SBIR Phase I Final Report

Transcription

1 SBIR Phase I Final Report Ultrasonic Phased Array Technique for Accurate Flaw Sizing in Dissimilar Metal Welds Interwav Report: 15 Phase I Contract DE-FG2-4ER83984 March 11, 5 Interwav, Inc. 39 East Main Street Bedford, Virginia SBIR/STTR RIGHTS NOTICE These SBIR/STTR data are furnished with SBIR/STTR rights under Grant No._ DE- FG2-4ER83984_ (and subcontract if appropriate). For a period of 4 years after acceptance of all items to be delivered under this grant, the Government agrees to use these data for Government purposes only, and they shall not be disclosed outside the Government (including disclosure for procurement purposes) during such period without permission of the grantee, except that, subject to the foregoing use and disclosure prohibitions, such data may be disclosed for use by support contractors. After the aforesaid 4-year period the Government has a royalty-free license to use, and to authorize others to use on its behalf, these data for Government purposes, but is relieved of all disclosure prohibitions and assumes no liability for unauthorized use of these data by third parties. This Notice shall be affixed to any reproductions of these data in whole or in part.

2 1. Abstract The lack of a manual crack sizing technique compatible for use on dissimilar welds is a significant problem for the nuclear industry. The crystallographic structure associated with these welds results in beam redirection that produce significant errors when through-wall sizing is attempted. This report describes the results of a study where the through-wall sizing accuracy is compared for two different Time Gated Diffraction (TGD) techniques using both machined reflectors and dissimilar metal weld samples. The samples used for this study represented various weld material, cross-sectional configurations, pipe diameters and thickness commonly found in nuclear applications. The ultrasonic technique utilized portable phased array technology and transducer configurations designed specifically for the detection of tip diffracted signals. Results measured during testing were entered into a model that calculated the severity of beam redirection and provides and adjusted tip depth measurement. Test results indicated that crack height accuracies of approximately.41 RMS are possible using this technique. These results suggest that the use of a TGD technique with beam simulation as specified in this report can result in the successful completion of an ASME Supplement 1 qualification for crack sizing in dissimilar metal welds. Feasibility is claimed based on the fact that ASME qualification Supplement 1 accuracy requirements for qualification are over three times that measured for this technique (must be less than.125 ) 2. Background Dissimilar metal welds are used throughout nuclear power plants wherever a ferritic component is joined to an austenitic component. For example, the reactor vessels of commercial nuclear power facilities are fabricated from thick-sectioned carbon steel materials and cladded for corrosion prevention. In contrast, most piping used to carry coolant water and steam to and from the reactor vessel is fabricated from a stainless steel alloy. Where these two components attach, is a 2

3 weldment that secures two materials that have different material properties and thus, are considered to be dissimilar. Differences in mechanical properties such as thermal expansion coefficients, Young s modulus, metallurgical grain size and orientation, hardness, resistance to fatigue failure, etc., make these welds highly susceptible to crack initiation caused by high residual stresses, inter-granular stress corrosion cracking, or other mechanisms. Dissimilar metal welds have long been identified as a difficult component to inspect using conventional ultrasonic techniques (the only applicable method for single surface inspection) due primarily to the anisotropic nature of the weld. The actual inspectability of these welds has not been fully realized until recently when the NRC (Nuclear Regulatory Commission) adopted Appendix VIII of Section XI of the ASME code as a requirement for in-service inspection of nuclear facilities. As a result, all vendors that perform inspections on specific safety critical components after November 22, 2, must have successfully passed a series of blind tests using their respective techniques on samples containing real flaws. This performance based criteria is designed to improve flaw detection and sizing capabilities of vendors while preventing inferior techniques from being deployed to sites. On January 21, 3, the NRC issued a Regulatory Issue Summary (RIS) 3-1 titled Examination of Dissimilar Metal Welds Supplement 1 to Appendix VIII of Section XI of the ASME Code. In this document it is stated that, The NEI (Nuclear Energy Institute) representatives indicated that licensees had not qualified any procedures or personnel to meet the requirements of Supplement 1 (Supplement 1 pertains to DM weld inspection from the OD surface). The NEI further projected that the earliest any qualification could be completed was the end of November or December 2. 3

