PRIMARY LOOP ACOUSTIC EMISSION PROCEDURE: AN UPGRADED METHOD AND ITS CONSEQUENCES ON THE IN-SERVICE-INSPECTION Laurent Truchetti, Yann Forestier, Marc Beaumont EDF CEIDRE, EDF Nuclear Engineering Division; Saint-Denis, France Phone: +33 1436 97235, Fax: +33 1436 98293; e-mail: laurent.truchetti@edf.fr ABSTRACT In France, according to pressure vessel regulations, the PWR Primary loop is tested with a hydrotest at 207 bar pressure hold during every decennial outage. The primary loop should stand without leakage. During this test, many different welds (about 450) are checked in different Primary Loop areas such as the RPV and Pressuriser. Some of them are not easily reachable even for a visual examination. Therefore EDF uses an Acoustic Emission Procedure to ensure that the level of safety of the primary loop is satisfactory. This AE procedure has been defined to monitor the RPV head vessel welds, the RPV BMI and the Pressurizer heater sleeves. A large part of the initial development and qualification of this method relies on tests made during the first years of the French nuclear program during the 70s-90s. Nowadays due to stricter considerations on radiation during in-service inspections (ISI), the method has been upgraded in order to facilitate the Non-Destructive Examination (NDE). The objective of this paper is to present this new method and its challenging development. More specifically, a focus will be made on the evolution of the calibration signals moving from a wideband noise (historical inputs) to artificial waveforms, its consequences on the performances and the ISI itself. Keywords: Acoustic Emission, non destructive examination, Main Primary Loop, reactor pressure vessel, nuclear, pressurized water reactor, Hydraulic Pressure Test INTRODUCTION According to the pressure vessel regulation, the PWR Main Primary Loop is tested with a Hydaulic Pressure Test (HPT) during every decennial NPP outage. The main primary loop should stand without leakage. During this test, many different welds (about 450) are checked. A visual examination has to be performed. But some areas are not accessible because: either the areas is locked for radioprotection considerations as in the reactor pressure vessel (RPV) bottom head, or the density of welds is too high as in the RPV head vessel or the pressuriser (PZR) bottom head. For these areas, another NDE is put in place: a global non destructive examination which relies on the acoustic emission method. This paper will briefly present this method adapted to the RPV and PZR welds, its purpose and the correlated ISI with its constraints. As the French Ministerial Order of November 10 th, 1999 requires the qualification of all NDE carried out on the Primary and Secondary loops of a Nuclear Power Plant in France, this NDE was qualified in 2006. To add the EPR to the scope of the qualification and to take into account the ISI feedback, an updated qualification has been undertaken. This paper will focus both on the improvements made to the performance demonstration, and on the different gains obtained on the NDE and the ISI. THE ACOUSTIC EMISSION DURING HYDRAULIC PRESSURE TEST METHOD Acoustic Emission General Principles The principle of acoustic emission relies on the wave propagation that occurs inside a component under pressure if a leak is present. This is the result of different physical phenomena linked to the fluid flow 667
through the leak path, e.g.: cavitation, vapor bubble collision on the surface, or a disruptive flow, see figure 1. Depending on the depth of the material different wave form will travel across the material. In our case, considering that the sensors are located on the main bulk of the RPV or the PZR, only surface mode propagation is considered (Rayleigh waves). These waves will be detected with piezoelectric transducers; the RMS value of the transducer signal is registered. The RMS is integrated with an integration time constant of one second in order to avoid spikes. The RMS intensity is correlated to a reference flow rate. The Qualification Scope Figure 1: Principle of Acoustic Emission As said, the objective of the Acoustic Emission during the hydraulic pressure test examination, called Main Primary Loop (MPL) AE Testing in the following document, is to detect leaks at the highest pressure hold of the hydraulic test. This method is a global one meaning that leaks are looked for by areas of interest (zonal location) instead of precise planar location. The MPL AE Testing occurs at every ten-year outage. The following welds have to be inspected: - The RPV head welds - The Heater Sleeves to the Pressuriser Bottom Head - The RPV BMI welds to the RPV Bottom Head Figure 2: Acoustic Emission Channel Positioning 668
