Measure for measure - getting it right

Christoph Wiesner After obtaining his engineering degree in Germany and his doctorate at the Swiss Institute of Technology, Christoph joined TWI in 1991. Following 12 years (as Manager since 1998) in the Structural Integrity Technology Group, Dr Wiesner became Director, Research and Technology in 2002/3.

Bulletin presents an abridged script of a presentation given at the Ministerial Networking Event of the 2004 National Measurement Awards, to an audience including Lord Sainsbury of Turville, Parliamentary UK Under-Secretary of State for Science and Innovation.

The talk, by TWI's Director of Research, Christoph Wiesner, was to a mainly non-expert audience and is therefore very much of an introductory conversational nature. However, detailed TWI expertise and previous work provided the foundation for this overview.

Through a series of TWI case studies, it is intended to illustrate the importance of measurement for the benefit of sustainable and safe economic development.

Why measure?

A vital purpose of measurement is to avoid failure. The historical example in Fig.1 shows the remains of a 42m diameter storage tank which failed catastrophically during hydrotesting. Luckily no-one was injured, but the economic damage was very significant. What happened was that it was acceptable practice then to cut samples from weld metal to check whether their composition was to specification.

The cut-out was then repaired by welding. Unfortunately, the repair was incomplete and the heat cycle of the repair embrittled the steel surrounding it, providing an ideal crack starter. No toughness measurements or non-destructive examinations were carried out at the time, but this was changed as a consequence of this and similar failures of the period.

TWI has contributed much to the understanding of brittle fracture for 40 or so years by carrying out many large-scale experiments simulating structural behaviour. The set-up in Fig.2 shows a large steel plate being loaded in two directions to simulate the biaxial stress state which exists in pressurised vessels and pipes.

Brittle fracture is a failure mode which occurs without warning and can completely fragment the component, which is why its avoidance is so important. Toughness is therefore an important measurement to take on all steel structures. It is very exciting to witness tests like this as the failure point is never quite certain and most witnesses tend to jump whenever the failure occurs.

Failure of large structures is spectacular and always dangerous, but nowadays failure of micro-components can be just as risky. Figure 3 shows that the range of measurements at TWI spans from the very large to the very small. For micro-components, precision equipment is used to determine failure strength in support of improved microelectronics design and reliability.

Large-scale experiments are expensive. Representative small-scale measurements have therefore been developed. Figure 4 shows a toughness test specimen made of plastic - materials joining technology at TWI covers all material types.

This is a drop-weight experiment where the speed of loading is aimed at representing what could happen to a joint like this in practice during a collision or an impact event. Materials behave quite differently when loaded at high speed and it is important to check this rate dependence.

Figure 5 shows a further example of tests aimed at avoiding failure. Figure 5a illustrates how lampposts can bend due to wind loading. Many repeated sway-cycles cause fatigue damage, particularly at certain design features, such as attachment welds where local stresses can be very high. A number of lamp post failures have occurred which prompted the relevant public body to get TWI involved to investigate.

TWI simulated the loading experimentally ( Fig.5b), identified the weak areas and recommended design improvement to avoid failure.

Household goods can also suffer from fatigue. The manufacturer of the exercise cycle in Fig.6 did not measure the appropriate property for its base and, ironically, vigorous exercise by TWI's Fatigue Technology Manager, Dr Stephen Maddox, resulted in this failure, luckily not catastrophic. As TWI has repair technology expertise, we were able to help in the repair, see Fig.7.

The final example in this section is a case of specialist property determination to facilitate 'controlled failure', see Fig.8. The idea at the time was to decommission steel rigs by toppling them and create a sub-sea marine environment. A large-scale multi-axial collapse rig was designed and built to simulate the buckling events of tubulars and provide material property data for the validation of computer models.

Measure what?

It is an easy thing to say, but often finding the right thing to measure is not obvious. The first example to illustrate this concerns fume exposure during welding and what the appropriate safe limit should be. A lot of attention has been focused on how exposure is affected by weld parameters, type of materials and the weld environment, while overlooking the significance of the welder's head position ( Fig.9).

By measuring exposure using a sensor installed behind the helmet, it became clear that other factors are actually secondary to head position, see Fig.10.

A further failure in 1984, that of a 17m high refinery tower, which used to stand in Lemont, Illinois, see Fig.11, also provides an example illustrating the need to measure the right thing. The refinery tower developed a gas leak, which was noticed by an operator who tried to close off the main inlet valve, and evacuate the area. Unfortunately, as the company fire fighters arrived, the gas ignited, creating a massive explosion which sent the tower into the air like a rocket, landing over 1km away: 17 people were killed.

One course of the tower had been replaced earlier, and fracture occurred in the weld heat affected zone by a hydrogen cracking mechanism. When analysing the failure, fracture mechanics calculations were used, together with fracture toughness measurements made on samples from the failed plant to see whether it could have been predicted. However, these measurements did not take into account the presence of hydrogen that tends to reduce toughness.

Only when the toughness of hydrogen-charged material was measured, could the failure be explained. The failure exposed the need to have well-qualified repair procedures and take into account environmental effects when measuring relevant properties.

The last example in this section shows an impact test sequence the of an Al-alloy sample, representing a structural train coach component during 100kg drop-weight testing at 30mph ( Fig.12).

The structure is designed to withstand impact. A novel mechanical fastening technique is used in this component, and the test is to show that the structure buckles and crumples in the right areas to absorb sufficient energy in a simulated crash situation. Deformation behaviour of many materials is dependent on the rate of loading and the right measurement has to apply this rate.

