Sensors and Instrumentation and Nondestructive Evaluation
Safety Related Applications
Nuclear Regulatory Commission
International Steam Generator Tube Integrity Program
Key objectives of the International Steam Generator Tube Integrity Program (ISG-TIP) at Argonne is
to evaluate advanced NDE and
signal analysis techniques for reliable in-service inspection (ISI) of original and repaired steam
generator tubes and to assess improved correlations between eddy current results and flaw morphology,
leak rates, and failure pressure. Degradation-specific management requires detailed knowledge about
the specific nature and severity of flaws. Improved techniques (eddy current and others) are needed
for more reliable inspection and interpretation of flaws. The reliability and accuracy of the techniques
need to be quantified so that plugging criteria and the consequences of degraded tubes remaining in
service after inspections can be evaluated. Further, the robustness of the voltage and other eddy-current
parameters and techniques needs to be evaluated with respect to their range of applicability.
For more information: Steam Generator Tube Integrity Program
A major outcome of nuclear reactor regulatory activity over the past 10 years has been the development and implementation of two key concepts, condition monitoring and operational assessment. Condition monitoring is an assessment of the current state of the steam generator (SG) relative to the performance criteria for structural integrity. Operational assessment involves an attempt to assess the state of the generator relative to the structural-integrity performance criteria at the end of the next inspection cycle. Predictions of the operational assessment from the previous cycle can be compared with the condition monitoring to verify the adequacy of the methods and data used to perform the operational assessment.
A key factor in establishing the reliability of operational assessment and condition monitoring is the NDE techniques used to determine the flaw distribution in terms of detection and characterization of flaws and the capability to assess their impacts on the structural and leakage integrity of SG tubes. An NDE round-robin exercise has been used to independently assess SG inspection reliability. This exercise employed the steam generator mock-up at Argonne. The purpose was to assess the current state of ISI reliability for SG tubing, determine the probability of detection as a function of flaw size or severity, and assess the capability for flaw sizing.
Eleven teams participated in analyzing bobbin and rotating probe data from the mock-up that were collected by qualified industry personnel. The mock-up tube bundle contains hundreds of cracks and simulations of artifacts such as corrosion deposits, support structures, and tube geometry variations that, in general, make the detection and characterization of cracks more difficult. An expert NDE Task Group from ISI vendors, utilities, EPRI, Argonne, and the NRC has reviewed the eddy current signals from laboratory-grown cracks used in the mock-up to ensure that they provide a realistic simulation of those obtained in the field. The number of tubes inspected and the number of teams participating in the round-robin are expected to provide better probability-of-detection data and characterization accuracy than are currently available from industry performance demonstration programs.
The mock-up tube bundle consists of 400 Alloy 600 tubes made up of nine test sections, each 0.3 m (1 ft) long. The test sections are arranged in nine levels, each having 400 tube sections. The lowest level simulates the tube sheet, while three other levels simulate tube support plate intersections. The remaining five levels are free-span regions. Bobbin coil data were collected on all 3600 tube sections of the mock-up by using magnetically biased (“mag-biased”) probes. A mag-biased, rotating, three-coil probe was used to collect data from all 400 tube-sheet and special-interest test sections. Eddy current data were collected by a qualified industry team and stored on optical disks. The round-robin teams later analyzed the data with an Argonne proctor present to monitor the analysis process. The intent was to make the analysis as close a simulation of an actual inspection as possible. The procedures and training sets were developed in cooperation with the NDE Task Group so that the inspection protocols and training would mimic those in current practice.
The reference state for each flaw in the mock-up, i.e., crack geometry and size, was established by calculations using a multiparameter algorithm developed at Argonne for analyzing eddy current data (see next figure). Results were analyzed for all eleven round-robin teams, including the team-to-team variation in the probability of detection, along with the population average. The detection results for the 11 teams were used to develop probability-of-detection curves as a function of maximum depth and the parameter mp, a stress multiplier that relates the stress in the ligament ahead of the crack to the stress in an unflawed tube under the same loading. Because mp incorporates the effect of both crack depth and length, it better characterizes the effect of a flaw on the structural and leakage integrity of a tube than do traditional indicators, such as maximum depth. The probability-of-detection curves were represented as linear logistic curves, and the curve parameters were determined by the method of maximum likelihood. The statistical uncertainties inherent in sampling from distributions and the uncertainties due to errors in the estimates of maximum depth and mp were determined. In addition, the 95% one-sided confidence limits, which include errors in maximum depth estimates, were determined along with the probability-of-detection curves.
eddy current map of a crack in a roll transition of a tube simulating the tube sheet level of
a steam generator. The image shows the eddy current signal amplitude as a function of position.
The lower image but not the top shows that the crack is circumferential in orientation.
Click on photo to view a larger image.
