Inspection of Compressed Gas Cylinders Using Ultrasonic Guided WavesA. Chahbaz, V. Mustafa, M. Brassard and D.R.Hay
Tektrend International Inc
Air Liquide Canada Inc.
Corresponding Author Contact:
|TABLE OF CONTENTS|
Mechanical integrity of cylindrical structures ranged from piping to storage and pressure vessels needs to be maintained at a high level to ensure safety and reliability. Current inspection techniques are less sensitive to common defects and unable to inspect bends and other insulated cylinders and cylindrical structures. This paper describes a novel automated inspection system based on EMAT-Ultrasonic guided wave method.
In this work, several test cases are conducted on steel cylinders with real and simulated defects. Contact-free EMAT probes were designed and mounted in the system. Results from the guided wave inspection technique demonstrated good detectability of defects in longitudinal and circumferential direction of the cylinders. Moreover, results from several real-time tests confirmed the guided wave inspection system as fast, reliable and cost effective inspection tool to locate and detect crack-like and wall thinning defects in compressed gas cylinders.
Ultrasonic inspection is a very flexible method where several inspection modes are available. Ultrasonic methods detect all flaws in the path of the ultrasonic stress wave within the limits of sensitivity of the method.
In the conventional configuration, straight beam probes scan the cylindrical wall to measure wall thickness and detect lamination and (planar) defects in the plane of the wall. Angle-beam probes are used to detect the defects oriented normal to the plane of the wall.
Conventional ultrasonic C-scan can inspect the straight sectors of the cylinder but is not suited for inspection of bends and hidden parts of a cylindrical structure. Moreover, inspection of the cylinder with C-scan is very time consuming since a point-by-point measurement is needed for a complete inspection of the structure and difficulties in inspecting around bends, are disadvantages of conventional ultrasonics. However, ultrasonic measurements are more appropriate and precise for small defects.
|Fig 1: Scan of cylindrical structure|
This work evaluates the ability of ultrasonic guided waves to detect several types of defects in cylindrical structures with different pre-selected wave modes. We also demonstrate ultrasonic guided wave method as an alternative to standard ultrasonic techniques and how to address their deficiencies. This approach is based on the use of ultrasonic guided waves or, generally, the resonant modes of propagation in the cylindrical geometry.
Guided wave or more general resonant mode propagation in cylinders is complex due to it curved geometry where complex three-dimensional propagation of waves is produced. In addition, the attenuation mechanisms (and the anisotropic nature of the material in some cases) through the thickness of their material lead to various distortion phenomena. The resonant modes of an isotropic cylindrical structure have been extensively studied [1,2,3]. However, in this section, we will only explain how resonant modes are related to Lamb wave modes of an isotropic plate and whether this basic understanding is still useful in the presence of moderate curvature.
The problem of a hollow circular cylinder has been investigated in an exact formulation without any of the approximation usually associated with thin cylindrical shell theories [4,5]. Dispersion curves for all cylindrical shell modes also were derived; it was demonstrated that a subset of these modes is reducible to Lamb waves in the limit of small cylinder curvature (large radius) .
Generally, the resonant wave modes of cylindrical structure belong to two groups. The first group represents resonant wave modes propagating in the axial plane (axially symmetric modes) and the other group are modes propagating in the transverse plane of the cylindrical coordinate system (non-axially symmetric modes). Each group by itself can be divided further into two types of resonant modes. One type is unique to cylindrical geometry and the other type is equivalent to Lamb wave modes of the flat plate .
The two sets of resonant wave modes which are equivalent to Lamb waves are called the planar and axial modes. The planar modes propagate circumferentially and the axial ones along the cylinder. In this work, the axially symmetric modes equivalent to Lamb waves are used L(0,m)(m=1,2,3...) where L(0,1) A0 and L(0,2)S0
Fig 2: Guided Wave propagation in a cylindrical structure
Fig 4: Group Velocity Dispersion Curves
While standard ultrasonic waves propagate through the thickness of a cylinder walls, Lamb waves propagate along the walls and are introduced in it via two angled probes (or EMATs probes) making a global interrogation of the line joining the two probes. With Lamb wave propagation, the vibration patterns through the thickness of the cylinder in the axial and circumferential directions are quite different for different Lamb modes.
They may also change for different frequencies. Thus, to demonstrate Lamb wave propagation in cylindrical structures, an initial experimental tests were carried out by measuring velocities in different axial and circumferential directions of propagation (parallel, perpendicular to simulated defects). Using the fundamental Lamb modes A0 and S0, detection and interaction of Lamb waves modes with defects in cylindrical structures were also investigated. However, it was necessary to predict theoretical dispersion curves in order to verify the velocity of the pre-selected modes, to excite and choose at which point(s) on the dispersion curves these modes are to be excited. It was essential also to analyze wave structures to determine the most suitable mode for a certain application or defects .
In the following section results show a strong detectability of defects in the straight cylindrical section as well as the end sections; a reproducible response to different cylinders inspected with different transducer types and high sensitivity to the chosen wave modes. From these results, the proposed ultrasonic guided waves method can be considered feasible for efficient detection and location of variety of defects in cylindrical geometry.