How to measure?

The next question focuses on new test methods to measure things more accurately, or more cheaply, or to allow measurements which were not previously possible. The first example is a fatigue gauge ( Fig.13) under development at TWI which allows the measurement of cumulative fatigue damage. It is well known that during the repetitive loading experienced by such structures as bridges, or the lamp post in Fig.5a, their fatigue lives are being 'used up'.

Because of uncertainty in loading ( eg how many heavy lorries travel over the bridge or how many storms shake the lamp post), it is not possible to know how much fatigue damage has occurred during a given time. This means inspection is carried out at regular intervals, often not finding anything. With this development, the amount of life 'used-up' can be measured and used for maintenance and life extension decisions, and comparison with laboratory data allows validation of the gauge.

Figure 14 shows the whole electronic packaging of the sensor, attached to an articulated lorry component for field trials, the gauge being attached to the steel beneath the electronics.

Another example of how to measure relates to the better understanding of welding processes. Laser welding is used in industry, but it is quite demanding with respect to the accuracy required for fit-up. By combining a laser with a conventional arc welding process, so-called hybrid laser arc welding, productivity and tolerances improve significantly, see Fig.15.

TWI has employed new powerful high-speed video equipment to look into the welding processes taking place. This region is a hostile place: temperatures in the arc are up to 20,000°C (hotter than the sun's surface), and the various plasma jets emanating from the laser key hole and in the arc move at more than 200mph. Visualisation of the process allows better understanding of the metal deposition and helps to improve it. A particular interesting feature here is the weld droplet emanating from the arc welding filler wire 'bouncing' off the plasma jet emerging from the laser key hole, see Fig.16.

A further example of new measurement techniques is from the medical sector where TWI's polymer experts have developed a surface treatment for biopsy needles which allows them to be more clearly visible in a clinical ultrasound scanner, Fig.17.

The effervescent coating produces bubbles when in contact with water or wet body tissue. It is these bubbles ( Fig.18) which scatter the ultrasound of the scanner and make the needle visible, as can be seen in Fig.19, allowing the surgeon to see where he is guiding the biopsy needle during examination.

When to measure?

Our next question is when to measure? The first answer to this question is when it is convenient! As illustrated by the graph in Fig.20, it costs disproportionately more to measure things later in a project than early on during manufacture. Once a problem occurs in-service, for which the relevant property has not been measured earlier, the cost of obtaining it is very significant, as manufacture or production may be interrupted. However, in some cases measurements are needed at very specific times. A further example of a failure illustrates this. The Cockenzie boiler drum ( Fig.21) failed in 1966 during hydrotesting. No one was injured thanks mainly due to the connecting pipework which prevented the failed pieces from being thrown too far. The origin of the failure was a large (330mm long by 90mm deep) crack which had formed during final post weld heat treatment (PWHT) by a now known re-heat cracking mechanism. The vessel had been inspected, but prior to the final heat treatment, before the flaw was formed. This failure caused changes to the design and manufacturing standard to require non-destructive measurements after final PWHT.

With whom to measure?

The question with whom to measure is a commercial question, although access to specialist facilities also has a political angle to it. What is clear is that very specialist test facilities and expertise are beyond the resources of many that need them or would not be justifiable for occasional use. For example, equipment for laser surface profiling of flip-chip arrays used in the microtechnology sector, is not something every small enterprise would be able to afford ( Fig.22).

Figure 23a shows how the TWI integrity assessment methods, developed for corroded pipelines, are being validated by testing to failure a section of corroded pipe. This enables the actual and predicted failure pressures to be compared, Similarly, Fig.23b shows a validation burst test for an Al-alloy rocket fuel tank sample that is part of the introduction of TWI's friction stir welding process into the manufacturing of this structure. Rather than developing such specialist facilities and financing and maintaining the resource requirements going with them, it is better to outsource experiments like this to expert providers such as TWI.

Interpretation and application of test results

The final point to make is the importance of using measurements wisely, using experience and appropriate judgement. The example used here is fracture mechanics-based flaw assessment, see Fig.24 for an illustration of the principle. By measuring toughness, the resistance of a material to fracture is obtained; and by determining the flaw size, using non-destructive test methods such as ultrasonics, the tolerable driving force or load can be calculated.

One complication of this type of analysis is the very significant scatter one obtains when measuring toughness, particularly in the transition region (see for example Fig.25). At -20C, the Charpy toughness varies between around 20 and 130J, more than a 6-fold difference between lowest and greatest. So which value should we use for calculations? This requires experience and engineering judgement, available from the Structural Integrity experts at TWI.

Equally, the measurement of flaw dimensions using ultrasonics can be very scattered. Figure 26 shows the results of an HSE-sponsored UK trial to quantify the accuracy with which manual ultrasonic testing can size flaws in steel welds. The operators were experienced, certified inspectors who knew that what they measured would be looked at. Still, the sizing was often inaccurate. The particularly worrying outcome was that large flaws were often undersized whilst innocuous small flaws were oversized. Again, specialist expertise is required to make sense of such measurement results, and this is available from the NDT group at TWI.

Summary and conclusions

The final example particularly illustrates that there is still much progress to be made in measurement and the UK National Measurement Awards are encouraging further improvements to measurement methods and their applications in all industry sectors.

To conclude, a summary is given below in the format of succinct answers to the broad questions raised. They appear obvious, but are not always easy to get right in practice.

• Avoid failures by measuring
• Measure the right thing
• Use the appropriate method
• At the right time
• Consider partnering and
• Train people to interpret and apply the results correctly.