Fatigue sensors take centre stage - 'seeing' the way forward...

TWI Bulletin, July - August 2008

CrackFirst^TM - a new fatigue damage sensor developed for welded joints in steel structures.

Yanhui Zhang graduated in China as a metallurgist in 1982. After obtaining his PhD in the United Kingdom, he undertook post-doctoral work at Cambridge University before joining TWI in 2001. Now he is a principal project leader working in the Structural Integrity Technology Group. He has many years research experience covering fatigue of small cracks in aluminium alloys, high temperature fatigue of silicon nitride ceramics and thermomechanical fatigue and creep of Ni-base superalloys and has published many scientific papers. His current work is mainly on fatigue issues of welded joints.

Chris Allen gained his Natural Sciences degree from the University of Cambridge then completed a doctorate in the area of Materials Science at the University of Oxford. Following postdoctoral research, he worked for two years as a project leader at Corus on rolled aluminium products, before joining TWI in 2002. Since then, he has specialised in the project management and delivery of laser welding projects for a range of TWI's clients, primarily in the aerospace, shipbuilding, defence, and automotive sectors.

Peter Tubby joined TWI in 1979 after a degree in metallurgy at Manchester followed by four years at Lucas Research. He manages the Fatigue Integrity Management Section of TWI's Structural Integrity Group, which comprises ten engineers specialising in assessment of structures and components with particular emphasis on the fatigue performance of welded joints.

Prof Jagath Mawella, MBA, PhD (Cantab), FWeldI, FIMMM, C.Eng., is the Head of Maritime Materials Technology at the Sea Systems Directorate, Ministry of Defence, Bristol, UK. He is also the Visitng MoD Research Professor at the University of Southampton.

The sensor, which comprises a small steel shim containing a pre-crack, is attached adjacent to a critical weld detail. Under the action of cyclic stress in the member to which it is attached, the crack in the sensor extends by fatigue, giving rise to a change in the electrical output of the sensor. As Yan Zhang, Chris Allen, Peter Tubby and Jagath Mawella report, interpretation of the output allows the cumulative damage in the target joint to be estimated, providing valuable information for assessing the safe remaining life of the structure.

In order to assess the sensor's suitability for application to ship structures, a series of fatigue tests was conducted on steel plate specimens incorporating transverse welded stiffeners, with sensors attached adjacent to the welded joint. The specimens were fatigue tested to failure under constant amplitude loading in air. Sensor performance was compared with the fatigue test results for the joints and with the Class F fatigue design curve in UK design codes. The results demonstrate that the fatigue sensor system is suitable for welded joints in steel and is capable of providing advance warning of the rate at which the design life is being consumed.

Although fatigue failure should not happen in correctly designed components, there are often uncertainties as to the actual loading, environmental conditions and material properties. It would clearly be desirable to develop a sensor, which, when attached to the structure at a point of significant cyclic stress, is capable of registering cumulative fatigue damage. Such a system provides valuable information on the rate at which the fatigue design life is being expended, and can therefore be used to set inspection intervals according to usage rather than elapsed time.

A recent collaborative project managed by TWI has developed a fatigue damage sensor suitable for welded joints in steel. Known as the CrackFirst^TM sensor, it consists of a thin steel coupon with a pre-crack at its centre, attached to the structure adjacent to a critical weld detail. Cyclic loading applied to the structure induces a cyclic stress in the coupon, resulting in propagation of the pre-crack. The extent of crack growth in the coupon, sensed electronically by the breaking of the conductive tracks on the surface, provides a direct measure of the Miner cumulative damage in the target joint. Sensor electronics allow the crack length to be measured and recorded and subsequently downloaded to a PC for analysis. Validation tests conducted within the collaborative programme have demonstrated that the sensor has a fatigue performance similar to Class F in the current fatigue design rules. This class represents the fatigue performance of steel plates with transverse non-load-carrying fillet welded attachments.

In a recent project at TWI for the Ministry of Defence to investigate the suitability of hybrid laser arc welding for ship construction, the fatigue performance of these joints were of primary concern. Several series of joints were fatigue tested, including specimens with transverse, T-type fillet welded non-load-carrying attachments simulating stiffener joints to the main hull plating. During fatigue testing of these specimens, the performance of the CrackFirst^TM sensor was evaluated with respect to that of the hybrid laser-arc (HLA) welds made by TWI and metal-cored arc welds provided by another company. This paper reports the experimental details and the findings.

