Vibratory detection system of cavitation erosion: historic and algorithm validation

Proceedings of the 8 th International Symposium on Cavitation CAV2012 Abstract No. 150 August 14-16, 2012, Singapore Vibratory detection system of cavitation erosion: historic and algorithm validation Lafleur François IREQ: Research Institute of Hydro-Québec lafleur.francois@ireq.ca +1 (450) 652-8802 SUMMARY Cavitation erosion inspections and repairs of hydraulic turbine runners require long periods of outage that limit hydropower generation. Although a current design tendency is to lower the cavitation of new runners, cavitation erosion continues to impose costly repair and loss of revenues. A vibratory detection system of cavitation erosion was installed 10 years ago for continuous monitoring of 4 high-power hydro-generators (> 300 MW) in a Hydro-Québec power plant. A new hardware version of the system was developed and installed in 2010. This new system configuration is more reliable than the previous generation. This situation improves the availability of the system and then allows more accurate evaluation of the cavitation erosion of the runners. The vibration measurements at the lower guide bearing and the analysis of the Mean Square Value (MSV) along with its modulation at the high frequency range (generally 5 to 15 khz) allow the estimation of the cavitation erosion. The algorithm can be use to evaluate the relative cavitation signature vs. of operating conditions of the power plant or the continuous monitoring of the absolute cavitation erosion rate (in kg/10 000h). The absolute evaluation appeals for a vibratory transmissibility function between the blade and the guide bearing. In this research project the validation of the erosion evaluation algorithm is a continuous task. On planned inspection activities, the cavitation erosion was directly measured on each blade of the runners using a cavity filling method giving the exact volume of eroded material and by a normalized IEC 60609 method that allows an estimation of the same cavity by overall dimension measurements. The evaluation of the vibratory transmissibility function was also performed during this inspection. These evaluations are compared and will be used to enhance the erosion evaluation algorithm. This paper presents the historic of this vibratory detection system of cavitation erosion and will discuss the present work of validation of the erosion evaluation algorithm. Future work will also be discussed. INTRODUCTION Cavitation erosion which results from repeated collapse of transient vapor cavities on solid surfaces is a constant problematic in hydraulic turbine runners and continues to enforce costly repair and loss of revenues. Furthermore cavitation erosion may reduce the power efficiency of the runner. Only in repair cost several M$ a year is affected to this problematic and the lost of revenues for long repair or inspection periods are even greater. Also, the inspection plan proposes prescheduled periods of at least 5 years between inspections. This situation makes very limited information available to the power plant exploitation about the erosion status of the runner blades. One way to gain information about the erosion status is to implement a cavitation detection system. Vibratory detection of cavitation in hydraulic turbine runners is a long time activity at Hydro-Québec and was in several power plants. The advancement in this field that will be described in this paper was pioneered by Paul Bourdon (researcher at IREQ) and lead to a series of publications and internal reports [1, 2, 3]. HISTORICAL REVIEW Cavitation erosion detection is a field of interest since several years at Hydro-Québec. Several types of intervention were performed at different power plants and on prototypes for the characterization of cavitation to develop a vibratory algorithm of cavitations detection. Most of the work was performed on the investigation of the trailing edge cavitation for Francis and Kaplan runners. Over the years, the system has been used in different ways: 1

The evaluation of aggressiveness of cavitation on different type of materials. The evaluation of the cavitation signature vs. of operating conditions at several power plants. The continuous monitoring of the absolute cavitation erosion rate (in kg/10 000h) of 4 groups in a power plant. The expertise and methodology were implemented in the form of a complete cavitation detection system including multichannel high speed acquisition hardware and used in conjunction with specific software modules. The following paragraphs summarize some information on the system algorithm and data analysis. The data analysis II required the knowledge of the runner characteristics in term rotation speed, number of guide vanes and blade to calculate the pertinent fraction of the signal that correlates with the cavitation erosion. This pertinent fraction is related to the spectral analysis of the acceleration signal. The blades and vanes passing frequencies harmonics along with the number of associated side lobes with each these harmonic components are taken in to account in the cavitation erosion calculation. A typical result shows the relative aggressiveness of cavitation (normalize MSV) vs. the guide vanes opening for a group at the Sept-chutes power plant (showed on Figure 3). The same type of results is available in term of relative agressivness vs. the output power of the group in MW. a) Algoritm of cavitation detection This system is based on vibration measurements analysis at the lower guide bearing and