8.52 existing system could reliably detect surface cracks as small as 2.5 mm during the manufacture of the plates. Flaws 10mm x 10mm are reliably detected in industry in 40 mm thick plates at lm/sec inspection speed by special 50 mm water-coupled ultrasonic transducers (8.29). Scaling down this system to handle one meter wide plates requires an array of 158 standard American-size 0.250 in.transducers to cover the entire plate width. Using large arrays of transducers is the only practical way to obtain high-accuracy, wide-area, high-speed automatic inspection. 0.250 in.ultrasonic transducers currently can detect 5 mm cracks (8.25), and future electromagnetically coupled ultrasonic transducers should allow detection of the assumed 2.5 mm cracks. If these transducers become available in larger standard sizes, the array can be redesigned and simplified to use the new transducers. VIII.8.3.2: Fastener Inspection: All fasteners are ultrasonically inspected by a large single-chip transducer array. Flaw location accuracies of about 4 mm can be obtained at 25 mm/sec scan rates (8.28). The transducer chip is held stationary while the fastener is rotated, allowing 360° single-pass scans. For a 30 mm diameter fastener, the inspection time will be about 3.7 seconds per fastener, or almost 1000 fasteners per hour, well within the capability of available automatic small parts handling and inspection equipment (8.30). VIII.8.3.3: Plate Proof Testing: Samples of plates are proof tested, some to destruction, to determine the actual achieved material characteristics of the plates. If the steel has an ultimate strength of 150 kpsi, a lm 2 x 2 cm thick plate would require a load of nearly 5xl0 6 lbf to break it in tension. Constructing a five million pound tensile testing machine in space would be thoroughly impractical. It is much simpler to return an occasional sample of plates to Earth via the Shuttle and use existing earthbound facilities. The assumed Shuttle carries about 150 plates, a generous sample size. The most efficient sampling schemes are adaptive: the size of the sample, number of samples required, and even the level of confidence in the accuracy of the measured data vary according to the data itself (8.31). In a general sense, the less variation in
RkJQdWJsaXNoZXIy MTU5NjU0Mg==