Eddy current noise varies from plant to plant and can interfere with the quality of the data used to assess the integrity of the steam generator tubing. How much EC noise can be tolerated before data quality is affected and detection capability degraded is a key current issue and part of the Argonne ISG-TIP effort. Possible actions for low signal to noise of data include changing technique, determining if flaws can be detected in the presence of noise, and adjusting probability of detection and sizing uncertainty. The current effort has shown that noise levels that may not significantly affect detection can have a profound effect on sizing results.
Computer-Aided Data Analysis
In conventional eddy-current data analysis of SG tubing, human analysts are trained to differentiate flaw-induced signals from artifacts by analysis of the behavior of signals, typically extracted from their impedance plane trajectory and signal amplitude, through application of a series of rules. Manual analysis of multiple-frequency eddy-current data is a tedious and challenging process. No qualified technique, manual or automated, currently exists that could provide reliable estimation of flaw size over a wide range of SG tubing damage. Conventional data analysis methods become rather subjective when dealing with complex forms of degradation such as stress corrosion cracking. Signal distortion by interference from internal/external artifacts in the vicinity of a flaw further complicates discrimination of flaw signals from noise. Analogous to the manual analysis of eddy-current inspection data, computer-aided data analysis algorithms can be developed based on analysis of characteristic behavior of eddy current signals as a function of frequency. Because all available information can be examined rather than a selected subset in the case of a manual analysis, such algorithms allow more effective identification of subtle forms of degradation. Computer-based algorithms that imitate some form of human decision-making process are generally labeled as “expert systems.”
Multiparameter algorithms for the computer-aided analysis of eddy-current NDE data have been developed and implemented in a PC-based software called MATLAB, a high-level scripting language that provides an efficient environment for data manipulation and computation, together with convenient graphical user interfaces and graphical displays of the results. The effectiveness of these multiparameter algorithms has been evaluated in studies on electrodischarge-machined notches, laboratory-grown flaws, and in-service flaws. The Argonne expert system for the sizing and imaging of flaws from pancake-coil data is currently being further improved. The algorithms are being modified for use with other eddy current probes, and the resulting flaw-sizing accuracy and ligament-sizing capability is being evaluated and compared with those for the existing algorithm using the pancake-coil probe. Advanced signal processing and filtering techniques are being further evaluated for use with eddy current data from bobbin coils and other rotating probes to improve flaw detection and sizing for ISI of steam generator tubes. In response to inquiries by the industry partners, efforts are also being made to further develop the analysis methods implemented at Argonne for field application.
Barrier Integrity Research Program
The objective of this program is to reevaluate the basis for all existing leakage requirements and establish a technical basis for improved leakage requirements in nuclear power plants. The work being performed includes (1) a review of leakage operating experience and current leakage requirements, (2) evaluation of barrier integrity monitoring systems, (3) evaluation of leak rate studies, and (4) development of problem cases for leakage calculations. This program supports the implementation of the NRC Assessment of Barrier Integrity Action Plan and will provide a technical basis for improved requirements needed to maintain the integrity of the reactor coolant pressure boundary.
In one of the tasks, the sensitivity, reliability, response time, accuracy, and feasibility for leak detection systems (e.g., sump monitors, gas and particulate monitors) are being established and compared with technologies that can meet new requirements that may be necessary to detect the types of leakage that occurred recently in a nuclear-plant pressure vessel head. Leak detection may be improved by use of moisture sensitive tape, acoustic monitoring and advanced radioactive particulate, and gas detectors. The advantages and disadvantages of systems commercially available but not currently used in the U.S. are being determined. Two such systems are the FLUS humidity sensor and ALUS acoustic sensing, both from Seimens. While the FLUS system monitors the increase in local humidity caused by a leak, the ALUS system listens to changes in sound levels picked up by piezoelectric sensors attached to wave guides. The sensitivity depends on the background noise, which varies widely in a plant. Another system is called ARMS, which draws air through a filter to concentrate the particulates that are detected by a radiation monitor. A few ARM systems have been installed in the U.S.
For leak rate studies, we have developed correlations between crack size, crack tip opening displacement, and leak rate. The emphasis is on leak rates associated with stress corrosion cracking. The results are expected to show that the leak rates associated with through-wall cracks can be very low. The first set of computations was carried out for stainless steel piping with intergranular cracking in boiling water reactors and leak rates ranging from 0.1 to 100 gal/min. The output is length of circumferential through-wall cracks. The results obtained for crack lengths at various leak rates are compared against the critical crack length for the specified value of the bending moment. This work is being carried out as a collaboration with Engineering Mechanics Corp.
A barrier integrity database is being developed. The database contains three types of information: LWR leak events, leak detection systems, and crack monitoring systems. Sources being used to provide input to the database include Licensee Event Reports and NRC reports covering prior work. Literature searches are also being carried out to identify other relevant publications.
Last Modified: Fri, October 22, 2010 7:27 PM