the algorithm is described in details in the work of Bourdon [1, 4]. The measurement and analysis of the Mean Square Value (MSV) and its modulation at high frequency (generally 5 to 15 khz) is the base of the cavitation erosion algorithm. There is three type of data analysis: Data analysis I is basically the high speed data acquisition. Data analysis II is the calculation of the relative aggressiveness of cavitation vs. the operating condition (cavitation signature) using the pertinent fraction of the modulated MSV (Mean Square Value in m 2 /s 4 ) of acceleration. Data analysis III is use to evaluate the absolute cavitation erosion, by continuous monitoring, the knowledge of a vibratory transmissibility function between the blade and the guide bearing is necessary. Figure 1: Typical accelerometer installation at the Robert Bourassa power plant b) Cavitation signature detection vs. operating conditions As stated before, cavitation signature refers to relative aggressiveness of cavitation vs. the operating conditions. This is done by following an acquisition procedure using the DATAMAX software and an analysis procedure using the CAVIMAX software. The instrumentation and data acquisition procedure is well defined [4]. One or more accelerometers with mounting frequencies of 45 khz or higher are installed at the lower guide bearing (see typical installation at Figure 1). Also a tachymeter is used to synchronize the acquisition to the runner rotation and the data from the group operating conditions are usually transferred from the power plant monitoring system or directly measured. Figure 2 shows the installed instrumentation to monitor the guide vane opening of a group of the Sept- Chutes power plant. Figure 2: monitoring of the guide vane opening at the Septchutes Power plant The data acquisition procedure includes several steps from 0% to 100% of guide vane opening. At every step there are 30 blocks of 6 seconds time data recording at a rate of 100ksample/seconds [4, 5]. A sensors verification step at 200ksample/seconds is also performed. 2

Relative aggressiveness of cavitation 1 0,9 0,8 0,7 0,6 0,5 0,4 0,3 0,2 0,1 0 0 10 20 30 40 50 60 70 80 90 100 Guide vane opening (%) Figure 3: Relative aggressiveness of cavitation vs. guide vane opening of a typical group at the Sept-chutes Power plant c) Continuous monitoring of the absolute cavitation erosion The continuous monitoring of absolute cavitation erosion is performed using the CAVICIEL software module. This monitoring requires similar instrumentation than relative cavitation detection and the same information about the runner characteristics as well as the knowledge of a vibratory transmissibility function between the blade and the guide bearing. A series of 32 blocks of time at a sampling rate of 65536 samples/second are performed sequentially on the monitored group of the power plant. Figure 4 shows the display panel for the continuous monitoring of 4 groups of high-power hydro-generators (> 300 MW) at the Robert Bourassa power plant. This display includes indication of the status of the groups that are monitored, the status of data acquisition, the number of hours of operation and the absolute cavitation erosion on each of the monitored groups. At the date of 15 April 2012 we notice that three groups are in operation (green indication) and that the one group is not running (red indication). The number of hours of operation for each group was reset at zero after the visual cavitation inspection on the summer 2011. For example, total metal lost of group 7 (for all 15 runner blades) during this period is estimated to 3.97 Kg in 5976 hours. Figure 4: Display panel for the continuous monitoring of 4 groups of high-power hydro-generators at the Robert Bourassa power plant on April 15 2012 Other graphical information is presented as the synchronous average, time and signal with their frequency content over 1 rotation of the runner. Especially the synchronous average shows a visual estimation of the importance of the cavitation erosion. POTENTIAL IMPROVEMENT TO THE CAVITATION DECTECTION SYSTEM The Hydraulic Plant Life Interest Group (HPLIG) of 2009 [6] compare on technical parameters the existing on-line monitoring system for turbines. The report identifies two detection technologies: The system based on accelerometer measurements The system based on acoustic emission measurements Both types of technologies are appropriate for the cavitation monitoring with the general objectives indentified in the report: Optimization of operating loads Optimization of cavitaion repair activities Optimizing or validating the turbine aeration. As stated by the report, the Hydro-Québec cavitaion erosion evaluation system which assessment is based on acceleration measurements has obviously a good correlation with the actual erosion on Francis turbine. But the lack of reliability of the first continuous monitoring system (operation based on older exploitation systems) lead to interruption in data acquisition due to the computer malfunction made the cumulative lost of metal incomplete in the period of 2000 to 2008. After this period, a fatal failure of the computer occurred and the continuous monitoring system was not maintained. NEW HARDWARE IMPLEMENTATION In 2009, a new hardware version of the system was developed and installed. This new system is more reliable than the previous generation. Basically the same software module were used but adapted to the new data acquisition hardware. Attention was paid to optimize the availability of the cavitation system. The reliability