Subscribe to our newsletter to receive the latest news and events from TWI:

Subscribe >
Skip to content

Ensuring the Strain Capacity for Threaded-End SENT Specimens

   

Ensuring the Strain Capacity in Threaded-End Single Edge Notched Tension (SENT) Specimens

 

Andreea Crintea and Philippa Moore

TWI Ltd, Granta Park Great Abington, Cambridge, CB21 6AL, UK

Paper presented at ISOPE 2016, June 26-July 2 2016 Rhodes, Greece.

Abstract

Single Edge Notched Tension (SENT) fracture toughness test specimens are being used for a wide range of applications due to the higher fracture toughness that is measured by SENT specimens compared to an equivalent specimen under bending (SENB). The testing of SENT specimens is now standardized in BS 8571:2014 (BSI, 2014) and there is potential to use SENTs for high and low temperature tests, in a wider range of material thickness, and for different applications.

This paper describes the validation of a method for carrying out SENT tests at very low temperatures, using threaded ends to allow testing inside a temperature controlled test chamber, while preventing the specimen from yielding at locations away from the intended notch tip. Two designs of threaded-end specimens were assessed for their feasibility; directly machining a round threaded portion onto a square section specimen is the simpler approach, while the alternative is attaching to the square section specimen (in this case by friction welding), a wider diameter round bar for threading.

Numerical models of the notch location and of the threads under load were compared for their respective strain capacity. A successful SENT specimen design has greater strain capacity in the threaded end than around the notch. The results of the numerical models were compared to experimental tests on both designs with different notch depths.

Machined threads were shown to be limited to applications with low weld strength over-match, deeper notch depths and single point fracture toughness testing (including multi-specimen R-curve testing). Friction welding the threaded portion allowed threaded-end SENT specimens to be successfully used for unloading compliance tests R-curve tests, shallower notched specimens and where there is weld strength over-matching.

Nomenclature

a measured initial notch depth, mm
B thickness of test specimen, mm
CTOD Crack Tip Opening Displacement, mm
CTOD0.2 resistance to crack extension expressed in terms of CTOD at 0.2 mm crack extension including blunting, mm
CTODmax resistance to crack extension expressed in terms of CTOD at the maximum crack extension, mm
E Young’s modulus, MPa
H distance between clamped grips, mm
KI stress intensity factor
M weld strength mismatch ratio
Rp0.2 0.2% offset yield strength of the parent metal at the temperature of the fracture toughness test, MPa
SENB Single Edge Notched Bend specimen
SENT Single Edge Notched Tension specimen
VP1, VP2 plastic clip gauge displacements, mm
UTS Ultimate Tensile Strength, MPa
Δamax UC maximum crack extension estimated from the unloading compliance method, mm
Δamax FF maximum crack extension measured from the fracture face, mm
ε applied strain
ν Poisson’s ratio.

1. Introduction

Single Edge Notched Tension (SENT) fracture toughness test specimens (Fig.1) are now being used for a wider range of applications. The test specimen became established for use with fitness-for-service assessments of flaws in pipeline girth welds under high strain conditions during installation, as described in DNV-RP-F108:2008 (DNV, 2008), and were later further established for all subsea pipelines in DNV-OS-F101:2013 (DNV, 2013). The lower constraint of the SENT specimen, compared to the historically more common Single Edge Notched Bend (SENB) specimen, results in higher values of fracture toughness being measured in SENTs at ambient temperature. The constraint of the SENT specimen was nonetheless still higher than a flaw in a pipe under uniaxial loading, and the fracture toughness from an SENT test could be conservatively used for the assessment of pipelines. Fracture toughness testing using SENT specimens is now standardized in BS 8571 (BSI, 2014) and there is potential to use SENT specimens for higher and lower temperature tests, for a wider range of material thickness, and for different engineering applications. Provided the structural application to be assessed has lower constraint than the specimen, SENT test results can be considered suitable for the fitness-for-service assessment.

