1、10FTM13AGMA Technical PaperGear Design for WindTurbine Gearboxes toAvoid Tonal NoiseAccordingtoISO/IEC61400-11By Dipl.-Ing. J. Litzba, HansenTransmissions International N.V.Gear Design for Wind Turbine Gearboxes to Avoid TonalNoise According to ISO/IEC 61400-11Dipl.-Ing. Jrg Litzba, Hansen Transmiss
2、ions International N.V.The statements and opinions contained herein are those of the author and should not be construed as anofficial action or opinion of the American Gear Manufacturers Association.AbstractPresent wind turbine gearbox design usually includes one or two planetary gear stages and at
3、least one highspeed helical gear stage, which play an important role regarding noise and vibration behaviour. Next to theoverall noise of the gearbox and the structure-born noise on the gearbox housing also tonal noise isbecoming a much more important issue in recent years. Since tonal noise is prob
4、lematic due to the humanperception as “uncomfortable”, avoidance is important. Conventional theories regarding low noise geardesign are not developed in view of tonal noise. This leads to the question: How to deal with tonal noise in thedesign stage and which gear parameters can be used for an optim
5、isation regarding good tonal noisebehaviour?Within a research project measurements have been performed on different gearboxes using different geardesigns. These measurements have been evaluated according to ISO/IEC 61400-11 and the results havebeen analysed in view of the influence of different gear
6、 parameters. It was also investigated if it is possible torank gearboxes in wind turbines according to their tonal noise behaviour as observed on the test rig.The paper will give an introduction into the definition of tonal noise according to ISO/IEC 61400-11 and giveinsight in measurement results f
7、rom test rigs and from gearboxes in the field, where noise behaviour is alsoevaluated according to ISO/IEC 61400-11. Furthermore, the paper will show and discuss the link betweenmeasurement results and different gear parameters, which are affecting tonal noise behaviour. In additionsimulation result
8、s will be presented, showing how tonal noise can be estimated within the design stage usingstate-of-the art calculation software.The paper will give recommendations regarding a gear design process that is considering tonal noise in thedesign stage and will compare an, regarding tonal noise, improved
9、 gear set with an older one.Copyright 2010American Gear Manufacturers Association500 Montgomery Street, Suite 350Alexandria, Virginia, 22314October 2010ISBN: 978-1-55589-988-23Gear Design for Wind Turbine Gearboxes to Avoid Tonal Noise According toISO/IEC 61400- 11Dipl.-Ing. Jrg Litzba, Hansen Trans
10、missions International N.V.IntroductionTypical wind turbine gearbox (WTG) designsinclude one or two planetary gear stages and atleast one high speed helical gear stage (HSS).These high speed gear stages play an importantrole in overall gearbox noise and vibration behavior.Noise behavior is character
11、ized by overall soundpressure level as well as tonal noise behavior. Thelatter aspect is gaining importance in the windturbine industry. Hence requirements are beingspecified in international standards 2 5 6, whichare still under development.Figure 1 illustrates how the gear mesh cangenerally be see
12、n as the main source of noise andvibrations. Shafts, bearings and housing are thetransfer path to the environment, resulting inairborne and structure-borne noise.Conventional theories regarding low noise geardesign focus mainly on reduction of overall soundpressure. This approach, which has been use
13、dover the last years, leads to a gearbox design withlower overall sound pressure level (includingreduction of gear mesh excitation), but pure tonesmay become more problematic due to a reduction ofbackground noise levels, e.g. by improved gearquality. Figure 2 shows a comparison of overallsound press
14、ure level and tonal noise for a gearboxdesign where tonal noise is masked due to highsound pressure level (Figure 2a) and a gearboxdesign where tonal noise is more visible due to a lowoverall sound pressure level (Figure 2b). Althoughoverall noise and gear mesh excitation levels havebeen reduced in
15、the latter gearbox design(Figure 2b), pure tones are clearly visible in thefrequency spectra.The term “tonal noise” is used for a sound whichincludes a specific frequency component with suchhigh amplitude that it is perceived as “uncomfort-able” for human hearing. Current evaluationmethods are based
16、 on standard 1/3rdoctave bandanalysis or procedures according to 2 and 6specifying that no pure tone is allowed and depend-ing on the level of the pure tone a penalty is appliedon the total sound power of the gearbox 1.Figure 1. Model of noise origin coming from the gear mesh of a 13.2 MW HSS gear s
17、et with770 mm center distance4Figure 2. Tonal noise compared to sound pressure level a) gearbox with high sound pressurelevel where tonal noise is masked b) gearbox with low sound pressure level where tonal noise isless maskedTonal noise problems should be prevented ingearboxes from the design stage
18、. However, experi-ence on how to avoid tonal noise in wind turbinedrive trains is limited and only few guidelines areavailable. The available information is more relatedto automotive 9, helicopter 7 or train applications8, or focuses merely on damping e.g. “How toeliminate tonal noise by active damp
19、ing” 7.Currently used parameters for low noise geardesign include:S low transmission error (TE) 9 10 18 20S large total contact ratio, preferably an integernumber 13S overlap ratio preferably an integer number 1012 15Furthermore, international standards (such as 4)specify minimum safety factors for
20、tooth flank andtooth root loading, as well as the design load for tiprelief - one of the main flank modifications, relatedto noise and vibration behavior. These specifica-tions lead to a basic gear design optimized for nom-inal load. However, since noise plays typically amore important role in the p