Fig 5: a) S0 transmission mode b) A1transmission mode
Fig 6: Pulse-echo response for cylinder a) without defect b) with defect
Fig 7: EMAT pulse-echo response for cylinder with crack-like defect
Fig 8: PE response for a steel cylinder a) with defect b) without defect c) two edge without defect
Fig 9: Circumferential pulse-echo response
In the following, tests were carried out on 1000x230x7.3 mm steel cylindrical geometry. Figure 5 shows the time response in terms of a sinusoidal waveform (RF radio frequency signal). These waveforms are measured using a fixed angle probe for the S0 and A1 modes in the pitch-catch arrangement. From the time response measurement (Figure 5a) and comparing the theoretical with the calculated group velocities for the selected mode, it was confirmed that at this frequency-thickness combination only the S0 mode is propagating along the tested area in the cylinder.
Figure 5b illustrates the RF signal of A1 mode. The experimentally calculated group velocity of this mode was equally equivalent to the theoretical group velocity obtained from the dispersion curves. The parameters used to excite these modes were used in most of the performed tests. They were selected from the calculated dispersion curves for given material properties (steel in our case) and optimized by the analyses of their wave structures. However, at this point we demonstrate that the selected modes can be excited and detected at considerable distances, results of the detected defect are presented in the following figures. It should be noted that for comparison reasons some of the following experiments were performed using EMATs probes where mode excitation is exercised with perturbation directly inside the steel material of the cylinder.
Figure 6 shows the time response obtained from the S0 wave mode using the pulse-echo arrangement. We observe significant differences between the inspected area on the steel cylinder without defect (Figure 6a) and the inspected one with a 25x1.5x1.5 mm crack (Figure 6b). The time response shown in Figure 6a for a non-defected area of the cylinder shows only the S0 mode reflected approximately at 1200µs (micro seconds) from the end of the specimen. Waves arriving before the S0 wave mode represent the excitation (main bang) and the reverberation produced in the interface between the transducer element and the Plexiglas wedge. In Figure 6b, an additional wave packet corresponds to the reflection of the S0 mode from the crack is observed. This reflection from the defect was confirmed by comparing the actual and calculated distance between the transducer and the defect. This distance is calculated from the product of time-of-flight measurement and the group velocity of the excited S0 mode. An immediate conclusion from the time response for the pulse-echo arrangement test is that we can directly see any changes in the time response waveform. Existence of an additional packet of waveform is a strong evidence of the existence of defect. Therefore, a guided wave pulse-echo propagates over long distances of these cylinders and can detect flaws along its path.
Figure 7a shows the results obtained from inspection of the same cylinder with previously detected defect using an EMAT probe to excite S0 wave mode in pulse-echo arrangement . Slight differences in the position of inspection probe and interpretation of the results are occurred because EMATs transducer transmits in both directions. In Figure 6a, EMAT transducer is placed faced to the defect resulting in an extra reflection of S0 time response. The first wave packet is the main bang exercised to excite the EMAT, the second wave packet represents a reflection produced from defect while third one is a reflection from the near end of the cylinder. For a longer time response (Figure 7b) we would be able to see another two reflections one after another. One is produced from the defect due to the reflection of the previously reflected wave from the near edge and the second is produced from the reflection of the secondly transmitted pulse from EMATs at the far edge of the cylinder.
Tests were continued on a half section cylinder whose ends were cut, giving a straight section 1000 mm long and sliced in the middle so that internal defects are easily introduced. The results in the following figures illustrate the detectability as well as the long range inspection capability of the guided wave procedure. Figure 9 shows results from crack-like defect detected at 190 mm away of EMAT transducer. Clear reflection from the defect is obtained as well as the near edge of the cylinder. In the second figure the transducer was placed on the other side of the cylinder so no reflection from the defect is observed. However, Figure 8 shows the reflection of the second far end of the cylinder.
A number of tests were also carried out while the probe unit is scanned circumferentially to send its beam perpendicular to defects. In the following test, Figure 9 shows the time response obtained with the previously used S0 mode while using circumferentially pulse-echo approach to detect the same type of defects. Reflection of S0 mode from defect can now be clearly observed.
To develop a global inspection technique which locates and detects defects in cylindrical structures, we have demonstrated the ultrasonic guided wave method as an appropriate and promising inspection technique. We showed how the ultrasonic guided waves method can be used as a highly sensitive, fast and cost effective inspection technique.
Under different conditions, several experimental tests utilizing S0 and A1 modes with different frequency-thickness product were performed. Experiments showed that results can be repeated on different cylinders which is encouraging for adapting this technique to real time inspection. Yet our experimental conclusion on the sensitivity and applicability of using guided Lamb waves confirms that they can be used to inspect large areas in the axial and circumferential plane. Various modes can be induced and selected appropriately to optimize their propagation as a function of their selected characteristics. Sensitivity can be demonstrated from ability to detect, locate and validate structural damage in different type of cylindrical structures, also by demonstrating how ultrasonic guided wave method assesses different types of defects like fine surface-breaking flaws, crack-like defects, holes as well as wall thinning.
This work was supported jointly by Tektrend International Inc. and Natural Resources Canada under an agreement entitled "Evaluate the Performance of Guided Ultrasonic Waves in Detecting Cracks or Flaws in Steel Cylinders" funded by the Efficiency and Alternative Energy Program Agreement.