Much of the established data for SENT tests has been generated on ductile materials as tearing resistance curves (R-curves) in terms of J and CTOD at ambient temperatures. But as SENT specimens become used more widely, lower temperature tests will be performed more often. It is known that a change in constraint between SENT and SENB specimens will affect not just the upper shelf fracture toughness values, but also the ductile-to-brittle transition temperature. This is why characterizing the low temperature behavior of SENT specimens is important. Fig. 2 shows the difference between SENB and SENT ductile-to-brittle transition curves in three different steels tested at TWI (Moore and Crintea, 2016). Standard deeply notched specimens (a0/W=0.5) were used for this comparison in order to capture the potential risks of the simple replacement in engineering critical assessments (ECAs) of SENB fracture toughness with SENT fracture toughness. SENT specimens may show a lower transition temperature than SENBs, but also have a very steep transition between ductile and brittle behavior.

Fig. 1 SENT test specimen of square cross section instrumented with a double clip gauge ready for test under clamped loading at ambient temperature

Fig. 1 SENT test specimen of square cross section instrumented with a double clip gauge ready for test under clamped loading at ambient temperature

Fig. 2 Comparison of SENT and SENB transition curves in three different steels (Moore and Crintea, 2016)

Fig. 2 Comparison of SENT and SENB transition curves in three different steels (Moore and Crintea, 2016)

Clamped SENT specimens are usually tested at low temperatures using cooling pads locally attached around the notch. Experience of TWI laboratories in testing SENT specimens at low temperatures has shown that depending on the weld-strength over-matching, test temperature and notch depth, locally cooled specimens can fail by yielding in the arms of the specimen. This is mainly because the cooled region around the notch results in higher strength than the arms of the specimen which would experience only ambient temperature.

The thermal chamber method allows testing to much colder temperatures (as low as -160°C), or even high temperatures with good thermal stability, and the potential to cool and soak subsequent test specimens while the first is being tested, shortens the overall testing time for a set of specimens. One solution to enable SENT specimens to be tested inside a thermal chamber was described by Delliou and Geniaut (2014) using a shorter SENT specimen gauge length and bolt holes in a wider clamped area at the end of the specimen to enable it to be gripped inside a thermal chamber. They carried out SENT tests from -40°C to -100°C.

An alternative solution investigated in this work is to adapt the design of a plane-sided clamped SENT specimen to have threaded ends which can then be ‘clamped’ and held in the threaded grips of a universal test machine and tested inside a temperature controlled chamber (Fig. 3). However, the method is not without its own challenges, and the optimization and validation of this specimen design is discussed in this paper with respect to ensuring its strain capacity.

Fig. 3 Machined threaded-end SENT specimen gripped inside a thermal chamber

Fig. 3 Machined threaded-end SENT specimen gripped inside a thermal chamber

2. Manufacturing Threaded-end SENT Specimens

Two different production methods were used to investigate threaded ends on SENT specimens. The first method was to machine round threaded portions directly onto the ends of the square cross section of the SENT specimen, leaving a specimen length equal to ten times the specimen width (10W) between the threaded portions, which is the specimen length defined for clamped SENT specimens in BS 8571 (BSI, 2014) and DNV-RP-F108 (DNV, 2008) (Fig. 4). The second method was to use rotational friction welding to weld lengths of round bar onto the ends of a length of square cross section. Threads were then machined onto the round bar (Fig. 5). The length of the square cross section part of the specimen was cut 10mm longer than 10W initially, to allow for some material loss in the flash formed during friction welding, while ending up with the correct specimen dimensions for testing.

Fig. 4 Machined threaded-end SENT specimens, shown with shims attached to the notch mouth ready for the screw attachment of a double clip gauge before testing

Fig. 4 Machined threaded-end SENT specimens, shown with shims attached to the notch mouth ready for the screw attachment of a double clip gauge before testing

Fig. 5 Friction welded threaded end SENT specimens, shown with shims attached to the notch mouth ready for the screw attachment of a double clip gauge before testing

Fig. 5 Friction welded threaded end SENT specimens, shown with shims attached to the notch mouth ready for the screw attachment of a double clip gauge before testing

The machined thread method had the advantage of being quicker and simpler to produce, and in these specimens the strength in the threaded portions was the same as the strength in the main part of the test specimen. The concern in these specimens was that the loss of cross section in the threaded portion might mean that the load bearing capacity of the threads was less than in the notch region.

