On line and global structural health monitoring of high temperature steam lines

目前，蒸汽管道焊缝的超声波探伤需要在停电时取下管道，搭设脚手架，这些过程所涉及的时间和费用意味着每次停电只检测20%的焊缝，需要10年时间才能实现100%的完全检测。因此，一些缺陷很可能一直没有被发现，直到它们被发现(超过管道CSA的50%并不少见)，并且故障概率相应地很大，需要进行大的和昂贵的维修变得迫切和必要。100%焊接检查的最明显的方法，第一眼是在焊缝附近放置一个传统的MHz频率压缩和角度波束探头的环形阵列，即在停电时应用的现有检查方法中使用的探头类型。然而，为了覆盖整个焊缝厚度，角度束探头必须机械地扫描垂直于熔接线的一个跳跃距离。此外，还需要在高温下具有有限寿命的流体耦合。可以使用寿命不确定的替代永久高温胶粘剂，但禁止机械移动。更好的解决方案是使用位于两个焊缝之间的管段中央的单个环形阵列来传播导波模式。通过使用更长的波长，阵列中均匀间隔的换能器的数量可以减少到8个，甚至减少到4个，同时仍然能够以较佳的波模式相当均匀地激励整个管道焊接，从而以比先前选项(1)和(2)低得多的成本提供阵列。在250 MHz，这意味着波长约15 mm，传播范围是到达每个管道末端的焊缝，通常长达6米。频率降低，例如降至25 kHz，波长约150 mm，将增加覆盖几个管道长度的量程，成比例地降低100%检测的设备成本，但代价是降低检测灵敏度(增加可检测的最小缺陷尺寸)。对于缺陷反射系数的准确预测以及因此最小的可检测缺陷尺寸，只有使用诸如有限元分析之类的波形建模技术才有可能，但是使用基于灵敏度下限和上限的经验法则来获得数量级的估计是更快和有用的。上述两种理论估计严格地只适用于阵列附近，因为它们没有考虑阵列缺陷的深度或沿该深度的模式转换损耗。然而，当波被引导时，波的强度不会像三维或二维伸展波那样下降到深度-2或深度-1，这确实是为什么对于任何给定的频率和波长，导波的射程更长的原因。此外，通过安排所传输的波包含实际可能的最高比例的单个非色散模式，可以最小化模式转换损耗。这取决于阵列设计。在管道中用于导波的实际工作频率下，即使在最高预定工作温度下，管道材料中的波吸收也很低，因此对降低缺陷检测灵敏度几乎没有贡献。综上所述，该估计值可以可靠地应用于~6m的光焊间距。影响实际获得最高可行灵敏度的另一个因素是波强沿半径从内墙到外墙的分布。内表面和外表面具有强极大值的分布有利于同时检测内壁和外壁蠕变和疲劳裂纹细节。这说明了在项目中建模的重要性，以确定可能连续使用的最佳模式短列表，以实现最佳的100%墙体积复盖率。

Presently ultrasonic inspection of steam line welds involve taking off the pipe lagging and erecting scaffolding at outages.The time and expense involved in these processes mean that only 20% of the welds are inspected at any one outage and ittakes 10 years to achieve complete 100% inspection. So it is likely that some defects remain undiscovered until they are solarge (in excess of 50% of pipe CSA is not uncommon) and the failure probability correspondingly large that the need formajor and expensive repair becomes immediate and essential.The most obvious approach to 100% weld inspection, at first sight is to place close to the weld a ring array of conventionalMHz frequency compressional and angle beam probes i.e. the type of probes used in existing inspection methods appliedat outage. However, for coverage of the entire weld thickness the angle beam probes must the mechanically scannedthrough one skip distance normal to the weld line. Also fluid coupling is required with a limited lifetime at high temperatures.Alternatively permanent high temperature adhesives, of uncertain lifetime can be used but mechanical movement is thenprohibited.A far better solution is to use a single circumferential array located centrally on a pipe section between two welds phased topropagate a guide wave mode. By the use of longer wavelengths the number of transducers in the array, evenly spaced,can be lowered to 8 or even as little as 4 whilst still being able to insonify the entire pipe welds fairly evenly with a preferredwave mode, so to provide an array at far less cost than the previous options (1) and (2). At 250MHz, which implieswavelengths ~15mm, the propagation range is reach the welds at each pipe end, typically up to 6m. Reduction of thefrequency, for example down to 25kHz, wavelength ~ 150 mm, would increase the range to cover several pipe lengths,proportionality reducing the equipment costs of 100% inspection, at the expense of reduced inspection sensitivity(increased minimum detectable defect size). Accurate prediction of defect reflection coefficients and hence minimumdetectable defect sizes is only possible with a wave modeling technique such as finite element analysis but it quicker anduseful to obtain order of magnitude estimates using rules of thumb based on lower and upper sensitivity limits. Both theabove theoretical estimates apply strictly only close to the array because they take no account of the depth of the defectfrom the array or mode conversion losses along that depth. However as the waves are guided the wave intensity does notfall off as depth-2 or depth-1 as would 3 or 2 dimensional extensional waves, which indeed is why guided waves are oflonger range, for any given frequency and wavelength. Also mode conversion losses can be minimized by arranging for thetransmitted waves to contain as high a proportion as practically possible of a single non dispersive mode. This depends onthe array design. At the practical operating frequencies used for guide waves in pipes the wave absorption in the pipematerial is low, even at the maximum intended operating temperatures and will thus contribute little towards reducingdefect detection sensitivity. So in conclusion of these considerations the estimated can be credibly applied to the intendedarray-weld separation of ~6m. A further factor affecting the practical achievement of the highest feasible sensitivities is thedistribution of wave intensity along a radius from the inner to the outer wall. A distribution with strong maxima at the innerand outer surfaces is advantageous for simultaneous detection of inner and outer wall creep and fatigue cracksparticular.This illustrate the importance of modeling in the project to determine an optimum short list of modes which mightbe use sequentially to achieve the best 100% wall volume coverage.