4 Although some vendors have been able to successfully satisfy the flaw detection criteria of Appendix VIII Supplement 1, no vender to date (verified by EPRI on 1//5) has passed the flaw through-wall sizing requirements using manual ultrasonic examination methods. This has become a significant problem for the commercial power utilities as nuclear plants in the United States are commonly 3- years old. An increasing number of cracks have been found in dissimilar metal welds over the last 5-1 years in both Pressure Water Reactors and Boiling Water Reactors. The most recent of these was at the Hope Creek Nuclear Power Station in New Jersey on 11/4. There are cases where an axially oriented crack has propagated completely through the weld resulting in water leakage before being detected by visual inspection or through the use of leak detection sensors (Vermont Yankee Power Station 3). Although a less likely scenario, a circumferentially oriented crack in a DM weld has the potential to lead to catastrophic failure of the entire weld resulting in a potentially serious loss of coolant to the nuclear core. The general process that occurs when a crack is found starts with a crack height sizing exam. This step can take two days to two weeks depending on the availability of a automotive sizing crew. The results of this exam are used to 1) determine if the repair can be delayed one outage cycle based on fracture mechanics analysis using the height measurement, 2) determine whether the crack size appropriate for compressive remediation ( MSIP - a process where the pipe material is placed in compression through mechanical deformation), 3) determine the required thickness for an overlay repair (overlay is basically when weld metal is deposited on the OD of the pipe which results in a new pressure boundary), or 4) determine the need for total weld replacement. Currently if an utility discovers a flaw in a dissimilar metal weld, they are forced to gather sizing data using an automated approach where a motorized scanning apparatus is mounted to the pipe and ultrasonic data is collected and stored. This process is very costly since the number of qualified automated procedures and analysts are limited due to the difficulty of the qualification. If no automated resources are 4

5 available, or if access is limited on the pipe preventing the mounting of the automated scanner, the utility is forced to perform a costly repair based on worst case assumptions. This report describes an approach to manual flaw sizing that is significantly different from that commonly found in the industry to date. If successful, this technique could offer the utility an option of determining flaw through-wall height without the cost and time requirements associated with an automated inspection. In addition, the technique described in this report is considered a manual technique that can be used for space restrictive applications where scanning hardware is not an option. 3. Objective The objective of this project was to determine feasibility of using a Time Gated Diffraction (TGD) technique that utilized a beam redirection model for accurate flaw height measurements. Feasibility was determined by comparing the crack height measurement error obtained by measuring crack heights associated with implant flaws on dissimilar metal weld (DM) samples to that required by ASME code. 4. Technical Approach 4.1 Ultrasonic Sizing Methodology Inaccurate sizing using ultrasonic testing of flaws in dissimilar metal welds is a phenomenon typically caused by either the misidentification of the tip diffracted signal, or by beam redirection effects resulting from columnar grain structure in the weld and Inconel butter material. A technique that is to be successful at sizing in these materials must address these two 5

6 issues. The following is a discussion of how the Time Gated Diffraction technique performs this task. Failure to properly identify the correct tip diffracted signal is a common problem especially when inspecting a weld using a manual conventional ultrasonic instrument. The inspection of dissimilar metal welds from the OD is typically performed using single or dual element transducers operated in a pulse-echo configuration as illustrated in Figure 1. In a pulse-echo test, a crack is detected and sized using sound energy that returns along the same general path to the transducer from which it originated. When evaluating the response from an ideal ID surface connected crack, two signals are observed: the reflection from the corner created by the crack plane and the ID surface, and the tip diffracted signal. While the corner reflection is typically high amplitude, directional signal, the tip-diffracted signal is commonly very weak and is irradiated omnidirectionally from the crack tip. However, in a pulse-echo configuration, additional low amplitude signals are reflected back to the transducer from the multi-faceted crack surface originating from multiple points along the crack face. Knowing the angle of sound propagation, θ, and the difference in an arrival time of the corner and tip signals, the flaw height can be determined either mathematically or directly from an UT instrument that has been accurately calibrated. A problem can occur if the operator confuses a facet reflection for a tip diffracted signal. In such a case, the flaw is under sized. In order to minimize the possibility of misinterpreting a reflective signal for a diffracted signal, two transducer configurations were selected that are designed to be more sensitive to diffracted energy and less to reflective energy. Both configurations use separate receiver transducers positioned so that energy reflected off the crack face is directed away from the receiving transducer. However, since tip diffracted signals are emitted 6

7 omni-directionally, the receiver transducer positions are appropriate for the reception of this diffracted signal. Figure 2 illustrates the two transducer configurations tested in this investigation. The Delta and Forward Scatter configurations were chosen for this study in an effort to minimize errors associated with operator inexperience by presenting only those signals with diffracted characteristics. The second, and most common problem with sizing a flaw in a DM weld, is the beam redirection effects caused by the large grain size and columnar crystallographic orientations in the weld volume. Beam redirection creates a situation for the operator where the actual angle of propagation of the sound path to the flaw is unknowingly modified by the material as illustrated in Figure 3. Since trigonometric relationships are used for flaw depth measurements, a change in the angle of propagation can lead to significant flaw height measurement inaccuracies. The key to being able to successfully size a crack is to have a method by which the operator can detect measure and lastly, compensate for material induced change in the angle of propagation. The following is a description of a method that uses a phased array instrument and beam simulation model that is capable of performing this function. Fundamental to this method is the use of a portable ultrasonic phased array instrument. Phased array systems use transducers that have many small elements that are fired independently of each other. The firing sequence and relative delays are determined by focal laws, or calculated firing delay times, that are entered into the instrument. These calculated firing sequences determine the angle of propagation of the wave front as well as beam focusing characteristics. Phased array systems are unique in that one transducer can sweep through a range of angles (Sector Scan) without moving or physically modifying the transducer. The system can be 7

