This paper provides a state-of-the-art review of the techniques in the recent development in infrared measurement technology for experimental stress analysis. These techniques are based on the use of radiometric systems to determine by non-contact means the thermoelastic effect in structures and materials subjected to dynamic loading. The recent advent of highly sensitive infrared systems, some of which are capable of measuring dynamic temperature changes as small as 1 mK, has led to versatile instrumentation for the evaluation of stresses in engineering models and actual components. The most important feature of the techniques is that it provides a unique method of direct measurement of stress in a material of a test component by non-contact means. Methods have recently been developed to determine dynamic stresses over a frequency of less than 1 Hz up to many KHz. It is now recognised that the technique has become a new important tool for stress analysis and already considerable experience has been gained in practical application.

The last few years have seen the development of the thermoelastic technique into a quantitative stress analysis tool of major importance. The contribution to this development from the University of Manchester is reviewed in this paper.

It has been demonstrated that the response of the SPATE 8000 depends not only on the alternating stress experienced by the object being examined but also, to a small extent, upon the mean stress about which the alternating stress is superposed. The 'thermoelastic constant' varies linearly with mean stress for each of the aluminium, steel and titanium alloys studied.

Over recent years the thermoelastic effect has been exploited successfully to measure full-field surface stress distributions on structures subjected to uniform cyclic loading conditions. The technique can provide engineers and designers with considerable insight into the quasistatic behaviour of structures. Software developed at the National Engineering Laboratory has enabled the thermoelastic technique to be extended to the variable-amplitude loading situation which is typically encountered in service applications. This paper analyses the theoretical basis of random signal processing, and presents experimental data from thermoelastic stress analysis tests performed under random-loading conditions, including modal behaviour.

The amplitude of surface temperature oscillation induced thermoelastically by cyclic loading is analysed through solutions of the heat-conduction equation. Significant attenuations may occur under typical operating conditions as a result of internal conduction and the presence of surface coatings. Attenuation due to heat loss to the environment will be significant only under rather particular circumstances. Comparison is made with analytical and experimental work of earlier investigators.

In this paper the experimental methodologies based on linear response theory are applied to the thermomechanical behavior of a solid body subjected to external load histories. This is possible due to the linearity of equations of thermoelasticity provided macroplastic flow is avoided. In particular the techniques of spectral analysis of noisy signals passing through linear systems are employed to obtain the Bode diagram of the stress-temperature transfer function of a thermoelastic solid. Both sinusoidal and white noise excitations are used and coherence function tests are included to assess the statistical goodness of the results. From these measurements the Gruneisen parameter of the solid can be estimated whithin the linear thermoelastic region. The experimental apparatus, based on a PDP 11/03 microcomputer system, designed and built to reach the above goals is described in some detail together with the specifically developed software. The problems connected with the calibration of thermal sensors are outlined. Preliminary experimental results are briefly presented. Although the above methodology does imply linearity, the apparatus has also been used to perform local temperature measurements on the surface of compact tension specimens in non linear elastic - plastic conditions, namely during sinusoidal low-frequency fatigue experiments. In the case of fatigue pre-cracked specimens the thermal resolution of the apparatus was high enough to detect non linear phenomena, like e.g. local plastic flow or crack closure, in the vicinity of the crack tip. The non linear part of the thermal response can be nicely pinpointed as harmonics generation in the power spectrum of the local temperature signal.

Thermoelastic stress analysis of plastic models was successfully employed in the design and development of steel counter parts of an automobile. SPATE 8000 facilitated non-contact high resolution stress measurements of cast and vacuum formed plastic models. The stress analysis of the model was used for predicting stresses on steel prototypes. Design sensivity analysis was performed on plastic models to evaluate different design iterations. Problem areas were identified before the design was comitted to metal, thereby minimizing impact on cost and scheduling.

This paper describes a series of laboratory tests undertaken to demonstrate the application of thermoelastic measurements for use on typical woven carbon epoxy composites used in the aerospace industry. The tests undertaken showed that the woven carbon epoxy produced linear thermoelastic output with increasing load. No significant changes in thermoelastic output were observed with changes in warp direction relative to load direction. A carbon aluminium honeycomb specimen was examined and clear differences in the stress distribution were seen between the two surfaces of the honeycomb. These are attributed to local geometry changes resulting from the manufacturing methods. The results demonstrate the suitability of the thermoelastic effect to assess non-homogeneous anisotropic materials. Considerable time and cost savings could be made by the application of thermoelastic techniques to existing composite fatigue and mechanical testing programmes.