The friction welded method had the advantage of being able to select round bar with a larger cross section than the square section specimen so that the threaded portion cross section was at least that of the specimen. For these specimens, commercially available S355 grade structural steel round bar (identified as M03) of diameter 30mm was used. Common steel rod materials are either high strength tool steels (unweldable) or mild steel (low strength); the S355 was selected as a compromise. M03 steel  had a yield strength of 368MPa, lower than the materials being tested, again raising concerns about load bearing capacity of the threaded ends compared to the notch. A hardness traverse across the friction weld (Fig. 6) confirmed that the weld and heat affected regions are stronger than the parent metal in the threaded region, so the latter is the limiting yield strength in the joint. The friction welded specimens were more complex to produce and used fairly specialized friction welding equipment for their production.

Fig. 6 Macrograph showing a cross section through the friction weld attaching the threads (M03) to an SENT specimen (M02). The hardness traverse across this weld is shown superposed

Fig. 6 Macrograph showing a cross section through the friction weld attaching the threads (M03) to an SENT specimen (M02). The hardness traverse across this weld is shown superposed.

3. Avoiding Failure away from the Notch Location

3.1. Analysis Approach

The feasibility of each of the two threaded-end SENT specimen designs was established by performing a strain capacity analysis comprising both finite element modelling and experimental trials. The specimen design was considered suitable only if the strain capacity of the plane-sided SENT specimen before the point of unstable tearing around the notch proved less than the strain capacity of the threaded end. The challenge in using threaded ends for SENT specimens is to ensure the load transmission to the notch location by preventing the failure of the threads, or yielding in the arms of the specimen.

The strain capacity of the plane-sided SENT specimen was obtained by a combination of elastic-plastic finite element simulation, used to determine the crack driving force during the fracture toughness test, and experimentally-obtained unloading compliance tearing resistance curve (R-curve) test results, necessary to predict how the notch will start to tear and the crack to grow.

The principle of tearing analysis assumes that a flaw in a structure propagates in a stable manner until the crack driving force has increased to a point where any increase in driving force cannot be supported by an increase in fracture resistance. At this point, the propagation is unstable and results in the failure of the structural component (Minaar et al, 2007). Crack propagation was simulated by using a series of numerical models with stationary cracks in which the crack depth was incrementally increased. The overall strain capacity was evaluated for specimens with initial notch depth to width ratios a0/W of 0.2, 0.325, 0.4 and 0.5. When the cracks in these models were grown to simulate the crack propagation, the final a0/W ratios ranged between 0.5 and 0.7, with the model assessments being made at increments of a0/W of 0.025.

The strain capacity of the threaded ends was evaluated by means of elastic-plastic finite element analysis. The limiting stage was defined as the level of applied strain at which the material UTS was obtained at some position along the axis of the specimen. This is a conservative approach, since in reality the stresses would re-distribute somewhat to accommodate the yielding in the threads, so UTS would not be experienced uniformly across the cross section of the specimen at this point to cause failure.

3.2. Numerical Modelling of the Notch Region

Firstly, a tearing analysis was carried out to determine the onset of unstable tearing of the plane-sided SENT specimen under increasing load. Elastic-plastic finite element analysis was performed on models for a range of conventional BxB (B=20mm) specimens of length H=200mm (where H is the daylight between grips), made of homogeneous material with the tensile properties of materials M01 and M02. The plane-sided SENT specimen was simulated as a 2D deformable body. The symmetry conditions allowed for the use of a half model, with appropriate constraints applied to the remaining crack ligament.

The crack was modelled with a blunt, U-shaped notch tip of 0.05mm radius in order to avoid the singularity effect at the crack tip. Given that the scope of the analysis was to determine the specimen’s plastic yielding limitations in comparison to those of the thread, a brittle fracture in the specimen (in the case where the notch is actually sharper than 0.05mm) would mean fracture was more likely to occur in the notch location than the threads. Therefore assessing a blunter notch is conservative for the failure criterion being analyzed in this work. The subdivision of the geometry allowed the mesh to be fine around the crack tip region and coarser farther away. The mesh of the SENT specimen model was composed of 8-node biquadratic plane stress quadrilateral elements.

