Performance of the FAST System for Stress Analysis by A.K. Wong and T.G. Ryall ABSTRACT--A state-of-the-art thermographic system has been developed based on a 512 x 512 focal-plane array thermal imager. The system was primarily designed to per- form stress analysis, but has applications in other areas where small cyclic temperature variations are encountered. A detailed analysis of its performance relative to an existing equipment (SPATE) revealed that, at least for low fre- quency applications, the proposed system out performs SPATE by some two to three orders of magnitude such that high-resolution area scans can now be achieved in minutes instead of hours. It is suggested that higher frequency ap- plications (i.e., order of 10" Hz) can be achieved with more up-to-date imagers. KEYWORDS--Focal-plane array, thermography, thermo- elastic effect, experimental stress analysis, SPATE Introduction The application of thermometric techniques for structural assessment has been practiced for some time. Early inves- tigations have primarily concentrated on the study of plas- ticity (e.g. Bever et al.~) or qualitative nondestructive in- spections (e.g. Marcus and Stinchcomb2). However, it was the emergence of SPATE 3 which has made thermometric (or more specifically, thermographic) techniques a univer- sally accepted means for quantitative stress analysis. Since the early 1980's, SPATE applications have steadily grown both in volume and in scope. For a review of the appli- cations of SPATE, see Harwood et al. 4 Based on a single cadmium-mercury-telluride infrared detector, SPATE can achieve full-field measurements via a set of digitally controlled xy scanning mirrors which al- lows a raster-like scan of up to 256 • 256 pixels to be formed. One advantage of using a single detector is 'that there is no variation in the calibration (in terms of detector sensitivity and offset) from pixel to pixel. On the other hand, being able to analyze only one point at a time usually means that forming an area scan often takes quite a long time. For work requiring both high spatial and high stress resolutions (e.g. Refs. 5 and 6), it is not uncommon for scan times to A.K. Wong is Senior Research Scientist and T.G. Ryall is Principal Research Scientist, Defence Science and Technology Organisation, Aeronautical and Maritime Research Laboratory, P.O. Box 4331, Melbourne, Victoria 3207, Australia. Original manuscript submitted: November 16, 1993. Final manuscript received: August 26, 1994. Permission has been granted to EXPERIMENTAL MECHANICS to publish this paper on a nonexclusive one-time basis only. This agree- ment is subject to the Commonwealth of Australia retaining Copyright of the paper, the author and source of the paper has been acknowl- edged. No meaningful changes have been made to the article without the prior consent of the author. reach five to six hours. Such long scanning times not only translate directly to high costs of analysis, but they can in- troduce other difficulties. For example, because there can be a large time lapse between the start and the end of the scan, slow changes in experimental conditions such as am- bient temperature or even specimen configuration (e.g. propagation of the fatigue crack being analyzed) can greatly distort the resultant scans. Analyzing components which are not designed to have a long fatigue life, as in an illustrative example described in a later section, could also be a prob- lem. In the preceding paper (Ref. 7), the design and devel- opment of a system based on a 512 x 512 focal-plane array thermal imager and a real-time digital video processor was described. This system, dubbed FAST (focal-plane array for synchronous thermography), is able to deliver high res- olution full-field thermoelastic stress scans in a fraction of the time which would otherwise be required by SPATE. In this paper, a detailed analysis of the system is presented and its performance compared to that of SPATE. Appli- cation examples are also given to demonstrate its capability as a stress-analysis tool. Experimental Procedures To demonstrate the working of the FAST system and to compare the stress resolution of the system to that of SPATE, a series of tests was conducted on a metallic specimen of uniform rectangular cross sections. The uniform specimen was selected as the resultant signal is expected to be a con- stant throughout the region and thus provides a convenient means for measuring the noise relative to the signal. The specimen was a 140 mm (w) x 400 mm (h) x 6 mm (t) aluminum plate and, as usual, was painted matt black to improve its emissivity. Loading was applied by a 50-kN servohydraulic testing machine and a range of frequencies from 1 to 20 Hz and loads from 10 to 30 kN were studied. As a first step, the noise-to-signal ratios (n/s) of both systems for any given load and analysis time were estab- lished. In general, it is expected that the signal would in- crease linearly with the applied stress, and that the system noise would be inversely proportional to the square root of the analysis time. Both SPATE and FAST scans were made on a working section approximately 100 mm x 80 mm in the central part of the specimen from a stand off distance of approximately 500 mm. These scans confirmed that the region considered has an essentially uniform stress distri- bution so that the arithmetic mean and the standard devia- tion of each scan may be taken as good measures of the signal and noise respectively. Noise Analysis for SPATE While full-field area scans were made by the FAST sys- tem, it was not feasible to make high-resolution SPATE 148 9 June 1995