The basic theory for determining stress intensity factors by thermoelastic stress analysis is outlined. The contribution of higher order terms to the solution is included. Attention is focused in the main upon Mode I, crack tip opening, failure. Various possible methods for obtaining stress intensity factors from experimental results are discussed. The advantages and disadvantages of each technique are commented upon. The use of some of the methods is demonstrated by a theoretical study of a crack tip stress field. This highlights any difficulties that might be encountered in the analysis of data. It also gives an indication of the accuracy that can be achieved under ideal experimental conditions. Consideration is also given to the way in which experimental data can distort results. In particular attempts are made to address the problem associated with scanning at increments smaller than the available resolution.

A thermoelastic technique has been used for the appraisal of a recently proposed specimen for the determination of the edge-sliding mode stress intensity factor K_II. A relationship between the received thermoelastic signal in the neighbourhood of a crack and the associated stress intensity factor is given. A method for the interpretation of the thermoelastic data is presented. Non-dimensional stress intensity factors, obtained from specimens with machined slots, are compared with those obtained from photoelastic and finite element analyses. For the edge slot there is good agreement; for the central slot there are appreciable differences.

The equations relating the output of the SPATE* 8000 thermoelastic stress analysis system to the strains in multidirectional laminated fibre composites have been developed. The system has been calibrated for use with a particular carbon fibre/epoxy resin material. For multidirectional laminates under a generalised state of stress, it has been observed that the sensitivity of the SPATE output to the various strain components differs significantly such that even the qualitative interpretation of SPATE scans is difficult. However, there are a number of special cases which are applicable to practical structural laminates for which the interpretation of scans is greatly simplified.

In course of the optimization of a proof concept for units of industrial processing techniques the thermoelastic stress analysis (THESA) is tested as the experimental part. The possibilities of the interpretation, the comparability with the calculations and the measurements by strain gauges (SGM) as well as the mutual supplementation of calculation, use of strain gauges and THESA are investigated by tests in the laboratory and at the actual structure.

Structural dynamics for integrity assessment and condition monitoring of electrical power station plant can be approached using a variety of methods combining theoretical modelling with experimental measurements. In recent years experimental approaches have broadened to include non-contacting full-field response measurement techniques using laser holography and more recently, stress pattern analysis from thermal emission (SPATE) to obtain dynamic stress information. This paper presents two examples of the application of a SPATE 8000 camera system to the determination of the dynamic stress distributions on nuclear reactor components. In the first project a Magnox reactor compensating bellows unit was dynamically tested in the laboratory using an electromagnetic shaker to excite resonances in the frequency range 300 to 400Hz. The dynamic stress data collected is compared with finite element model prediction. The second example describes a similar modal response investigation performed on an AGR gas circulator inner casing ring which is designed to isolate the stress concentrations present. This test was carried out in the power station maintenance facility during a routine reactor overhaul.

Joints in ladder frame chassis have been studied as part of an SERC Teaching Company Schene. The joints between the cross members and side members are complex structures involving bolts, welds and/or rivets, as the cross member section can be tubular, box or C-section. It is therefore difficult to apply simple analytical methods to such joints. This paper compares the stresses obtained by brittle lacquer, strain gauge and SPATE measurements with those found from a finite elenent analysis of the joints.

This paper describes the use of thermoelastic stress analysis techniques on a cast steel tubular joint, and compares the results with other stress analysis methods. The joint analysed was a 4 scale replica of four cast joints installed on Conoco's Victor JD platform in the Southern North Sea. These Joints represented the first use of cast joints in a fixed offshore platform in the North Sea. The use of SPATE was part of a comprehensive programme of tests on two cast joints to determine the overall distribution of stress under various loading conditions and to assess their fatigue performance. Because of the complex geometry of these components a number of stress analysis techniques were used to build up an accurate picture of stress distributions and the location of stress concentrators. The use of thermoelastic techniques enhanced the understanding of the behaviour of cast x-joints under various loading conditions. Overall it is considered that the results of the research programme will greatly assist in the development of a rational design and inspection procedure for cast joints in an offshore environment.