8 programmed so that the transmitter and receiver transducers sweep independently of one another and focus their beams along a defined plane. This constant offset focusing capability is another key aspect to this method. When testing using a constant offset focus, the instrument is programmed so that the transmitted beam and receiver focal zone are swept along a defined plane as shown in Figures 4 and 5. These figures illustrate the beam sweep pattern for both the Delta and Forward Scatter techniques. During an inspection, the transmit beam is propagated over a range of angles appropriate to provide coverage from the ID to the OD surface of the pipe. Simultaneously the receiver is adjusted so that its focal zone intersects the transmitted beam on the predetermined focal plane. It is important to note that both beam angle changes and changes in the focal distance of the two transducers are constantly modified in an effort to create a small vertical column of material that a crack must reside in order to be detected. For the Delta configuration, the detection column is directly beneath the receiving transducer. Similarly, the Forward Scatter technique has its detection column defined between the two transducers (for this technique the detection column can be defined at any location between the two transducers although all tests for this investigation were performed with the focal column located at the mid point). The ability to define a column of material from which signals must originate reduces the possibility that non-crack related reflectors are detected and mixed with crack related signals. More importantly, if a signal is detected and peaked by moving the transducers, the operator knows that the signal of interest must lie in the detection column of material. Although the depth of the reflector is not known, its lateral location relative to the two transducers is. The ability to laterally locate the flaw is key to the determination of tip depth. 8

9 Once a crack tip signal is located and peaked, the operator has knowledge of its lateral location. In order to determine depth, two additional signal characteristics are then measured; signal peak angle and signal absolute time-of-flight. Signal peak angle is a direct measurement taken off the phased array system Sector Scan display. The Sector Scan is a plot of signal time-of-flight verses transmit angle. An example of a geometry corrected Sector Scan is shown in Figure 6. The color displayed is a function of the amplitude of the signal. The signal absolute time-of-flight measurement is also provided by the phased array system. The absolute time-of-flight is the amount of time required for the sound wave to travel from the transmit transducer, to the flaw and on to the receiver transducer. Since the velocities values are known for both the weld material and transducer wedge material (these are the two materials that the sound must propagate through on its trip from the transmitter to the receiver) and the lateral position of the flaw is known as discussed above, there exists a discrete time-of-flight value for every angle of propagation. In fact, assuming no beam redirection effects, the time-offlight of the sound wave is very predictable and can be easily modeled. Two different time-of-flight models were produced, one for each transducer configuration. The computer models, like that shown in Figure 7, calculated time-of-flight values for a given range of propagation angles. The model utilized Snell s Law of refraction to determine the sound path angles in wedge material based on a given angle of propagation in the weld material. The model utilized sound velocity values measured for both the weld and wedge materials. Another feature in the model was time-of-flight corrections based on the input of varying degrees of beam redirection. Input of beam redirection would force the model to adjust its time-of-flight values based on a diverted sound path that did not obey Snell s law, but rather simulated effects similar to that produced by 9

10 material conditions. The model also displayed calculated flaw depth based on beam parameters entered. The model is used to provide the corrected flaw depth value once all parameters are enter correctly. Since the lateral location of the flaw is fixed, and the time-of-flight is a predictable entity for a given angle of propagation, the measured signal peak angle and time-of-flight must closely match that calculated by the model. If the measured values match that calculated by the model then the operator knows that no beam redirection is occurring. However, if model generated and measured values do not match, then the model is adjusted so that beam redirection is simulated. Redirection angles are entered until the time-of-flight value is equivalent to that calculated by the model. If this procedure is performed correctly, the model is capable of calculating the corrected flaw depth since the amount of beam redirection has been determined resulting in a more accurate sound path simulation. 4.2 Description of Ultrasonic Hardware The test setup consisted of specially designed phased array transducers and a portable phased array system. Descriptions of these components are discussed below Transducer Two separate but identical 1-D ultrasonic phased array transducers were procured for this investigation from General Electric Inc. One unit was used for transmitting and the second for receiving. The transducers operated at a frequency of 2.34 MHz and consisted of 16 elements per unit. The transducers were mounted to machined Lucite wedges with a wedge angle appropriate to generate a natural angle of propagation 1

