Integrity Assessment of Headcovers and Headcover Fasteners

Transcription

1 Integrity Assessment of Headcovers and Headcover Fasteners Annette Karstensen, Quinton Rowson, Nathan Stanbridge, Kevin Stevens Quest Integrity Group (Quest Integrity) Robert Dillon Hydro Tasmania Abstract Since the catastrophic failure of the headcover in the Russian Sayano-Shushenskaya hydropower plant, the integrity of headcovers and headcover fasteners have become of particular interest to plant owners across the world. Fatigue initiation in fasteners can lead to failures as seen in the Sayano-Shushenskaya incident. This paper considers factors that affect the fatigue strength of fasteners as well as issues surrounding the material performance and material degradation such as embrittlement of bolting material. The concerns and the lessons learned from the Sayano-Shushenskaya incident have resulted in a series of assessments of headcovers and fasteners from various power stations operated by Hydro Tasmania. The assessments analysed a variety of integrity issues such as: fatigue performance of the fasteners, wall thinning caused by erosion and weld cracking. The assessments that are presented employed fitness-for-service methods to determine under which conditions the fasteners will not be able to carry the intended load resulting in a potential failure. The conditions that were analysed were the effect of pretension and corrosion in the fasteners. The assessments included headcovers from 3 different power stations; some of which previously exhibited cracking. A finite element analysis was employed to determine the through thickness stress distribution which was used in the fitness-for-service analysis. Additionally, using the results from the stress analysis it was also possible to investigate the effect on the integrity of corrosion within the headcover. From an initial analysis of one of the headcovers, it was determined that a vibration assessment should be conducted. This assessment included vibration measurements taken from the headcover during startup and steady state operation. The results of the measurements were incorporated into the stress analysis and a subsequent fatigue analysis was carried out on the critical regions of the headcover identified in the analysis. The results of the analysis provided Hydro Tasmania with quantitative evidence with respect to areas within the headcovers that should be targeted for inspection and maintenance strategies. 1

2 Introduction On the 17th of August 2009 turbine 2 of the Russian hydropower plant Sayano-Shushenskaya violently broke apart initiating a chain reaction of devastating events. The failure occurred due to fatigue of the fasteners which held turbine 2 s headcover in place. This disaster caused the turbine hall ceiling to collapse, the turbine hall and control room to become flooded, 9 of the 10 turbines were damaged or destroyed, and 75 people lost their lives. The entire plant output totalling 6400MW and a significant amount of power to the local grid was lost resulting in power failures throughout the local area. The Russian event increased the awareness of the consequences of headcover failure amongst hydrostation operations around the world. Quest Integrity have used finite element analysis and fitness for service methods to assess the integrity of several headcovers in Australia and New Zealand. Many aspects were investigated including stresses, vibrations, fastener corrosion, susceptibility to fatigue, and critical flaws. Sayano-Shushenskaya Background [1] [2] Although the vibration monitoring equipment on turbine 2 was out of service at the time of the incident, the most likely cause of the fatigue occurring in the head cover fasteners was due to the severe vibrations the turbine was experiencing. An upward force caused the runner, shaft, head cover, wicket gate upper trunnions and operating arms, both bearings, the generator rotor, and all associated structure of Unit 2 to be forced upward with enough violence to crush the rotor spider and destroy the stator. The entire unit had a combined weight of approximately 1500 tonnes and required a large force to lift it, suggesting that something triggered Unit 2 to fail suddenly and violently. A large force was most likely present at the time of failure and several hypotheses have been put forward as to the origin of this force. One hypothesis is the possibility of a water hammer occurring with enough force to lift the turbine assembly. A water hammer occurs when the downstream pipeline flow is rapidly brought to rest resulting in a rapid pressure increase which has the effect of hammering on the pipeline or cover. The occurrence of water hammering phenomena in combination with fatigued bolts could possibly have resulted in the destruction seen at Sayano-Shushenskaya. Another plausible hypothesis is that the Sayano-Shushenskaya failure could have occurred due to a water column separation. Water column separation occurs when the gate at the entry point of the draft tube is closed too quickly. The water column then becomes separated in the draft tube, followed by an extremely violent pressure rise as the water column is re-joined under the head cover. Observations in the hydropower industries throughout the past have indicated that plant operators sometimes adjust the equipment in the plant to improve productivity. One such adjustment is to modify the orifice control of the wicket gates in order to speed up the movement of the gates, which has occasionally resulted in water column separation occurring. The fact that turbine unit 2 at Sayano-Shushenskaya had a new governor installed in early 2009 suggests the possibility that the new governor was used to speed up the wicket control resulting in the violent failure. Hydro Tasmania s headcover analysis Quest Integrity investigated the integrity of several headcovers from Hydro Tasmania s hydropower plants. The stations that were considered are summarized in Table 1. Table 1. Hydro Tasmania stations included in this analysis. Station Number of units Max output Year of commission Cethana Lake Echo Liapootah