of the computer and the development of a fail proof reboot procedure and remote control of the system improved the availability of the system. It has to be remembered that the Robert Bourassa power plant is approximately 1400 km from the IREQ research center and that remote control and access is one of the key parameter in the success on the monitoring This availability increase of the system permits more accurate evaluation of the cavitation erosion of the runner. At the present moment the system has run continuously since 26 month (February 2010) except for an early failure of the hard drive that was replaced by a more reliable solid state hard drive plus a 3 weeks period where the acquisition manually stopped by a system operator error. The reliability of the original accelerometers has been proven when they were change at the summer of 2011. These accelerometers installed in 2000 on the lower guide bearing were checked for variation of mounting resonance frequencies and for sensibility variation. This validation shows that the sensor selection and 3

the installation procedure are adequate for the long term continuous monitoring activities. Figure 5 shows the cavitation erosion monitoring system located at the Robert-Bourassa power plant. Fig. 5: Continuous cavitation erosion monitoring system located at the Robert-Bourassa power plant A) Filling method A filling method was developed over the year at IREQ using standard water based latex caulking tube. This product is specified for porous surface and is easy to use for filling of the cavitation defects of the runners. The caulking tubes were calibrated to allow the determination of the defect volume for each blades of the runner. The metal loss in then calculated from the evaluation of volume of the defect. Typical analysis is presented at Table 1 for a group of Robert-Bourassa power plant. For this group, the total evaluation of metal loss is 12.45 kg. A period of less than 4 hours is required to perform this inspection and the metal loss is available automatically using an anlysis template. Figure 6 illustrates the method. Figure 7 and 8 show the cavitation defect of a typical blade of a runner at Robert-Bourassa power plant respectively before and after latex caulk filling. COMPARISON OF CAVITATION INSPECTION RESULTS PRESENT WORK The present work includes several topics: Filling method IEC 60609 method The follow up of reliability of the new system including software module updates to add some functionalities to the software and to revisit software interface [4]. Updating and documenting the methodology of cavitation damage inspection for the purpose of modeling the cavitation erosion. Perform the evaluation of damage on a regular basis to be able to make adjustments to the model parameters. Evaluation of the transmissibility function results between the blades and the lower guide bearing in different conditions. We will look at the historic result of the transmissibility function in water and in air and make some comparisons in different frequency range at time period. Validation of the cavitation erosion model: Laboratory work [7, 8, 9] on the cavitation erosion resistance along with the result of the periodical damage detection will allow us to update the model. CAVITATION INSPECTION RESULTS Planned inspection activities were schedule at the summer of 2011 on the 4 monitored groups at the Robert-Bourassa power plant. Cavitation erosion was directly measured on each blade of the runners using a cavity filling method giving the exact volume of eroded material and by a normalized IEC 60609 method that allows an estimation of the same cavity by overall dimension measurements. The evaluation of the vibratory transmissibility function was also performed during this inspection. These evaluations are compared and will be used to enhance the erosion evaluation algorithm. Blade no. Volume of latex caulk (cm3) equivalent weight of inox steel (kg) Blade no. equivalent weight of inox stell (kg) 1 147,9 1,18 1 1,66 2 66,5 0,53 2 1,14 3 88,1 0,71 3 1,25 4 88,1 0,71 4 1,30 5 327,4 2,62 5 5,07 6 44,9 0,36 6 0,47 7 147,9 1,18 7 2,69 8 236 1,89 8 3,32 9 307,5 2,46 9 2,64 10 41,5 0,33 10 0,46 11 58,2 0,47 11 1,21 Total 1553,9 12,45 Total 21,20 Estimated time since last repair of cavitation erosion damage: 18335 hours Calculated Erosion rate (Filling method) 6.79 kg/10000hours Calculated Erosion rate (IEC 60609 method) 11.56 kg/10000hours Table 1: comparison of cavitation inspection results for a group of Robert-Bourassa power plant. Filling method vs. IEC 60609 method 4

B) Normalized IEC 60609 method [4] Figure 6: Illustration of the filling method procedure For this evaluation, the cavitation defects volume are evaluated by an approximation of the overall dimensions of the defect. A correction factor (usually 0.5 is applied) to the results. Figure 9 show the cavitation defect with some overall dimension for blade no.2 of the group (reference Table 1) at Robert-Bourassa power plant. Notice that the identification of the overall dimension of the cavitation defect is difficult and that one should suspect the metal loss evaluation by this method to be over-evaluated. Typical analysis is presented at Table 1 for a group of Robert-Bourassa power plant. The total evaluation of metal loss is 21.20 kg. The time period to perform the IEC 60609 measurements is similar to the filling method but the analysis is longer. The over evaluation of the IEC method is of 1.70 in average from real cavitation defect volume from the filling method. Notice the IEC method is