Crack propagation was simulated by using a series of models with stationary cracks in which the crack depth was incrementally increased. The relation between the CTOD and the applied strain was recorded for each of the models. The critical point for unstable crack propagation in the SENT specimens for M01 and M02 were determined by plotting experimental R-curves alongside numerical plots of CTOD against crack extension obtained for a given value of strain (iso-strain plots), and finding the point of tangency of the curves, i.e. the strain at crack instability.

In the finite element model, CTOD results were calculated with the same method used by the double clip gauge in the experiments, using the principle of similar triangles with their apex at a given point of rotation for the specimen (Figs. 7 and 8). Even though the actual clip gauges were not modelled, their corresponding displacement measurements were taken at the location of the last two nodes at the crack mouth opening. For all SENT models created, the relative distance between the two ‘clip gauge’ displacement measurements was 1mm.

Fig. 7 Determination of CTOD from the double clip arrangement (BSI, 2014)

Fig. 7 Determination of CTOD from the double clip arrangement (BSI, 2014)

Fig. 8 Double clip gauge arrangement in the FE model

Fig. 8 Double clip gauge arrangement in the FE model

According to BS 8571 (BSI, 2014), CTOD is calculated as the sum of the elastic and plastic components following the formula:

Ensuring the Strain Capacity in Threaded-End Single Edge Notched Tension (SENT) Specimens - Equation 1

(1)

In this equation, the first part represents the elastic component of the CTOD based on the stress intensity factor, K, and elastic tensile parameters (Poisson’s ratio, ν, yield strength, Rp0.2, and elastic modulus, E). The other terms in Eq. 1 sum up to the plastic component of CTOD, based on the crack mouth displacement measurements, as defined in Fig. 7. Previous research papers studying the tearing behavior of SENT specimens have shown that the value of the elastic component of the CTOD is considerably lower than the plastic one (ExxonMobil, 2010); consequently, its contribution has been neglected in this study for simplicity.

3.3. Numerical Modelling of the Threads

An elastic-plastic finite element analysis was carried out for the threaded end of the SENT specimen in contact with the grips of the tensile test machine. The parts have been meshed with 4-node bilinear, axisymmetric, quadrilateral, reduced integration, hourglass control elements. The reduced integration linear type elements have been chosen due to their ability to better establish the contact between the elements, while their excessive distortion has been restricted using the hourglass control option.

The interaction between the thread and the grip has been set up as a standard surface-to-surface contact defined by a tangential behavior and a ‘hard’ pressure-overclosure relationship. A friction coefficient of 0.15 was chosen based on the previous published works on experiments and finite element simulations for bolt-nut friction (Fukuoka and Nomura, 2008).

The density of the mesh was increased in the contact region, with a finer mesh on the side of the thread where the actual load transfer takes place (Fig. 9). The thread was more finely discretized overall in comparison to the grip, due to the necessity to accurately estimate stresses across the cross-section of the threaded-end SENT specimen where failure must be avoided.

The strain capacity of the threaded ends of the SENT specimen was determined by making the assumption that local plastic deformation can be allowed at the root of the thread. With the increase of load on the specimen, the stress would be redistributed to the unthreaded core of the ends of the specimen. Consequently, the strain capacity was evaluated in terms of critical applied displacement necessary to cause a part of the unthreaded core to reach the ultimate tensile strength of the material.

Fig. 9 Axisymmetric model of the threaded end of the SENT specimen and the tensile machine grip with FE mesh shown and zoomed view of the FE mesh in the contact region. Finer mesh on the side of the thread where the load is transferred

Fig. 9 Axisymmetric model of the threaded end of the SENT specimen and the tensile machine grip with FE mesh shown and zoomed view of the FE mesh in the contact region. Finer mesh on the side of the thread where the load is transferred

3.4. Experimental Results

When a SENT specimen is machined with a shallow notch, there is more area of the ligament ahead of the notch to carry the load, and therefore a greater risk that the arms will fail before the notch location. BS 8571 (BSI, 2014) permits a0/W ratios between 0.2 and 0.5, so model cases representing this range of notch depths were assessed for their strain capacities, and corresponding experimental data at room temperature was generated.