Sensor design

Figure 1 shows a fully assembled CrackFirst^TM sensor. It comprises a steel shim 0.25mm thick, 20mm wide and 51mm long, with openings at each end, and a central slit, which acts as a pre-crack. A series of coatings is applied to the shim to provide corrosion protection. An array of twelve evenly distributed conducting tracks on either side of the central slit provides the means of detecting crack length in the shim, namely the tracks severed by crack propagation from the slit result in loss of electrical continuity. The design of the track array allows the crack length on each side of the slit to be registered independently. The fully assembled sensor incorporates a steel reinforcing pad at each end of the sensor shim in the area where it is attached to the structure, and connectors for the sensor electronics.

The sensor was designed to suit welded joint details with a fatigue strength close to the Class F design curve according to BS 7608. Joints in Classes E and F2, ie one class either side, can also be assessed with reasonable accuracy. The sensor's fatigue performance therefore matches that of many of the most common fatigue-limiting joints in welded structures.

Interrogation system

The crack length on each side of the sensor is checked and recorded in memory by sensor electronics installed on the structure with the sensor. The sensor electronics is shown in Fig.2. Sensor data can be downloaded to a PCvia a communication port.

The sensor is attached to the structure by a combination of threaded studs and adhesive bonding. A capacitor discharge welding (CDW) gun is used to attach four threaded studs to the structure at the required locations. The sensor is then mounted on the studs and retained in place with nuts tightened to a specified torque. Adhesive bonding is achieved by injecting a two-part low temperature curing epoxy resin through the central opening in each reinforcing pad to provide a bond between the sensor and the substrate at each end.

Since welded joints in the as-welded condition contain high tensile residual stress at the point of fatigue crack development, the sensor is pre-stressed in tension during installation using a jig, Fig.3, in order to achieve the same response from the CrackFirst^TM sensor. To protect the sensor and electronics from mechanical damage and corrosion, a sealed enclosure is installed over the sensor installation, Fig.4.

a) prior to fixing the electronics in the protective enclosure

b) prior to sealing the protective enclosure (the sensor shim is beneath the sensing electronics)

c) after sealing the protective enclosure (showing the communication port)

PC software allows data stored by the sensor electronics to be downloaded and interpreted. The reading number is displayed, together with a status code indicating the result of a routine which checks the integrity of the system at the time each reading was taken. The data indicating the crack length for each side of the sensor shim are used to estimate the expended fatigue life, expressed as a percentage of the design life for the appropriate joint class according to BS 7608.

Experimental details

Fatigue testing

Two series of test samples, each consisting of steel plate with a transverse, non-load-carrying fillet welded attachment, which is designated as Class F according to BS 7608, were tested in the development of a hybrid laser-MAG welding method. Nine joints were made by laser-arc welding at TWI and nine by metal-cored arc welding at another company. The specimens were 8mm thick, 100mm wide and 500mm long with a transverse non-load-carrying stiffener attachment weld at the specimen centre, see Fig.4. Angular misalignment was observed in all specimens. The degree of misalignment was measured and was used subsequently in the calculation of stress concentration factor, Km, as will be described later. The metal-cored arc welded specimens experienced greater welding distortion, resulting in larger angular misalignments.

A total of ten sensors were evaluated by attaching each on one test specimen: five on the HLA welded specimens and five on the metal-cored arc welded specimens. Each sensor was attached adjacent to the joint, with the centre of the sensor 50mm from the weld toe, and with the axis of the sensor aligned parallel to the longitudinal axis of the specimen.

Cyclic load was applied to the test specimen to determine its fatigue endurance and to evaluate the fatigue performance of the sensor attached. Fatigue tests were conducted under constant axial stress amplitude in air, using hydraulic testing machines, and continued until complete separation of the specimen. A stress ratio (ratio of minimum to maximum stresses) of 0.1 was applied to all test specimens. Two strain gauges were installed adjacent to the weld on each specimen to determine the local stresses. They were installed back-to-back on the centreline of the specimen, with the gauge adjacent to the weld, centred 5mm from the weld toe.

The stress concentration factor (SCF), K_m, near the weld due to misalignment was determined using the information from the strain gauge measurements as follows.

Assuming the measured strains at the weld cap and at the opposite weld root are respectively ε_T

and ε_R , the average strain in the section is

Thus, the nominal applied strain is increased at the weld toe by a factor of: 

The local stress range was determined by multiplying the nominal applied stress range by the corresponding K_m.

Sensor installation

The attachment method for the sensor developed previously was adopted in this project. Specifically, each sensor was pre-stressed to a tensile stress of ~120MPa before it was installed on the CDW studs and tightened. The magnitude of this tensile pre-stress was chosen such that the maximum total stress (the sum of the pre-stress and the maximum stress in the cyclic loading) will not exceed the yield strength of the shim steel. A two-part epoxy resin adhesive was injected through the opening and was allowed to cure for a few minutes at room temperature before the pre-stressing jig was released. As each test plate specimen was concave towards the stiffener side due to welding induced distortion,and the sensor was applied on the concave side, additional mean stress was introduced in the sensor when a tensile load was applied to the test plate.