used by power plant authority to evaluate the total mass of metal that will be necessary for the blade reparation and is not suitable for the exact evaluation of the cavitation defect volume. Figure 7: Typical blade of a runner at Robert-Bourassa power plant before latex caulk filling Figure 9: Cavitation defect with some overall dimensions for blade no.2 of the group (reference Table 1) at Robert-Bourassa power plant. C) evaluation of the erosion rate Figure 8: Typical blade of a runner at Robert-Bourassa power plant after latex caulk filling The last repair of cavitation erosion damage this group was performed in 2007 and the total operation time evaluation is 18335 hours. The erosion rate calculated using loss mass of the filling method is of 6.79 kg/10000hours. This value cannot be compared to the continuous monitoring system because it was not operational during this entire period. The erosion rate evaluated with the data of the IEC 60609 is not accurate for the cavitation model validation. At most this value (IEC 60609) can be use to estimate the amount of material that will be necessary for the repair of cavitation erosion damage. 5

TRANMISSIBILITY FUNCTION. Evaluation of the transmissibility function results between the blades and the lower guide bearing in different conditions were performed. Per example, these results of the transmissibility function of a group (reference table 1) in air and in water measured at installation of the system (Figure 10). Similar results were obtained in 2011. Presently the cavitation erosion algorithm is based on the transmissibility function in water. Due to reduction of production downtime, it is more difficult to have the opportunity of measured the water transmissibility function. One of the possibilities is to use the transmissibility function in air that is much easier and faster to measure for the calculation of the cavitation erosion. The discrepancies of the between the use of the transmissibility function in water or in air are presently in evaluation. AvrageTransfmissibility function 0,00001 0,000001 0,0000001 0,00000001 Air Values order 40 Eau Values order 40 AVG Ratio air/water 5-15 khz 10.40 0,000000001 0 5000 10000 15000 20000 25000 30000 Frequency Hz Figure 10: Results of the transmissibility function of a group (reference table 1) in air and in water measured at installation of the system CAVITATION MODEL VALIDATION The cavitation model validation plan is based on the results of the next cavitation erosion visual inspection planned in 2013. Theses information will give the erosion rate in this period and will be compared with the cavitation system result. The enhanced reliability of the new system will allow having complete results over the same period. A statistical model will also be study to find a relation between the operating condition and the cavitation erosion. year at IREQ using standard water based latex caulking tube. This method is rapid and gives the exact volume of eroded material needed to validate the cavitation model over the time. Future work will focus on model validation, new approaches of absolute cavitation erosion estimation based on the used of the air transmissibility function and the uses of a statistical model. REFERENCES [1] Bourdon, Paul, Détection Vibratoire de l érosion de cavitation des turbines Francis, Ph. D. thesis, École Polytechnique de Lausanne, thesis no 2295 (2000), 406 pages. [2] Bourdon, P., Simoneau, R., Lavigne, P., A vibratory Approch to the Detection of Erosive Cavitation, ASME International Symposium on Cavitation and Erosion in fluid system, San Francisco, California, December 1989. [3] Bourdon, P., Simoneau, R., Avellan, F., Hydraulic Turbine Cavitation Pitting Detection by Monitoring Runner Vibration, Canadian Electrical Reseasch Projet 307 G 657 Report, June 1993, Montréal, Canada. (Available from CATI International). [4] IREQ internal reports, IREQ: 99-191C, 99-197C, 2000-129C, 2000-132C, 2001-039C, 2001-078C, 2011-0067. [5] Mossoba, Y., Centrale Sept-Chutes: Expertise sur la Cavitation du Groupe 2 par la Detection Vibratoire, Hydro-Québec internal report (APP 2010 1065), 2010. [6] Mossoba, Y., On-Line Cavitation Monitoring, Hydraulic Plant Interest Life Group (HPLIG), CEATI Report No. T062700-339, January 2009. [7] Tôn-Thât, L., Experimental comparison of cavitation erosion rates of different steels used in hydraulic turbines, 2010 IOP Conf. Ser.: Earth Environ. Sci. 12 012052. [8] Tôn-Thât, L., Cavitation erosion behavior of high strength steels, 2012 International Symposium on Cavitation, Singapore. [9] Simoneau, R., "Austenitic stainless steel with high cavitation erosion resistance," USA Patent US 4,751,046, 1988. [10] CEI/IEC 60609-1:2004, Internal Standard, Hydraulic turbines, storage pumpsand pump-turbines Cavitation pitting evaluation Part 1: Evaluation in reaction turbines, storage pumps and pump-turbines. CONCLUSION A detection system gives more frequently updated information about the cavitation erosion of blades. This information can be used to optimize the power plant exploitation. A vibratory cavitation system was developed at Hydro-Québec. This system allows performing relative and absolute measurement of cavitation damage. The reliability of the initial system was not optimum due to computer and exploitation system failures. The new hardware version of the system is proven to be more reliable. The validation of the cavitation erosion algorithm is based on visual inspection performed every 2 years. This work showed the efficiency of a filling method developed over the 6