11 that bisected the range of angles needed for a full sweep on the samples tested. The wedges were removable whereby allowing the same transducers to be used for both the Delta and Forward Scatter configurations. The transducers were mounted in a small test frame that allowed for quick and accurate separation adjustments. The frame allowed for the transducers to be secured at a fixed separation during testing Phased Array System The phased array system used for this investigation was the X-32 manufactured by Harfang Microtechniques shown in Figure 8. The X-32 is a phased array system designed for portability and field conditions. The X-32 is capable of supporting 32 separate channels operated in the pitch-catch configuration. The system offers a data storage and post collection analysis capabilities. This feature allowed for the storage of data once the signal of choice was peaked. Time Gated Diffraction plots, time-of-flight, and peak angles were then extracted from the data without requiring the transducers to be maintained on the sample. The PhaseFX focal law generation software by Harfang Microtechniques was also used for this project. This software was capable of generating the focal laws required for both the Delta and Forward Scatter transducer configurations. 4.3 Description of Samples A total of nine dissimilar metal weld samples were used for this investigation. The samples ranged in diameter from 4 to 24 inches and.55 to inch in thickness. The weld samples contained a combined 11

12 total of 24 implanted cracks randomly placed throughout the weld volume. All cracks were ID connected and were oriented in the circumferential direction. No axially oriented flaws were tested due to the financial limitation of this study and the expense associated with the fabrication of curved calibration blocks and contoured transducer wedges. A direct correlation between these results on circumferential flaws and projected performance on axial flaws can be expected. Figures 9-14 provide the weld cross-sections associated with each sample. It should be noted that these samples represent the majority of pipe and weld configurations found in the ASME Section XI, Appendix VIII PDI qualification exam at the Electric Power Research Center. Also many samples contained Inconel buttering on one or more sides of the weld making these samples the most difficult to inspect ultrasonically. 5. Test Method 5.1 Transducer Adjustment Prior to the performing any calibrations, the transducer configuration was adjusted. Transducer spacing was determined primarily by the weld width and component thickness dimensions. When using the Delta configuration, the transducer separation was adjusted so that the receiver transducer could be positioned of the entire weld volume with the transmitter always position over base material. When using the Forward Scatter configuration, the center position between the two transducers should be capable of scanning over the entire weld volume without either transducer leaving base metal. Unlike the Delta technique, the Forward Scatter technique can be used with the weld crown present as long as the transducer spacing is large enough. However, it should be noted that depth accuracy decreases with increase in transducer spacing for both techniques due to the reduced range of angles to span the part thickness. 12

13 The transducer spacing was carefully and accurately measured and entered into both the phased array system and the computer time-of-flight model System/Model Calibration Prior to the collection of data, the ultrasonic system and model was calibrated. The focal laws, or individual transducer element pulse delays, were calculated using measured dimensions associated with the transducer setup and material velocity values. Since the focal laws are theoretical calculations, confirmation must be made to access and finetune their accuracy. This was performed using a special calibration block that provides the operator with the capability to directly measure the beam angle produced by the transducer for a specific focal law. Once confidence in the accuracy of the phased array system was established, the time-of-flight model must was also calibrated. This was performed by peaking the response from a side drilled hole in a stainless steel block. It is important to note that the calibration block was not fabricated from cast material since this would invalidate the calibration by introducing beam redirection effects. The peak angle and time-of-flight of this reflector was compared to that calculated by the model for a reflector of equivalent depth. Adjustments to global phased array parameters were made to force the phased array system to accurately match the beam angle calculated by the model. Lastly, the time-of-flight measured by the phased array system was compared to that calculated by the model. The arrival time measured by the phased array system was typically greater than that calculated by the model since the phased array system required additional time for focal law processing. This additional time delay constant was added to the time-of-flight values calculated by the model as a global parameter. Once calibrated, several reflectors were peaked and 13

14 corresponding time-of-flight and angle values are measured and compared to the model. If performed correctly, the model and measured parameters would correlate closely. 5.3 Technique Accuracy Tests The first tests performed were designed to determine the accuracy of the Time Gated Diffraction technique model using both the Delta and Forward Scatter transducer configurations when applied to a material where beam redirection was not present. The test was performed using a special block that contained three 1/32 diameter side drilled holes (SDH) located at various depths as shown in Figure15. Due to offset location of these holes in the block, testing from opposing sides provided a total of reflectors of varying depths. The depth of the holes ranged from.516 to inches. The test block was fabricated from forged stainless steel, thus making it immune to beam redirection effects. The system calibration process as described in Section 5.2 was performed on a hole that was in the middle of the depth range. Following calibration, the remaining whole depths were measured and compared to their actual positions. This test was designed to provide the best case accuracy expectations for both transducer configurations. It also served as validation for the accuracy of the time-of-flight values calculated by the two models. The results of this these tests are described in Section 6.1 of this report. 5.4 Dissimilar Metal Weld Sample Tests A second series of tests involved the use of dissimilar metal weld samples that contained implanted flaws. These samples as described in Section 4.3, were identical in both material and fabrication methods to that used in ASME Appendix VIII PDI qualifications. An attempt was made to test as many open samples that could be located in an effort to demonstrate the 14

