Docstoc

WPM_Hamburg_case_studies_2010_Commented

Document Sample
WPM_Hamburg_case_studies_2010_Commented Powered By Docstoc
					         Wind Turbines Condition Monitoring:
         Illustrative Case Studies



         Frederic CHAMPAVIER
         Monitoring & Diagnostic Services
         01dB-Metravib, AREVA Group




01dB-Metravib WPM 2010 - Hamburg forum – F.Champavier – 09/09/2010 - p. 1
                                              Condition Monitoring of WT
                                                            Introduction

   ► Objective of Condition Monitoring for WT
           To detect, identify and characterize, at an early stage, any
           relevant change in the condition of monitored components
           that may represent deviations from normal operational
           behaviour and lead to their premature failure.

   ► Main expected benefits
              Prevention of secondary damages = Surveillance
              Identification of the faulty components = Diagnostic
              Ability to schedule maintenance operations = Prognostic
              Ability to recommend maintenance or operation work =
               Decision Making Support
   ► Additional expected features
            Ability to verify equipment condition after maintenance
             work
            Understanding degradation mechanisms: Root cause
             analysis
            Continuous Reliability improvement



01dB-Metravib WPM 2010 - Hamburg forum – F.Champavier – 09/09/2010 - p. 2
                                                Condition Monitoring of WT
                                                   Case Studies Overview



 Case study #1: Tooth
breakage on sun gear




                                                                              Case study #3: High Speed Shaft
                                                                              Bearing Outer Ring defect




     Case study #2: Main
Bearing Outer Ring defect




  01dB-Metravib WPM 2010 - Hamburg forum – F.Champavier – 09/09/2010 - p. 3
                                            Condition Monitoring of WT
                                           Case Study #1: A gear failure

                                                                 ACC5




                                                                            ACC4    Periodic impacts are generated by
                                                                                   the broken tooth when meshing with
                                                                                             the planet gears




                            ACC2
       ACC1
                                                         ACC3




01dB-Metravib WPM 2010 - Hamburg forum – F.Champavier – 09/09/2010 - p. 4
                                               Condition Monitoring of WT
                                              Case Study #1: A gear failure
                                                                     Periodic impacts can be
                                                                   revealed by FFT … but with
   ► Failure pattern in FFT                                      very low amplitudes compared
                                                                  to other « normal » frequency
                                                                    components : The Energy
                                                                 generated by the defect is very
                                                                               low


                            This is the broken tooth            Blue trace :
                                          signature !           No defect              1X 3rd stage
                                                                                        gearmesh
                                                                                        frequency
                                                                   Red trace :
                                                                   With defect




                                                                        3X 2nd stage
                                                                         gearmesh
                               1X 2nd stage      2X 2nd stage            frequency
                                gearmesh          gearmesh
                                frequency         frequency




01dB-Metravib WPM 2010 - Hamburg forum – F.Champavier – 09/09/2010 - p. 5
                                            Condition Monitoring of WT
                                           Case Study #1: A gear failure

   ► Failure pattern in time domain                                          … but it requires unpredictable
          Periodic impacts can be seen in time                               specific filtering to remove the
                    domain signals …                                        “normal" vibration components :




                                                                  Blue = Unfiltered timewave
     Defect is likely missed by spectral analysis and            Red = Filtered timewave
      energy descriptors trending
     Defect is detectable by time waveform analysis
     Timewaves have to be processed using specific                                     Specific processing is
      band pass filters                                                                 required for defect detection
     The defect pattern is series of periodic spikes


01dB-Metravib WPM 2010 - Hamburg forum – F.Champavier – 09/09/2010 - p. 6
                                            Condition Monitoring of WT
                                           Case Study #2: Main bearing

                                                                 ACC5




                                                                            ACC4



                                                                                   Repetitive spikes are generated
                                                                                   by the rolling elements passing
                                                                                    on the cracks of the defective
                                                                                               outer race




                            ACC2
       ACC1
                                                         ACC3




01dB-Metravib WPM 2010 - Hamburg forum – F.Champavier – 09/09/2010 - p. 7
                                            Condition Monitoring of WT
                                           Case Study #2: Main bearing

   ► Failure pattern in FFT
                                             Blue trace :                   Some harmonic spectral lines
                                             Bearing defect                    and broadband noise are
                                                  Red trace :                revealed by FFT … with very
                                                  Bearing replaced                 low amplitudes




01dB-Metravib WPM 2010 - Hamburg forum – F.Champavier – 09/09/2010 - p. 8
                                            Condition Monitoring of WT
                                           Case Study #2: Main bearing

   ► Failure pattern in time domain                                         … and once again, filtering is
           Repetitive impacts can be seen in                                helpful, to remove “masking”
               time domain signals …                                            vibration components




