Conference report Measurement and control technology for hostile environments in the metals industry S. Riaz*, M. S. Millman, I. Baillie, M. Adderley, J. van Boggelen, A. W. Smith, A. Chown, D. Smart, P. Kitson, O. Milling and K. G. Crudgington This informal conference, one of a regular series organised by the Iron and Steel Society of the Institute of Materials, Metals and Mining, was held in London on 22–23 November 2006. The meeting was wide ranging in scope and aimed to familiarise delegates with issues in manufacturing and online application of measurement and control technologies under the harsh environments typical of the metals industry. The 44 delegates from 11 countries included 21 delegates from various steel producing companies, 15 from engineering companies and marketing and eight from research organisations and universities. In all, 18 presentations were made covering vision, acoustic, electronic, and ultrasonics technologies for vibration control, shape measurements, temperature measurement, surface quality and crack detection and characterisation. Condition monitoring Dr Steve Dixon (University of Warwick, UK), the conference chairman, welcomed the delegates and speakers then introduced Professor L. Gelman (Cranfield University, UK), the keynote speaker. The focus of Professor Gelman’s address was the condition monitoring of multistage gearboxes. A system developed in the laboratory to detect damage on the teeth of individual gear wheels was described. Results were presented for a gearbox having artificial damage on a single tooth. However, the results presented for tests on operational gearboxes were not conclusive in identification of damage. The cost of the system was said to be y£1000, though this figure appeared to be based on the prototype system that requires further development to ensure survival in an industrial environment. The novel processing of accelerometer data on which the system is based was not described in detail: significant issues remain with how accelerometers are being used and data analysis. Overall, the system in its present state appears less advanced than systems that are currently available commercially. In discussion, it was stated that the system works in static or adaptive mode. For moving environments good adaptation is required and accelerometers are used to filter out the noise. Conventional technology can detect defects/damage at 40–50% of the pitting size whereas the detection limit of the new system is 10–15%. The system can detect cracks, but cannot at present differentiate between different types of damage. Vision This session, chaired by C. Lester (Corus RD&T, Swinden Technology Centre, UK), began with a presentation on ‘Camber, manifest shape and thermoprofile’ by R. Dowdey (Shape Technology Ltd, Christchurch, UK). The use of stereoscopic linescan cameras in the hot rolling mill environment was described for camber, manifest shape and thermoprofile measurement. The number of cameras used depends on the parameter to be measured. It was shown that use of two or three stereoscopic cameras allows the actual camber of a rolled plate to be determined by elimination of lateral movement. Similar examples were shown for shape problems and thermoprofiling. The system described was capable of presenting up to 2000 measurements per second thus enabling the system to operate within the level 1 control and feedback system to enable online correction of shape and chamber problems. The system was capable of operating with infrared radiation (600–1000uC emission range), thus eliminating the need for expensive back-lighting systems. Lens distortion is taken into account by a special calibration procedure and the cameras are guaranteed for a lifetime of 15 years. J. Læssøe (JLI vision a/s, Denmark) gave a lively presentation on ‘Vision systems for hostile environments’, boldly stating that his company will design and install vision systems (light and infrared) that nobody else wants to or no other company is capable of. Examples were presented from the company’s early days back in 1982 to the present day. The first example, when cameras had resolution of only 2506250 pixels, had very poor accuracy of 20 mm, but software integration of the measurement allowed this to be improved to 2 mm. Many other problems were also overcome using software algorithms. For example, scale inaccuracies were effectively ignored by using ‘minimum peak picking’ and steam obscuration was overcome by waiting for three consecutive similar images before taking a reading. With this system JLI was able to measure slabs to ¡3 mm accuracy. Impressive for 1982. The second example presented, from 1989, was a bloom-cropping measurement system installed at Stocksbridge. The system had been developed to estimate the amount of discard material and identify ‘fish tail’ or ‘grit end’. It had an accuracy of ¡2 mm, used the cropping head to calibrate itself and had worked continuously for 14 years. The next example was a system to measure bar up to a maximum length of 120 m, for which it was necessary to use five cameras mounted high up suspended from the roof. The main developmental problem was the necessity to adjust the cameras during production. This required precarious Corus RD&T UK. *email shahid.riaz@corusgroup.com ß 2007 Institute of Materials, Minerals and Mining Published by Maney on behalf of the Institute DOI 10.1179/174328107X203868 Ironmaking and Steelmaking 2007 VOL 34 NO 4 273