15 accuracy of the Time Gated Diffraction method using both the Delta and Forward Scatter transducer configurations. Each sample contained cracks that were implanted on both sides of the weld centerline. All samples were provided by Flawtech Inc, a company that specializes to the fabrication of ultrasonic test samples. A defect map accompanied each sample showing the general location of each flaw. The transducers were coupled to the part using ultrasonic silicon gel couplant and scanned in the area of interest. Since only circumferentially oriented flaws were of interest, the primary scan motion was in the circumferential direction, parallel to the weld. Crack tip signals were most noticeable when scanning in this direction since the flaw related signal would appear and disappear in the Sector Scan display when traversing across the defect. Using the Sector Scan image, the transducers were moved parallel to the weld until the deepest most extension of the flaw tip signal was detected. Then, the transducers were moved perpendicular to the weld to maximize the signal amplitude. This adjustment was necessary to assure that the flaw location corresponded with the column of material were the beams converged. The X-32 instrument storage feature was activated which stored the Sector Scan image and all associated A-scans. This permitted the operator to remove the transducer assembly from the pipe and focus on data analysis. Data analysis consisted of three primary functions. First, the operator measured the beam angle at which the maximum flaw signal amplitude was detected. This was performed directly from the Sector Scan image. Secondly, the time-of-flight of the tip signal was measured. This was performed by moving the A-scan extractor cursor so that the A-scan associated with the peak beam was displayed. The signal time-of-flight was measured from the A-scan display using the signal peak location. Thirdly, these two measurements were entered into the time-of-flight 15

16 model which calculated the amount of beam redirection that was present and the corrected depth of the indication. The model also calculated the corrected time-of-flight values based on the amount of beam redirection present for all beam angles. These delay values were then used to generate a Time Gated Diffraction plot for each defect. The calculated flaw depth was compared to the true flaw depth and an error was calculated. The results of these tests are presented in Section 6.3 of this report. 6. Results and Discussion 6.1 Technique Accuracy Tests for Delta Technique Figures 16 and 17 provide the Time Gated Diffraction plots and associated Sector Scans for each side drilled hole using the Delta configuration. Figure 18 is a table that summarizes the depth measurement and associated error for each side drilled hole evaluated using the Delta configuration. Figure 19 is a sample Sector Scan from the inch hole. The Sector Scan clearly indicates the signal diffracted by the small side drilled hole. Superimposed over this image is a box that represents the approximate focal zone of the beam sweep. As the depth of the flaw changes, the flaw signal will move vertically in this zone as can be seen in Figures Shown beside each Sector Scan image in Figures are Time Gated Diffraction plots of the depth verses amplitude using the time-of-flight values provided by the computer model. The peak shown on each plot relates directly to the depth of the indication. These measurements were performed using values provided by the model without the use of beam redirection correction. 16

17 The results of this test indicate that when no beam redirection was occurring in the material, the technique produced flaw height measurements with and accuracy of.22 inch RMS and Standard Deviation of.14 inch. This measurement was based on a sample set of five reflectors. The technique tends to be less accurate for flaws that are close to the inspection surface. The tips of such deep flaws require higher beam angles. The ability of the phased array system to accurately produce specific beam angles decreases as the beam is swept farther away from the natural angle of the wedge. Also, small changes in the beam angle results in larger changes in the depth at high angles. This results in a reduction in accuracy since the angle increment is constant throughout the beam sweep. These results suggest that the accuracy of the Delta configuration using the time-of-flight model will not exceed.22 inch regardless of material. It was considered highly probable that the overall accuracy of this technique would decrease as the model is subjected to dissimilar materials where beam redirection effects will require ultrasonic beam simulation. Results related to these tests are discussed in Section ## of the report. 6.2 Technique Accuracy Tests for Forward Scatter Configuration Figures -21 provides the Time Gated Diffraction plots and associated Sector Scans for each side drilled hole using the Forward Scatter configuration. Figure 22 is a table that summarizes the depth measurement and associated error for each side drilled hole evaluated. Figure 23 is a sample Sector Scan from the inch hole. The Sector Scan clearly indicates the signal diffracted by the small side drilled hole. 17

