp was re-pressured to 10.5 MPa and held for 4 hours, as plotted in Fig. 13 b. This was carried out to confirm that the functionality of the clamp was not compromised after the long-term survival test.

Strain measurements made during the test are shown in Fig. 14 for the locations shown schematically in Fig. 5. ‘H’ gauges are hoop orientated, while ‘Ax’ gauges are axially orientated along the length of the clamp. The strain values fluctuated with the minor variation in pressure that was induced by temperature variations.

OPEN IN VIEWER

14

Strain measurements of the clamp during the survival test

All measured strain values are below 0.072%, viz. well below the 0.25% maximum strain limit requirement and they remained relatively constant during the entire 1000 hours. This is a sign that no structurally compromising event occurred during the test, especially because the hoop-carrying capacity of the clamp was maintained. H0, which is the hoop strain of the bottom clamp, recorded the highest value of approximately 0.072%.

Upon inspection of the clamp following completion of the test, some macrocracks, approximately 8 mm deep, were found to have occurred in the unmachined area around the bolt recesses, see Fig. 15, similar to the ones observed during the short-term pressure test.

OPEN IN VIEWER

15

Fracture at the base of the unmachined area surrounding the bolt recess

No other damage could be found on the clamp. It is highly likely that these cracks developed due to local stress concentrations introduced during the machining of the bolt recesses. During the test, this stress was relieved via the cracking, but did not propagate or interfere with the pressure containment efficacy of the clamp.