3 The fitness for service integrity program included conducting finite element analysis, vibration analysis, fracture mechanics and fatigue assessment of the headercover joints and fasteners. The following tasks were completed in the study: Determination of the stresses in the bolted and welded joints in the headcover considering steady-state operating conditions. This task was performed by creating a finite element model of the headcover, including allowance for modelling of the headcover fasteners and the amount of pre-tension in the fasteners. Determination of the effect of a variation in preload tensions in the fasteners. Estimation of the minimum effective bolt area subject to corrosion. Calculation of critical flaw curves for regions of peak tensile stress. Performing a modal analysis considering the first 10 vibration modes of the headcover in order to determine the optimal locations for mounting accelerometers for detailed vibration measurements during start-up, rough running, normal operation and shutdown. Finite element analysis The finite element analysis of the headcover was conducted in the as-designed condition to determine the resultant stress field associated with the steady-state operation. To obtain this stress field a model (to the design specifications) was created, as shown in Figure 1. After a finite element mesh was created, boundary conditions and loads were applied to complete the model. The model was then run and validated to ensure an accurate finite element solution was produced. Guide vane Bolt holes Figure 1: Headcover finite element model (Lake Echo), only 1/6 of the model was created, but this figure showed the model mirrored 6 times for a 360 perspective The boundary conditions of the model contained two symmetry conditions which prevented rotation and translation in the tangential and radial direction. The vertical force acting on the headcover from the washers and bolts was modelled using springs. To give an approximate model the springs were connected to the headcover by multi-point constraints, which constrained the surface nodes on the headcover within a defined radius of influence from the centre of the fastener. The stay ring was modelled as a non-deformable fixed surface with a contact interaction with the headcover bolt flange. The loads considered for the headcover consisted of the bolt preload, hydrostatic pressure, self-weight, draft tube pressure fluctuations, and external bearing static loads. 3

4 Summary of Integrity Assessment The calculated von Mises stress distribution in the headcover is shown in Figure 2. The applied load what has been estimated as the maximum load during operation. This loading is 15% fluctuation of draft tube pressure applied to lower flange above runner and top flange where shaft bearing load are transmitted. Once more testing have been completed the pressure will be confirmed. Figure 2: Von Mises stress distribution in Cethana due to maximum applied load during operation, units are Pa, maximum stress of 114MPa is shown (as a red region) in the weld between the vertical and the horizontal web. One effect of the variation of fastener preload was a reduction in contact area between the bolt flange and the stay ring, this is shown in Figure 3 where the contact pressure is shown for L1: 90% bolt preload plus body load (left hand side), L2: 90% bolt preload plus service load (middle) and L3: 50% bolt preload plus service load (right hand side). The areas of blue on the plots indicate regions where the surfaces are not in tight contact. The significant reduction in contact may cause leakage and expose the fasteners to water resulting in problems associated with corrosion and aging. Load Case Load Applied Percent (%) Reduction in Contact Area L1 (left figure) 90% bolt preload + body load 0% L2 (middle figure) 90% bolt preload + service loads 38% L3 (right figure) 50% bolt preload + service loads 57% Figure 3: Reduction in contact area due to variation in fastener preload. The magnitude of the stress found at the toe of the radial web support near the shaft inner bore exceeded the maximum allowable stress limit of the material as shown in Figure 4. This was partially due to refinement of the finite element mesh at the fillet weld. However, because the only points of reaction to the vertically applied steady state pressure loads were at the fasteners, although the stress was compressive, it was real. Application of 50% reduction in bolt preload resulted in no appreciable change to the stresses in the headcover. 4