                                              0,3 sec = BPFO

                             Impacts


                                                                      Blue = Unfiltered timewave
                                                                      Red = Filtered timewave
     Defect is likely missed by spectral analysis and
      energy descriptors trending
     Defect is detectable by time waveform analysis                                        Specific processing is
     Defect is more easily detectable in filtered signals                                  required for defect detection
     The defect pattern is series of repetitive spikes


01dB-Metravib WPM 2010 - Hamburg forum – F.Champavier – 09/09/2010 - p. 9
                                            Condition Monitoring of WT
                                           Case Study #2: Main bearing

   ► Time domain signal after bearing replacement



                       Blue trace :
                       Bearing defect




                     Red trace :
                     Bearing replaced




01dB-Metravib WPM 2010 - Hamburg forum – F.Champavier – 09/09/2010 - p. 10
                                          Condition Monitoring of WT
                                        Low speed anomaly detection

   ► Innovative processing for anomaly detection : ShockFinderTM
                                                                



           Long time domain data acquired during several                    Automatic process including adaptative
                      rotations of the turbine                         filtering, impact identification + events counting


                                                                


                      Result = Shocks detected ?                             You should investigate on this point !


01dB-Metravib WPM 2010 - Hamburg forum – F.Champavier – 09/09/2010 - p. 11
                                   Condition Monitoring of WT
                            Case Study #3: HSS Bearing failure

                                                                 ACC5




                                                                             ACC4

                                                                                        ACC6                  ACC7


                                                                                     Periodic impacts are generated by
                                                                                    rolling elements passing on surface
                                                                                                irregularities



                            ACC2
       ACC1
                                                         ACC3




01dB-Metravib WPM 2010 - Hamburg forum – F.Champavier – 09/09/2010 - p. 12
                                   Condition Monitoring of WT
                            Case Study #3: HSS Bearing failure
                                                                   … Envelope spectrum shows a nice
   ► Failure pattern in FFT                                        periodic impact pattern at inner ring
                                                                               frequency
         This time, Overall Acceleration has
          increased significantly because
           enough energy is generated...




                            … with a perfect modulation at HSS 1X
                               RPM due to load zone influence
                                                                              Failure is detectable using energy
                                                                               descriptors trending, e.g. Acc. RMS
                                                                              High Frequency Demodulation (envelope)
                                                                               is useful to remove masking components
                                                                              The defect pattern is series of discrete
                                                                               spectral lines and harmonics of Inner Ring
                                                                               defect frequency with 1XRPM modulation


01dB-Metravib WPM 2010 - Hamburg forum – F.Champavier – 09/09/2010 - p. 13
                                              Condition Monitoring of WT
                                              Case Studies: Conclusions
                                                                             Fair :
            Depending on the type and location of a
                                                                             Very low amplitude
            failure, the detection can be a challenging
                                                                             Low masking effects
            task.
           Reliable Automatic Detection is                                            Difficult :
            necessary                                                                  Low amplitude
           Specific processing are required to                                        Strong masking effects
            separate multiple vibration
            components                                                                             Easy :
                                                                                                   High amplitude
            Failure patterns are often low amplitudes                                              Strong masking effects
            spikes, likely to be unseen, ignored or
            confused with impacts due to operating
            changes.
           Managing operating conditions is
            mandatory
            Once detected, abnormalities have to be
            confirmed and diagnosed by monitoring
            personnel.
           Advanced Diagnosis and Advisories
            capabilities are required


01dB-Metravib WPM 2010 - Hamburg forum – F.Champavier – 09/09/2010 - p. 14
                                                               Condition Monitoring of WT
                                                                       Process Flowchart
      Detection = Automatic Process




                                                                 Data acquisition


                                                                                                             The main
                                                                                                            objective is
                                                                                                              to focus
                                                                 Data-processing                           efforts where
                                                                                                            they really
                                                                                                               count !


                                                                 Automatic State
                                                                    detection




                                                                   Diagnostic
Human Agent Processes




                                                               Health Assessment
 Diag./Progn./advisory =




                                                                                                             The main
                                                                                                          objective is to
                                                                                              Estimated       enable
                                                                   Prognostic                  Time to    unambiguous
                                                                Risk Assessment                Failure    and actionable
                                                                                                            advisories




                                                                 O&M advisories

                 01dB-Metravib WPM 2010 - Hamburg forum – F.Champavier – 09/09/2010 - p. 15
                    Thank you for your attention
                           Questions?




                            www.oneprod-system.com
                                   Frederic.Champavier@areva.com




01dB-Metravib WPM 2010 - Hamburg forum – F.Champavier – 09/09/2010 - p. 16

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:9
posted:9/4/2011
language:English
pages:16