The machined threaded-end SENT specimens and the friction welded threaded-end SENT specimens were tested in accordance with BS 8571 (BSI, 2014) using the unloading compliance method to generate R-curves with different initial crack depths (Table 1).

The R-curves generated using machined threaded-end SENT specimens displayed apparent negative crack growth due to the excessive rotation of the specimen under tension loading, which was not observed in equivalent specimens with the friction welded ends (Fig. 10). The machine setup was identical for the two specimen types, the only difference being the threaded connection: 20mm diameter thread for the machined threaded-end specimens and 30mm diameter thread for the friction welded specimens, along with their corresponding grips. The distance between grips was kept constant at ten times the width of the plane-sided SENT specimen (H=200mm) for both specimen types. The difference in specimen rotation raised concerns about the overall stiffness of the machined threaded-end SENT specimens for unloading compliance tests, and for this reason, the R-curve results from this specimen design were not considered further in the tearing analysis.

Fig. 10 Unloading compliance tearing resistance curves for 20mm thick specimens in M02 with a0/W of about 0.4, from specimens with machined threads, and with friction welded threads

Friction welded specimens were successfully tested using the unloading compliance method for specimens with a0/W ratios from 0.22 to 0.42. The test data shown here was obtained from specimens without side grooves, which is the standard approach for the single point fracture toughness testing used to generate the ductile-to-brittle transition curves. For unloading compliance R-curves it is usual to use side-grooved specimens. However, the R-curves in this work were generated on plane-sided specimens to replicate the behavior of the low temperature tests for the numerical model, and do not as such represent ‘standard’ R-curves. Side-grooving is a technique that can improve the straightness of the stable tearing front during R-curve testing and there was some uneven crack tearing on these specimens as a consequence.

The results from all the R-curve tests are summarized in Table 1. The results were fairly scattered, with no clear trend with a0/W ratio, so the R-curve data with the best match between the crack length measured from the fracture face and the one predicted from the unloading compliance method was used as the most accurate R-curve to characterize each of the two materials, M01 and M02, assuming the tearing resistance to be independent of notch depth. The resulting R-curves were fitted to a Power Law equation for crack extensions between 0.2mm and the maximum crack extension from the test data. The R-curves are presented in Fig. 11.

Table 1 Threaded-end SENT test results obtained at ambient temperature using the unloading compliance method. The results in bold indicate the data used to characterize the material behavior in the model. MA=machined specimen, FW=friction welded specimen

Design and material a0/W Δamax UC
[mm]
Δamax FF
[mm]
CTOD0.2
[mm]
CTODmax
[mm]
MA-M01 0.512 0.627 2.541 2.945 3.984
MA-M02 0.415 1.24 2.809 2.843 3.595
MA-M02 0.519 1.108 3.004 2.798 4.164
FW-M01 0.222 1.615 2.037 0.522 3.029
FW-M01 0.355 1.618 1.933 0.655 3.098
FW-M02 0.203 1.86 1.712 0.381 3.049

FW-M02

0.423 1.767 2.176 0.973 3.335
Fig. 11 Best fit Power Law curves for the friction welded SENT specimens tested using the unloading compliance technique

Fig. 11 Best fit Power Law curves for the friction welded SENT specimens tested using the unloading compliance technique

The strain capacity of the plane-sided SENT specimens was determined by overlapping the FEA iso-strain plots for different initial notch depths, with the experimentally obtained R-curves. An example is shown in Fig. 12.