18 Superimposed over this image is a box that represents the approximate focal zone of the beam sweep. As the depth of the flaw changes, the flaw signal will move vertically in this zone as can be seen in Figures -21. Although this scan appears very similar to those produced using the Delta configuration, there were two distinct differences. The first was the presence of a surface related signal at the top of the Sector Scan (see Figure 23). This signal is the direct result of a surface following wave that travels from the transmitter to the receiver. The presence of this signal can assure the operator that a crack does not penetrate completely through the wall to the OD surface. If such a condition does exist, the presence of this signal will disappear. A second signal not shown on Figure 23 (calibration block thickness is too large for the Sector Scan settings displayed), is a back wall reflection which can be a strong reflection if the ID and OD surface are close to parallel. The back-wall signal can assist the operator in locating an ID connected crack since this signal will decrease in amplitude or disappear completely depending on the crack depth. It should be noted that the ID reflection should not be used as a definitive tool in identifying crack location or depth since weld root geometries can also cause irregularities in the ID surface reflection amplitude. Shown beside each Sector Scan image in Figures -21 are Time Gated Diffraction plots of the depth verses amplitude using the time-of-flight values provided by the computer model. The peak shown on each plot relates directly to the depth of the indication. These measurements were performed using values provided by the model without the use of beam redirection correction. The results of this test indicate that when no beam redirection is occurring in the material, the technique can produced flaw height measurements 18

19 with an accuracy of.17 inch RMS and Standard Deviation of.16 inch. This measurement was based on a sample set of five reflectors. These results suggest that the Forward Scatter technique is approximately 23% more accurate than the Delta configuration on average. Like the Delta configuration, the technique tends to be less accurate for flaws that are close to the inspection surface. The tips of such deep flaws require higher beam angles. The ability of the phased array system to accurately produce specific beam angles decreases as the beam is swept farther away from the natural angle of the wedge. Also, small changes in the beam angle results in larger changes in the depth at high angles. This results in a reduction in accuracy since the angle increment is constant throughout the beam sweep. These results suggest that the average accuracy of the Forward Scatter configuration using the time-of-flight model will not exceed.17 inch regardless of material. The overall accuracy of this technique was thought to decrease when subjected to the inspection of dissimilar materials where beam redirection effects will require ultrasonic beam simulation. These test results are discussed in Section 6.3 of the report. 6.3 Dissimilar Metal Weld Sample Tests for the Delta and Forward Scatter Configurations Figures 22- provide the Time Gated Diffraction plots and associated Sector Scans for each crack analyzed using both the Delta and Forward Scatter configurations for all of the dissimilar metal weld samples. Figures 41 and 42 summarize the sizing results for each flaw evaluated. It should be noted that the Delta configuration was not applied to Sample due to irregular geometrical condition in the location where the weld crown was located. Also some samples contained axial defects which were not included in this study. 19

20 Analysis of the sizing data revealed very positive information. Both the Delta and Forward Scatter configurations produced acceptable sizing results. The Delta configuration demonstrated an RMS error value of.5 inch with a standard deviation of.29 inch. These results are based on a sample set of 22 flaws. The Forward Scatter configuration demonstrated an RMS error of.41 inch with a standard deviation of.28 inch. These results are based on a sample set of 24 flaws. The Forward Scatter configuration out performed the Delta with an RMS error that was approximately.9 inch better, or an improvement of 22%. This result is very consistent with that measured on machined reflectors as discussed in Sections 6.1 & 6.2 of this report. If similar results were produced during a PDI qualification, the Time Gated Diffraction technique using either the Delta or Forward Scatter configurations would have passed by not exceeding the.125 inch RMS error as required by ASME code. Another parameter measured was the amount of beam angle correction calculated for each flaw. Results for the Delta configuration indicated an average angular correction of 2.44 degrees RMS with a Standard Deviation of 2.34 degrees. The Forward Scatter configuration required an average of 7.4 degrees of redirection correction with a standard deviation of 4.7 degrees. The disparity between the two different configurations can be explained by the method by which data was collected. During testing, both techniques were applied with the transmitter positioned on both sides of the weld. Comparison of the signal received from either of the two sides was performed and the side exhibiting the best signal quality was used. This was a significant difference for the Delta technique since the receiver was positioned above the weld and only the transmitted signal had to penetrate through the weld/base material interface. In contrast, the Forward Scatter approach required both the transmitted and received

21 signals to pass through the weld/base material interface in two separate locations, regardless of position. It is hypothesized that this effect made a significant difference since the Delta could have been positioned used where penetration of Inconel butter material was not required. This result suggests that the Inconel buttering and/or the carbon-to-stainless interface region is the primary source of beam redirection in a dissimilar weld. More importantly, the Time Gated Diffraction model was capable of compensating for the increased beam redirection, thus allowing the Forward Scatter technique to exhibit higher accuracy. Another observation was related to signal quality. The Forward Scatter configuration consistently produced signals that were more isolated and easier to identify. Generally the signal-to-noise ratio of the flaw signals was very good for both techniques. Flaw signal identification typically was made within one minute from the initiation of an inspection. However, when using the Delta configuration, signals not related to the flaw were observed on some samples when the receiver was placed over top of the Inconel butter or carbon/weld interface region. These signals could be observed along the weld and seemed to be emitted directly from the interface. Flaws that were located along the interface boundary could be more difficult to interpret due to these signals. The Forward Scatter technique was much less susceptible to these signals especially if the transducer spacing was adjusted so that both the transmitter and receiver remained on base material while inspecting the weld volume. The reduction of interface signals is a goal of future design improvements. One significant observation related to the use of this method, is the speed at which an inspection can be performed. The X-32 phased array system s capability to store data was an important factor in the practicality of this technique. Based on the samples tested it is estimated that the time required to acquire data under field conditions should not exceed 21