The different channels are evenly spaced, set up as close as possible to the welds under examination and mounted on: - the RPV Head : 6 transducers, - the RPV Bottom end cover : 4 transducers, - the PZR Bottom end cover : 4 transducers. The AE Testing has been qualified in 2006 in accordance with the RSEM code. The qualification type is conventional which is linked to the Defense in Depth (DiD) concept. No specific size of the leak is targeted at detection. Nevertheless the sensitivity is given in terms of minimum detectable flow rate (in liters per hour) at a maximum pressure hold and for a given background noise level. The detection threshold is defined as an RMS intensity at least 3 db above the amplitude of the background noise, during a 5-minute period. The transposition between flow rate and the sensitivity is in the scope of the qualification. The localization and sizing of AE information is outside of the RSEM qualification scope. Nevertheless, a zonal localization is realized and given to the Monitoring Team of the HPT in order to check and characterize any detected event in the monitored areas. In Service Inspection Impact The AE channels are set up and checked for sensitivity a few days before the HPT for the (RPV Head and PZR Bottom Head. But due to their localization on the RPV Bottom Head, the BMI channels are installed before the fuel unloading. It is therefore performed at least one month before the HPT under a heavy irradiation. In order to achieve a good sensitivity, a sensitivity check is realized at the installation: the response to a calibrated emission has to be above a threshold value. To ensure that the sensitivity is maintained throughout the full HPT, relative measurements between transducers are then realized at each step: installation, HPT beginning and HPT end. The background noise should not exceed a threshold value to guarantee the accuracy of the sensitivity check. Therefore, the installation has to be made during a low activity shift on the main primary loop. To express the performances in terms of minimum detectable flow rate, the background noise level should also be kept below a maximum value during the HPT. It implies that the HPT monitoring team has to balance the different flows between the pumps to avoid an excessive noise. These three constraints make the ISI quite difficult to monitor. In order to achieve gains during the ISI, the constraints had to be decreased. This could be done only by improving the sensitivity performance. THE FORMER RSEM PERFORMANCE DEMONSTRATION The initial performance demonstration relied on tests which were mostly performed during the 1980 s and 90 s. First a relation between the RMS signal and the leak flow rate was determined in laboratory conditions for different kinds of leak aperture, pressure and temperature on simple mockups. Several designs of leak aperture were tested. Finally the worst case was kept, and the performance was expressed in terms of flow rate through a rectangular, single canal aperture with a fixed surface. 669
Figure 3: Noise Level, RMS, versus Leak Flow Rate at AE sensor position In order to transpose the leak detection performance at the actual AE sensor positions (and not at the AE source), the attenuation between AE sensors and a potential leak had to be determined. The different attenuation values were measured experimentally during different first in-service inspections of the French PWR NPPs with a Hsu-Nielsen source, also known as Pencil Lead Break. Once the equation fit on the experimental curve, the impact of the different influent parameters; mainly : component Temperature, Electronic parameters (frequency bandwidth, sampling rate, gain), final hold pressure, RMS reference value during the penultimate pressure hold, sensor sensitivity ; was then embedded into the mathematically derived equation. The final relation was expressed through a log-log equation between the noise background level (in db) and the flow rate for each weld under examination, see figure 4. Figure 4: Example of Relation between flow Rate Detection level and Background Noise level, BMI to guide tube 900MWe Welds As shown in the above graphic, minimal and maximal values of the flow rate were estimated. THE IMPROVED ATTENUATION ESTIMATION The New Sources It appeared that the weakest point in the former performance demonstration was the estimation of the real attenuation. A real fluid leak is a continuous AE source, and has a wideband frequency spectrum. The 670
previously used source (Hsu Nielsen) can be considered as a short duration pulse with very few different frequencies. In order to improve the attenuation estimation, two new kinds of sources were used. The first source was a pulsing source based on a standard AE transducer. This transducer can work either in emission (reverse piezoelectrical effect) or in reception mode. This method has the great advantage of being already used for the ISI to check the sensitivity (calibration source). It is therefore easy to implement on-site. The frequency spectrum of this pulsing source has only a few frequencies centered on the resonant spikes of the piezo electric sensors. The bursts have a short duration and the response is more closely related to a transient signal. The second source is an Air Burst system which spreads a 5 bar jet of air onto the component surface. This stream is totally continuous in time and has a wideband frequency spectrum. The response is therefore relevant of a stationary state with the different mode interactions and propagations. Figure 5: AE Sources, Air Burst on left, Pulsing source on PZR Bottom Head on right Trials have shown that its behavior is well representative of a real fluid leak. Tests and comparisons between an air burst and a water leak (with rectangular single aperture) were made, and the final frequency spectrums were well correlated. However, this