Fig. 12 FEA iso-strain plots obtained on material M01 and an initial notch depth of a0/W =0.2 overlapped with the experimental R-curve with an initial notch depth of a0/W =0.355. Tangency of plots defines the limiting strain capacity for the specimen

Fig. 12 FEA iso-strain plots obtained on material M01 and an initial notch depth of a0/W =0.2 overlapped with the experimental R-curve with an initial notch depth of a0/W =0.355. Tangency of plots defines the limiting strain capacity for the specimen.

Table 2 Strain capacity of the plane-sided SENT specimen for different notch depths

Material a0/W (FEA) a0/W (R-curve) Strain capacity, ε [%]
M01 0.2 0.355 1.82
M01 0.325 0.355 1.81
M01 0.5 0.355 1.8
M02 0.2 0.203 1.72
M02 0.4 0.203 1.71
M02 0.5 0.203 1.72

The estimated strain capacity for material M01 is 1.8%, with slight variations of about 0.01% between the different notch depths assessed. The same independence of notch depth was observed also in the case of material M02, where the strain capacity value stabilized around 1.71%.

If global yielding was the limiting criterion for the ligament of the specimen, the general tendency would be to assume that a shallow notched specimen would have a greater strain capacity than a deeper notched specimen. However, the results show that in this case the failure behavior is governed by unstable ductile fracture, which is independent of notch depth but governed by the R-curve behavior. For comparison, the strain equivalent to general yielding in the specimen ligament was calculated, based on reaching the flow stress of the material (average of yield and UTS).

The stress-strain curves displayed in Fig. 13 prove that for a flow stress of 880 MPa for M01, the corresponding strain is approximately 2.5%, while for M02 the flow stress of 700 MPa corresponds to a strain of 2%. This shows that the plastic collapse of the specimen’s ligament would happen much later than a potential unstable fracture and hence fracture would be the failure criterion for the plane-sided SENT specimen rather than general yielding.

Fig. 13 Stress-strain curves of materials involved in the analysis

Fig. 13 Stress-strain curves of materials involved in the analysis

Table 3 Strain capacity of the threaded ends of the SENT specimens

Material Thread diameter [mm]

 

Strain capacity, ε [%]

 

M01 (high strength steel)

 

20

 

1.77

 

M02 (lower strength steel)

 

20

 

0.97

 

M03 (round bar friction welded attachments)

 

30

 

1.94

 


Comparing the tabulated values obtained from the tearing analysis of the plane-sided SENT specimens and the numerical modelling of the threads, it appears that the machined threaded-end SENT specimens are not feasible for any of the notch depths tested (neither M01 nor M02 gave thread capacities greater than the equivalent notch location). Experimental tests confirmed that machined threads on EDM notched specimens with a0/W of less than 0.35 can suffer failure in the threads (Fig. 14).

However, successful experimental tests had been done on more deeply notched fatigue pre-cracked specimens with a0/W of around 0.5 (used to generate the data used to plot the transition curves in Fig. 2). It was concluded, based on these experimental findings that these are, in fact, an acceptable design for deeply-notched single point fracture toughness testing in M01 and M02. The machined threaded-end SENT specimens cannot be used for testing with the unloading compliance method, as they resulted in negative apparent crack growth due to insufficient stiffness.

The outcome discrepancy between model and practice can be attributed mainly to the conservative failure criterion of the threads. Failure in the model was defined as the solution increment when one position along the axis reached UTS, rather than upon the net section reaching UTS. This gives a lower strain capacity prediction in the threads. Therefore, the predictions from the model can be applied with confidence in the conservatism, despite being potentially over-conservative.

The friction welded specimens manufactured with threads of 30mm diameter and commercially available structural steel M03 proved feasible for all notch depths in the range of 0.2 to 0.5. They were shown through model and experiment to be suitable for testing using both for single point and unloading compliance methods.