Since the tests were conducted in the laboratory, the protective enclosure was not necessary. However, sensor No.8 was tested with the enclosure in place, Fig.4, to provide a realistic trial of the complete sensor system.

Test results

The fatigue test results for the welds are compared with the performance of the sensors in Table 1. Stresses are given in terms of the remotely applied stress range and the larger local stress range at the joint resulting from the angular misalignment. Lives quoted for the specimens are for complete separation. For the sensors, endurances quoted for each crack tip are for failure of the final track, after which the sensor output is saturated.

Table 1 Summary of the test results for the fatigue sensors

The test results, expressed in terms of local stress range, are presented in Fig.5 where the Class F mean and design curves are also included for comparison. It should be noted that results for all specimens, nine of eachtype, are plotted in Fig.5, although only five of each type were fitted with sensors. Angular misalignment results in a secondary bending stress which is tensile on the concave side of the joint and decays to zeroapproximately half way between the specimen centre and where it is gripped in the machine jaws. As a result, the local stress range at the joint was higher than that at the point of attachment of the sensors, and this was allowed forwhen plotting the results in Fig.5. Because only angular misalignments, not axial misalignments, were present in these joints, the secondary bending stress induced by misalignment can be reasonably assumed to be linearlydistributed. From the value of K_m at the positions of the strain gauges (5mm from the weld toe), the average SCF for each sensor was estimated at its centre position. The local stress range for each sensor wasdetermined by multiplying the nominal applied stress range by the corresponding SCF at the sensor. The SCFs and the local stress ranges at both the strain gauges and the sensors are summarised in Table 1.

Discussion

The results obtained demonstrated that the sensors performed well in laboratory tests. It was established that the electrical output of the sensor system changed reliably throughout its life in 12 steps, Fig.6 and 7. As indicated in Fig.5, failure of the final sensor track, ie the twelfth step change in its electrical output, corresponded closely with the Class F design S-N curve. In a structural monitoring application, with the sensor installed adjacent to a Class F detail, the twelfth step change in output would indicate that the structure or component had reached the end of its design life. At this point various actions could be considered, such as inspection and repair if necessary, increasing the frequency of inspection, retiring the structure from service or modifying the loading. Since the sensor output is related only to the cumulative applied loading, it is not an indication of the condition of the joint itself. The target joint may or may not show evidence of cracking; the joints tested here survived significantly beyond the Class F design endurance (see Table 1). In applications where the decision is taken to extend the life, the failed sensor could be replaced to allow monitoring to continue.

Because of the difference in angular misalignments, the metal-cored arc welded specimens experienced higher bending stresses than the HLA welded specimens. When plotted in terms of the local stress range at the sensor, Fig.5, it appears that sensors attached to the metal-cored arc welded joints had slightly greater endurance than sensors on the HLA welded joints. This suggests that the performance of the sensor, to some degree, depends on the loading mode. In situations where bending stress constitutes a large proportion in the total stress range the sensor life may exceed the Class F design curve, in which case, it may be cautious to base assessment on the life to separation of the eleventh track.

As discussed earlier, successful application of sensors to a structure will require knowledge of the fatigue design, critical locations, stress direction. In general, the requirements are similar to those for monitoring systems based on strain measurement, except that details of peak stress magnitudes and cyclic frequency are not required. Sensors should ideally be aligned with the direction of maximum stress range, so that cracking progresses straight across the sensor shim; however, the design allows for some deviation from the correct angle without loss of sensitivity. The physical dimensions of the sensor prevent its application to very small components.

Conclusions

• All sensors failed before the test specimens to which they were attached indicating that the sensors could be used to predict the failure.
• The sensor had a fatigue characteristic similar to the Class F design curve of BS 7608, and an approximately linear response to the number of cycles at a given stress range.
• The sensor provided a reliable and sequential indication of design fatigue life consumed under cyclic loading.
• Sensor attachment to the structure by a combination of CDW studs and epoxy resin adhesive gave satisfactory performance.
• The fatigue endurance of the sensor, to some degree, depended on the loading mode. Sensors mounted on specimens subjected to a higher degree of bending stress gave marginally longer lives.

Acknowledgement

The CrackFirst^TM system was developed through the collaboration of TWI Ltd., FMB, Micro Circuit Engineering Ltd., UMIST and Caterpillar Peterlee (a Division of Caterpillar (UK) Limited) in a project funded by the UK's Department of Trade and Industry LINK Sensor and Sensor Systems for Industrial Applications Programme.

This CrackFirst^TM validation project has been supported by MOD, UK.