Air Burst source is difficult to use during inservice inspection. Its size and noise restricts it to areas with no irradiation and lot of available room. Figure 6: Frequency Spectrum for Air Burst and Water Leak Sources THE ATTENUATION TEST CAMPAIGN A full trial campaign was conducted during two years with the two sources. Several testing were made either on real component or on mockups. The only restriction was to use only the pulse source for tests during ISI of PWR units. The testings took place at the CETIC for RPV Head, the Areva Factory of Châlon-sur-Saône for the EPR RPV head and EPR PZR bottom end cover, the EDF PWR NPPs during ISIs for the RPV Core Bulk. Mockups of PWR BMI and RPV Head adaptators, and EPR Heater Sleeves were used to confirm former attenuation results either for geometry or thin component attenuation (e.g. nozzles). 671
Figure 7: Attenuation Measurement Test Campaign on Real Component, RPV Head and Nozzles The sources were localized in the different welds; and the sensor position, sensor to source distance, and shadow effect were tested and analyzed with pulsing and air burst sources. ATTENUATION RESULTS The main results are summarized below, see figure 8 and 9. First and foremost, it showed a behavior equivalent in near field to the approximate method that was used historically (a weld-crossing attenuation coefficient plus a linear attenuation coefficient). But a different behavior was observed for sensors localized in the far field of a source. The attenuation law is almost exponential in the first meter of propagation whereas it becomes negligible further away from the source. This near field effect can t be easily separated from the transition geometry effect (crossing of welds). It is the main source of attenuation. Both kind of sources show that the attenuation has an exponential decay in the 1 st meter and then a linear decay. Figure 8: RMS Variation Versus Distance, Air Jet Source 672
Figure 9: RMS Variation versus Distance, Pulsing Source Secondly, the pulse method is much more sensitive to the attenuation and to potential shadowing by nozzles. It is explained by the great number of mode interactions during a long signal duration with the Air Burst source (continuous AE), whereas the few frequencies of the pulse source fade away more significantly (discrete AE). Finally, a full attenuation map was measured with the air burst source. The impact of the nozzle placement (weld density) on the RPV Head and the RPV / PZR Bottom end covers on the attenuation law was estimated. This parameter appears to have little effect when a long duration source is used. Figure 10: RPV Head Attenuation Map As a conclusion, the attenuation law has been re-evaluated more accurately, and the attenuation values were decreased compared to historically-used values. This opened up the possibility of reducing the constraints on the background noise level while keeping the same performances as before. 673
IMPACT ON THE ISI The increase on sensitivity performances also opened up the possibility of mounting AE instrumentation on other areas. Especially considering that the RPV Bottom head channel instrumentation carries the most constraints of all the ISI. The simplest area into which the instrumentation used to monitor the BMI welds with little impact on the ISI organization was finally found to be the RPV inlet areas, between the Bimetallic Welds and the RPV Core bulk on a flat ferritic area. It is easily accessible compared to the RPV bottom end cover room: the transducer access is at ground level. The heat insulation is always removed during decennial outages. Furthermore, from a radioprotection point of view, even if the irradiation level is significant, it is still lower than in the RPV bottom end cover room. Finally, the RPV inlet AE channels can be installed on the same day as the RPV head, and PZR Bottom end cover AE channels. This implies no more actions before the fuel unloading, and the channels are exposed to high irradiation levels for a much shorter period of time. All of these elements secure the ISI and the performances of the NDE. Nevertheless, due to a bigger distance between the sensors and the BMI welds, and in order to keep the same level of performance, the constraint on background noise level had to be reevaluated. The noise constraint is balanced with the increase in sensitivity. As a result, the final constraint on noise background level during HPT was reduced by a factor 6, or one third in log scale. This is a completely realistic in terms of industrial conditions, given the positive feedback from former HPT monitoring. CONCLUSION In a nutshell, the use of new AE sources more representative of a real leak allowed for better estimation of the attenuation inside the components, which is the parameter that has the highest influence on the NDE sensitivity. Therefore, the improved attenuation estimation induced better performances of the NDE. As a result, the feedback from the ISI could be taken into account and new areas of instrumentation were chosen in order to reduce irradiation (AE instrumentation and personnel) without reducing the sensitivity and coverage area performances. This new organization of the ISI will also have a good impact in the schedule and monitoring team during the 10-year outage ISIs. 674