Fig. 14 Examples of failure in the machined threaded ends in SENT specimens in M01, with (a) yielding in specimens with a notch depth of a0/W of 0.325, and (b) a thread fracture from specimens with notch depth a0/W of 0.227

Fig. 14 Examples of failure in the machined threaded ends in SENT specimens in M01, with (a) yielding in specimens with a notch depth of a0/W of 0.325, and (b) a thread fracture from specimens with notch depth a0/W of 0.227

Weld Strength Over-matching

For specimens with shallow notches, where the notch is located in the weld metal which has a much higher yield strength than the parent material, there can be a risk of yielding in the arms even for SENT tests performed at ambient temperatures. The weld strength mismatch ratio, M, is defined as the yield strength of the weld metal divided by the yield strength of the parent metal, and for over-matched welds it is recommended in BS 8571 (BSI, 2014) to restrict the notch depth so that:

Ensuring the Strain Capacity in Threaded-End Single Edge Notched Tension (SENT) Specimens - Equation 2

(2)

This risk would be exacerbated when threads were machined onto the specimen directly, and was investigated by modelling over-matching in SENT specimens. The weld strength over-matching was simulated considering M01 the high strength weld metal in a specimen of varying parent metal strength. Consequently, the tensile properties of the threaded ends were altered in relation to the tensile properties around the notch, so that cases with a mismatch ratio in the range of 1.1 to 1.5 could be analyzed. The strain capacity of the threaded ends obtained from the finite element analyses are presented in Table 4.

Table 4 Strain capacity of the threaded ends of the SENT specimens for different levels of weld strength over-matching when M01 is the over-matched material.

Mismatch ratio, M Thread diameter [mm] Strain capacity, ε [%]
1.1 20 1.11
1.2 20 0.77
1.3 20 0.63
1.4 20 0.6
1.5 20 0.59

A comparison was made between the strain capacity results of the threaded ends of a simulated over-matching weld in Table 4 and the 1.8% strain capacity of the specimens modelled out of homogeneous material M01 (Table 2). With the increase of the over-match ratio, the strain capacity of the thread reduces as the parent metal strength is lower and the relative strain capacity between notch and threads is larger.

The machined threaded-end SENT specimens seem to be unfeasible for even the lowest mismatch ratio, M=1.1. However, for the even-matched machined threaded-end SENT with a0/W=0.5, a discrepancy was observed between the outcome of the FE analysis and that of the experimental tests for both assessed materials. Although the strain capacity of the specimens exceeded the strain capacity of the threads by 0.03% for M01 and 0.74% for M02, the specimens were successful during test. This fact suggests there is conservatism in the FE analysis and that a thread strain capacity prediction of more than 1% is still safe from risk of yielding before the notch. Relating this finding to the case of the 1.1 over-matched specimen with a0/W =0.5, where the difference in strain capacity between the specimen and the threaded end is much smaller than in the previous case, 0.69%, it is considered that the machined threaded-end SENT specimen would be suitable for such a configuration (M=1.1 and a0/W sufficiently deep, such as 0.45).

The friction welded threaded-end SENT specimens proved feasible for both M01 and M02 even-matched specimens with a0/W in the range of 0.2 to 0.5. The round bar for the threaded ends has a yield strength of 368MPa, while the steels being tested had yield strengths of 850MPa for M01 and 640MPa for M02. Based on the analysis performed on the even-matched friction welded specimens, one could conclude that this design can be used for any weld strength mismatch, as long as the tensile properties of the weld metal are not higher than 850 MPa which has been validated, and the conditions in Eq. 2 also hold. Additionally, attention must be paid to the diameter of the round bar that is friction welded to the SENT blank, to ensure it has a cross section sufficiently larger than the specimen gauge.

4. Discussion

Comparisons between the load bearing capacity of the cracked specimen’s ligament and that of the effective cross-sectional area of the thread would be a too simplistic approach, which would completely disregard the way in which a sharp crack influences the failure behavior of the specimen. Consequently, a more detailed FE analysis was needed in order to reach a safe conclusion with regard to the feasibility of the specimen.

The machined threaded-end SENT specimens have the advantage of being a fast and inexpensive specimen preparation method. For both steels assessed in this project, this specimen design proved feasible when specimens with deep notches (a0/W =0.5) were tested for single point fracture toughness. Therefore, these specimens could also be used for multi-specimen R-curves. The finite element simulation of this specimen also suggested that for the same a0/W =0.5, a weld strength mismatch of no more than 1.1 has similar levels of conservatism with the even-matched specimen models and experiments, and could be considered acceptable. The tearing resistance curves of these specimens obtained by the unloading compliance technique raised concerns with regard to the overall stiffness of the specimen design, which may not be sufficient for preventing the rotation of the specimen during unloading compliance testing.