22 approximately 15 minutes. Flaw depth calculation would involve additional time but this would be out of a radiological area, thus minimizing exposure. The Forward Scatter configuration had a distinct advantage over the Delta since both transducers could be placed on pipe material with a consistent surface contour. In contrast, the receiver transducer for the Delta required positioning on an irregular surface created by manual removal of weld crown. The surface on one sample was too irregular to seat the transducer without significant tilt. This sample was not inspected using this technique. Even though the Forward Scatter configuration is compatible for use with the weld crown in place when inspecting circumferential flaws, it is anticipated that axial oriented defects will require weld crown removal. 7. Conclusions and Recommendations It can be concluded that the use of the Time Gated Diffraction technique for accurate sizing of cracks located in and around dissimilar metal welds is feasible. This conclusion is based on the measurements summarized below: An RMS accuracy of.22 was achieved when measuring reflector depth in stainless material where no beam direction was present using the Delta configuration. An RMS accuracy of.17 was achieved when measuring reflector depth in stainless material where no beam direction was present using the Forward Scatter configuration. An RMS accuracy of.5 was achieved when measuring crack height on implanted flaws in dissimilar metal welds when using the Delta transducer configuration. 22

23 An RMS accuracy of.41 was achieved when measuring crack height on implanted flaws in dissimilar metal welds when using the Forward Scatter transducer configuration. Beam redirection angles were measured to be to as high as 14.5 degrees using the Delta configuration and 17.7 degrees with the Forward Scatter configuration. ASME Section II, Appendix VIII requirements for a Supplement 1 test requires that the procedure be demonstrated to be capable of sizing cracks with an error no greater than.125. The RMS error associated with the Time Gated Diffraction technique is approximately 1/3 of that required for PDI qualification. It is assumed that this buffer will greatly increase the probability for successfully qualifying this technique using blind samples. Also, it is recommended that both the Delta and Forward Scatter techniques be used during blind qualification tests. It was clear from the data that both techniques provided useful information and could be used as confirmation to one another. However, when possible, the Forward Scatter transducer configuration should be used for final depth measurements. One observation that should be mentioned is the amount of beam correction required for some samples. When considering that some flaws required up to 17.7 of correction to be sized accurately, it is clear that without beam redirection simulation and correction, the ability to accurately size cracks in dissimilar metals is not probable. Although the results were very positive, there is room for significant improvements. For example, transducer design modifications can be made to improve tip diffracted signal response. The current transducer design works very well on materials with thickness of 1-inch and greater. A larger transducer size should help with improving signal-to-noise characteristics on thicker materials by 23

24 extending the near field thus resulting in better beam focusing. Also, a slightly lower frequency transducer is thought to decrease metallurgical related signals associated with the weld interface by increasing the sound wavelength without significantly decreasing signal resolution. Other improvements related to transducer fixturing and improved interface additions to the beam modeling software will significantly improve the ease of use of the technique. Aside from the relatively minor improvements mentioned above, it strongly recommended that this technique be subjected to the PDI qualification process as soon as possible in an effort to alleviate industry needs for such capabilities. 24

25 Tip Diffracted Signal Inconel Carbon Steel θ Stainless Steel Corner Reflection Figure 1: Pulse-echo technique using tip-diffracted signal and corner reflection for determining flaw through-wall extension. a) b) Figure 2: a) Delta transducer configuration; b) Forward Scatter transducer configuration. 25

26 Transducer Location of flaw if beam redirection was not considered. Beam is redirected or bent by material columnar grain structure. Figure 3: Columnar grain structure in weld material causes the ultrasonic beam to change direction resulting in an inaccurate crack height measurement. 26

27 Focal Zone Figure 4: Sweep pattern of the Delta configuration where transmit and receiver are focused along a column of material that extends from the ID to the OD surface. 27

28 Focal Zone Figure 5: Sweep pattern of the Forward Scatter configuration where transmit and receiver are focused along a column of material that extends from the ID to the OD surface. 28

29 Time-of-flight Beam Angle Diffracted Signal Time-of-flight A-scan Display Figure 6: Example of a Sector Scan showing beam angle, time-of-flight and A-scan display. 29