The application of a full field vibration stress measurement technique is described. The application considered covers the range 30 Hz - 5 kHz on blades and other high energy components. These measurements are made to: (i) Validate a finite element model (ii) Measure the effect of local modifications (iii) Problem solve development incidents The relative successes of measuring at different frequencies are discussed. Full field stress measurements - using SPATE (Stress Pattern Analysis by Thermal Emission) have never been made in isolation from other techniques. SPATE's part in the overall vibration investigation and its relation to other measurement techniques is described.

The SIRA SPATE equipment for thermal stress analysis is a successful example of the application of modern instrumentation technology to the development and application of classical concepts in physics. The use of Infra Red Optics to monitor small temperature changes over the surface of a solid material has great advantages in many applications as it is rapid, non contacting and non-invasive, and produces signals which can be recorded and analysed in many ways. These advantages have already been exploited to some extent in other applications such as conventional thermography, in which the distribution of heat in a structure is monitored by an I.R. imager using a rotating scanning system, or an I.R. pyroelectric vidicon tube. Such systems are widely employed in monitoring electrical power supplies or chemical plant, but have proved of less relevance to problems of nondestructive testing. In recent years this idea has been combined with that of introducing an alternating or on-off source of heat by means of a chopped laser beam for example, and using recording equipment such as SPATE to monitor the flow heat through a solid. It can be shown that such a source produces a train of critically damped waves of length A where 1/2 A = 2{1T4 where a is the thermal diffusivity of the solid and v is the frequency of chopping and the wave amplitude falls to l/e of its initial value in one wavelength. For waves of frequency 1Hz in copper, A is about 20mm, but normally it is necessary to work at frequencies of 100Hz or more in materials of much lower diffusivity. Consequently this method is chiefly of interest in the examination of thin sheets or coatings up to 1-2mm in thickness. In another development which is more specifically adaptable 'to many problems of NDT, heat is introduced over a chosen area at a high rate for a predetermined time of between lms and about 10s. The subsequent flow through the solid is then monitored by means of an I.R. imager compatible with current TV video standards. This system enables the thermal field to be recorded every 2Oms, and the detailed study of the resulting heat flow has provided effectively a completely new NDT technique. Successful applications so far demonstrated include the detection of disbonds and delaminations in coatings and laminates, voids in castings and cracks in welds. A problem of current interest is the examination of ceramic parts, including unfired materials in which ultrasonics is impracticable. Theoretical models are now being developed to enable optimised systems to be specified for particular cases.

The SPATE 8000 stress measurement system was purchased primarily to determine dynamic stresses on operational electrical power generating plant where difficulties have arisen in using conventional methods of stress measurement. In this context this paper describes some of the requirements to enable this system to be used to obtain the minimum resolvable stress level on a steel bellows unit vibrating at a resonant frequency of 372Hz. The influence of a range of parameters including, (i) surface coatings, (ii) liquid nitrogen level, (iii) high acoustic noise (108dB at 372Hz) and (iv) environmental conditions, on the measured stresses are discussed. Problems relating to laboratory and on site calibration of the system are also presented. Finally the modal analysis capabilities of SPATE are compared with conventional modal analysis techniques using a vibrating steel plate as an example.

A 'round robin' exercise was carried out by the UK SPATE users' group to determine factors affecting the calibration of SPATE equipments. The participants separately measured, using their own SPATE equipments and calibration techniques on identical materials, SPATE output per unit range of principal stress sum. Even after known differences in factors such as system responsivity or test temperature have been allowed for, there is a considerable scatter (of the order of 20%) in the results obtained. The cause of this is not yet known. This variation causes SPATE calibration using material properties to be less reliable than calibration by 'known stress' methods.

This paper offers an overview of SPATE activities in the USA. Examples of industrial applications range from small aero-engine turbine blades tested at excitation frequencies in excess of 20 KHz to composite pressure vessels loaded at 0.5 Hz. We briefly describe some activities of the University of Wisconsin - Madison and Virginia Polytechnic Institute and State University both of which are offering some pioneering work in SPATE applications research.