5 Figure 4: The toe of the Radial Web Support at the Shaft Inner Bore was shown to be above the maximum allowable stress (at locations indicated in red). Fracture mechanics assessment of the bolts indicated that the bolts were extremely sensitive to even small cracks. The bolt critical crack sizes for the as designed load case can be seen in Table 2. To investigate the sensitivity to embrittlement, the fracture mechanics calculation was carried out for two different toughness values: 866N/mm3/2 simulating embrittled material and 2500 N/mm3/2 simulating non-damaged material. This fracture mechanics assessment was combined with corrosion analysis as indicated in the left hand column of Table 2. Table 2. Bolt critical crack sizes for the as designed load case. % of original area due to general metal loss from corrosion Maximum stress in section (MPa) Critical crack depth for embrittled material mm Critical crack depth for non-damaged material mm Figure 5: 10th mode of the modal analysis of the Cethana headcover. 5

6 As a part of the integrity program Hydro Tasmania wanted to measure the vibration on the headcover associated with operation of the Cethana unit. In preparation for the vibration measurement a dynamic modal analysis was carried out to determine the optimum location for the accelerometers to be placed. An example of the displacement normalized eigenvector for the 10th mode can be seen in Figure 5. The conditions under which the vibration measurements were collected are summarized in Table 3. Table 3. Test Summary of test conditions for vibration measurement. Machine Power Output (MW) Sample rates (KHz) Sample time (Sec.) Explanation provided of test conditions CE Machine at stand still before testing there is still low level vibration as there is water leakage past the closed guide vanes - Machine starting up from shut down to 0MW CE Rough running, air is also being sucked into the draft tube CE Rough running, air is also being sucked into the draft tube CE Smooth running CE Smooth running CE Smooth running, guide vanes fully open CE Smooth running, guide vanes fully open CE Machine shutting down from 0MW to stand still CE Machine at stand still after testing there is still low level vibration due to water leakage past the closed guide vanes Figure 6: Photo showing the location of one of the accelerometers on the headcover web 6

7 Figure 7: Compression of vibration measurements from rough running (left) and smooth operation (right) The location of one of the 6 accelerometers is shown in Figure 6. Figure 7 shows data collected from rough running (left) and smooth running during operation (right). The data was analysed with a view to apply the vibration measurements to the FEA model and conduct a fatigue analysis based on the data. This work is still to be completed; however Figure 8 shows the power spectrum density (PSD) as a function of frequencies for the two running conditions shown in Figure 7. Figure 8: Comparison of power spectrum density (PSD) of POSITION 1 for rough running (blue) versus smooth operation (red). A rainflow analysis of the data collected shows that there is no significant variation of frequencies over the time of the data sampling. From the results presented in Figure 8 it can be seen that apart from a single high frequency (5KHz) the remaining frequencies in the signal have relatively little power. 7

8 Conclusion A series of structural integrity investigations were carried out on headcovers from three power stations operated by Hydro Tasmania. Quest Integrity s results for the finite element stress analysis show the magnitude of the stress found at the toe of the radial web support near the shaft inner bore locally exceeds the maximum allowable stress of the material and could potentially become a site for crack initiation. Therefore it is recommended that areas around these fillets should be subjected to inspection during scheduled outages. The assessment also showed that reducing the bolt preload showed no noticeable changes in the overall stresses in the headcover. Reduced preloading of the bolts resulted in reduction of contact area between the stay ring and headcover bolt flange which may result in the fasteners becoming exposed to water and could potential cause issues with corrosion. The maximum tolerable crack size in the fasteners for embrittled material is 1.3mm compared to 3.5 mm for non-damaged material. Vibration measurements undertaken on a headcover at Cethana power station showed that most vibration occurs at a frequency of 5kHZ. References [1] F.A.Hamill, Sayano Shushenskaya accident presenting a possible direct cause, [2] V.A.Pekhtin, E.N.Bellendir, B.B.Bogush, Lessons from the Accident at Sayano-Shushenskaya Hydropower Station and Rehabilitation Work Progress, Acknowledgement Quest integrity would like to thank Hydro Tasmania for permission to publish this work. 8