30 Figure 7: Model input and output parameters for pitch-catch transducer configuration on flat plate. 3

31 Figure 8: Harfang X32 and Transducers Used for Data Collection 31

32 Figure 9: Samples and weld configuration. 32

33 Figure 1: Sample weld configuration 33

34 Figure 11: Sample weld configuration. 34

35 Figure 12: Sample weld configuration. 35

36 Figure 13: Samples 1--3 and 1--4 weld configurations. 36

37 Figure 14: Samples and weld configurations. 37

38 Figure 15: Machined stainless steel reference standard containing several 1/32 inch side drilled holes used for calibration and model accuracy tests. 38

39 Figure 16: Time Gated Diffraction Plots and Corresponding Sector Scans for Side Drilled Holes at Indicated s Using Delta Configuration 39

40 plitude Am Figure 17: Time Gated Diffraction Plots and Corresponding Sector Scans for Side Drilled Holes at Indicated s Using Delta Configuration

41 Error on Side Drilled Holes Using DELTA Configuration ID Flaw Measured Error Comments N/A Average Error =.22 Used for calibration Figure 18: Measured depths of side drilled holes using Delta transdcuer configuration. Tip diffracted signal will always be first signal detected. Defect signal will display in the vertical focal zone unless beam redirection is present. Figure 19: Sector scan obtained using the Delta configuration. 41

42 Figure : Time Gated Diffraction Plots and Corresponding Sector Scans for Side Drilled Holes at Indicated s Using Forward Scatter Configuration 42

43 Figure 21: Time Gated Diffraction Plots and Corresponding Sector Scans for Side Drilled Holes at Indicated s Using Forward Scatter Configuration 43

44 Error on Side Drilled Holes Using FORWARD SCATTER Configuration ID Flaw Measured Error Comments N/A Average Error =.17 Used for calibration Figure 22: Measured depths of side drilled holes using Forward Scatter transducer configuration. Surface wave is present when crack does not reach inspection surface. Defect signal inside focal zone Figure 23: Sector scan obtained using Forward Scatter technique. 44

45 7 5 Flaw Flaw Figure 24: Time Gated Diffraction Plots for Sample Using Delta Configuration 45

46 8 7 Flaw Flaw Figure 25: Time Gated Diffraction Plots for Sample Using Forward Scatter Configuration 46

47 8 7 Flaw Flaw Figure 26: Time Gated Diffraction Plots and Corresponding Sector Scans for Sample Using Deltar Configuration 47

48 1 1 8 Flaw Flaw Figure 27: Time Gated Diffraction Plots for Sample Using Forward Scatter Configuration 48

49 9 8 Flaw Flaw Figure 28: Time Gated Diffraction Plots and Corresponding Sector Scans for Sample Using Delta Configuration 49

50 8 7 Flaw Flaw Figure 29: Time Gated Diffraction Plots and Corresponding Sector Scans for Sample Using Forward Scatter Configuration 5

51 7 Flaw Flaw Figure 3: Time Gated Diffraction Plots and Corresponding Sector Scans for Sample Using Forward Scatter Configuration 51

52 7 Flaw Flaw Figure 31: Time Gated Diffraction Plots and Corresponding Sector Scans for Sample Using Delta Configuration 52

53 9 8 Flaw Flaw Figure 32: Time Gated Diffraction Plots and Corresponding Sector Scans for Sample Using Forward Scatter Configuration 53

54 8 7 Flaw Flaw Flaw Figure 33: Time Gated Diffraction Plots and Corresponding Sector Scans for Sample 1--3 Using Delta Configuration 54

55 1 1 Flaw Flaw Flaw Figure 34: Time Gated Diffraction Plots and Corresponding Sector Scans for Sample 1--3 Using Forward Scatter Configuration. 55

56 8 7 Flaw Flaw Flaw Figure 35: Time Gated Diffraction Plots and Corresponding Sector Scans for Sample 1--4 Using Delta Configuration 56

57 1 1 Top Tip Bottom Tip Flaw Flaw Flaw Figure 36: Time Gated Diffraction Plots and Corresponding Sector Scans for Sample 1--4 Using Forward Scatter Configuration 57

58 7 Flaw Flaw Flaw Figure 37 a): Time Gated Diffraction Plots and Corresponding Sector Scans for Sample Using Delta Configuration 58

59 7 5 Flaw Figure 37 b): Time Gated Diffraction Plots and Corresponding Sector Scans for Sample Using Delta Configuration 59

60 1 1 Flaw Flaw Flaw Figure 38 a): Time Gated Diffraction Plots and Corresponding Sector Scans for Sample Using Forward Scatter Configuration

61 9 8 7 Flaw Figure 38 b): Time Gated Diffraction Plots and Corresponding Sector Scans for Sample Using Forward Scatter Configuration 61

62 8 7 Flaw Flaw Flaw Figure 39 a): Time Gated Diffraction Plots and Corresponding Sector Scans for Sample Using Delta Configuration 62

63 8 7 Flaw Figure 39 b): Time Gated Diffraction Plots and Corresponding Sector Scans for Sample Using Delta Configuration 63