The friction welded threaded-end SENT specimens allowed for the attachment of lower strength threads, but with greater diameter than the width of the specimen, such that it can be used for a broader range of notch depths and weld strength mismatch levels as borne out by experimental and numerical results. The disadvantage lies in the fact that the manufacturing procedure involved specialized rotational friction welding equipment, which can be expensive and may add to the specimen preparation time. However, this specimen design approach could be used with other metals of sufficient strength, welding processes and procedures as long as the quality of the weld is sufficient to avoid yielding of the specimen at the weld location.

This specimen design proved feasible for the entire range of a0/W between 0.2 and 0.5. Over-matching levels of up to 1.5 were also effective. The improved overall stiffness of the friction welded specimen compared to the machined threads enabled both single point fracture toughness testing and unloading compliance testing.

A decision flowchart for selecting a suitable SENT specimen design for low temperature testing is given in Fig. 15.

Fig. 15 Decision flowchart to aid selection of the appropriate SENT specimen design for low temperature tests

Fig. 15 Decision flowchart to aid selection of the appropriate SENT specimen design for low temperature tests

Further work could be carried out to develop this research to investigate the effect of the material tensile properties on the strain capacity of threaded end specimens, and to fully experimentally characterize the effect of notch depth.

5. Conclusions

  1. SENT specimens can be more easily tested at low temperature inside an environmental chamber when they have threaded ends to fit within established tensile test grips.
  2. Threads can be simply machined onto square section SENTs when testing single point toughness in deeply notched SENT specimens without significant weld strength over-matching.
  3. For specimens with shallower notches, weld strength mismatch, or for generating unloading compliance R-curves, round bar can be welded to the ends of the specimen for threads to be machined.

6. Acknowledgements

Thanks are owed to Phillip Cossey for carrying out the experimental SENT tests presented in this paper and to Tyler London and Simon Smith for their advice.

Part of this work was carried out with funding provided by the Industrial Members of TWI as part of the Core Research Programme, and through efforts of the NSIRC/Brunel University MSc programme in Structural Integrity.

7. References

  1. BS 8571, “Method of test for determination of fracture toughness in    metallic materials using single edge notch tension (SENT) specimens,” 2014.
  2. P. Delliou and S. Geniaut, “Tests on SENT specimens to study geometrical effects in the ductile to brittle transition,” in ASME 2014 Pressure Vessels & Piping Conference, PVP2014, Anaheim, California, USA, July 20-24, 2014.
  3. DNV-OS-F101, “Submarine Pipeline Systems,” 2013.
  4. DNV-RP-F108, “Fracture control for pipeline installation methods introducing plastic strain,” 2006.
  5. ExxonMobil Upstream Research Company, “Measurement of Crack Tip Opening Displacement (CTOD) fracture resistance curves using Single Edge Notch Tension (SENT) specimens,” September, 2010.
  6. K. Minaar, P. C. Gioelli, M. L. Macia, F. Bardi and N. Biery, “Predictive FEA Modeling of Pressurized Full-Scale Tests,” in Proceedings of the Seventeenth International Offshore and Polar Engineering Conference, ISOPE 2007, Lisbon, Portugal, July, 2007.
  7. T Fukuoka and M Nomura, “Proposition of Helical Thread Modelling with Accurate Geometry and Finite Element Analysis”, in the Journal of Pressure Vessel Technology, 2008, Vol. 130.
  8. P. Moore, “Optimisation of SENT specimen design,” in Proceedings of EUROJOIN9, Bergen, Norway, June, 2015.
  9. P. Moore and A. Crintea, “Single Edge Notched Tension (SENT) Testing at Low Temperatures,” in To be presented at International Pipeline Conference IPC 2016, Calgary, Canada, September, 2016.

For more information please email:


contactus@twi.co.uk