21、artial load range of awind turbine, the rules for basic gear design mightnot result in a design optimized for noise. However,standard parameters for low noise gear design areoften in conflict with WTG design restrictions, suchas bearing load capacity, overall dimensions andweight.Above consideration
22、s lead to the question: How tooptimize gear design with respect to tonal noisebehavior? In order to define a best practice for low5tonal noise gear design a research project was setup, which is described in the next paragraphs.Setup of the research projectAs mentioned earlier, high speed (gear) stag
23、es(HSS) are an essential part of current WTGs.Therefore the research project as described belowfocuses on HSS with the aim to give an answer tothe following questions:1. Which gear parameters influence tonal noise?a. What are the differences in the gear designof the investigated gearboxes?b. Is ther
24、e a direct link between measurementresults and differences in gear design?2. Is it possible to establish a design method, whichbalances load carrying capacity and noisebehavior, preferably by using state-of-the-artsoftware to analyze the gear mesh undervarious load conditions?a. Definition of simula
25、tion model and availablesoftware.b. Definition of output parameters andcorrelation with tonal noise measurements.3. Which measurement methods can be used toevaluate gear sets on tonal noise behavior?When applied in test rig conditions, are theresults also representative for field conditions?To answe
26、r these questions, a measurementcampaign has been set up on relevant gear sets,using standardized methods for tonal noise determ-ination. Gear design parameters were analyzed incombination with a dynamic simulation of the gearmesh. Ultimate goal was to find a correlationbetween design and simulation
27、 parameters on theone hand and measurement results on the other,both for test rig and field conditions, see Table 1.The gear stages used in the research project aredesigned according to the requirements of ISO/IEC61400-4 for WTG with a nominal power between1.5 MW and 2.5 MW.Table 1. General gear set
28、 parametersSymbol Unit Unit rangeHelix angle 11 - 17.5Normalmodulemnmm 7-8Nominalspeedn rpm 1450 -1900QualityaccordingDIN 3962Q - - 11 does not showtonal noise for 33% and 100% of loading. Behaviorat ground frequency and 1stharmonic is analogueto the two other parameters, overlap ratio andlength of
29、single contact.Improved gear design: geometry, test andDZP calculation resultsBased on the findings described earlier, the HSSgear design of gearbox 5 was changed with the aimto improve tonal noise behavior significantly. Thedifference in gear design can be taken from Table 4.The measured tonal nois
30、e values are shown inTable 5.Table 4. Design parameters of two gearbox 5gear sets basic and improvedGearbox 1(2 )s/mn5 (basic) 1.70 1.61 3.33 10.005 (improved) 1.60 2.09 2.50 11.38Table 5. Tonal noise as measured for basic and improved gear setGearbox33% load 100% loadTonal noisegroundfrequency fz d
31、BTonal noise 1stharmonic2xfzdBTonal noisegroundfrequency fz dBTonal noise 1stharmonic2xfzdB5 (basic) -0.59 -6.73 9.5 - 1 . 85 (improved) -6.78 -7.17 - 1 . 6 - 0 . 2Figure 17. Comparison of12 and tonal noise as measured according ISO/IEC 61400-11 atgear mesh ground frequency15Figure 18. Comparison of
32、12 and tonal noise as measured according ISO/IEC 61400-11 at 1stharmonic of gear mesh ground frequencyFigure 19. Comparison of s/mnand tonal noise as measured according ISO/IEC 61400-11at gearmesh ground frequencyFigure 20. Comparison of s/mnand tonal noise as measured according ISO/IEC 61400-11 at1
33、stharmonic of gear mesh ground frequency16When comparing both gear designs, it can be saidthat the improved design resulted in a significantimprovement of the tonal noise behavior4.Measurement results also revealed a significant de-crease in structure-born noise. These findingsshow a good correlatio
34、n with results from DZPcalculations (Table 6) where also a significantimprovement, especially of the TE, can be seen.Recommendations for low tonal noisegear designThe results stated earlier are based on benchtesting, which is somewhat different from field test-ing. Only few field measurements (accor
35、ding to 2)were available for this investigation which does notallow conclusions based on statistical values.However, by experience a good correlation can befound between standard test rig measurementsusing octave band analysis and field behavior.As a general conclusion it can be said, that the tonal
36、noise behavior of a gear set is largely influenced bygear macro geometry, although trends of various in-fluencing parameters are not completely similar forpartial and full loading conditions (as can be seen inFigure 15).If the gear design were to be optimized only for fullloading conditions, specifi
37、c gear mesh parameterscould be used for low tonal noise design. However,wind turbine gearbox operating conditions are notlimited to only full load, on the contrary, a consider-able amount of operating time is under partialloading conditions.Table 6. DZP results of basic and improved gear setGearbox3
38、3% loading 100% loadingLFzLFz dynTE LFzLFz dynTE5 (basic) 136.8 140.2 0.0047 144.3 150.8 0.00815 (improved) 136.0 140.0 0.0026 145.5 145.6 0.0066Figure 21. DZP calculation results of transmission error at 33% loading (taking into accountshaft-bearing system for the HSS gear set designs of gearbox 5
39、(basic and improved)4Measurements were carried out on the same gearbox and on the same test rig. Only the HSS gear set has beenchanged.17In current gear design practice, basic (nominal)gear parameters are chosen in terms of optimiza-tion of tooth flank and tooth root strength, to allowoptimal use of
40、 available gearbox space. Oncesafety and load distribution criteria are met, the gearset is evaluated regarding noise behavior, applyingflank modifications such as tip relief and crowning.The amount of tip relief is mostly determined bycustomer 1 and other specifications, e.g., thedefinition of the
41、load level for determination of tiprelief is specified, mainly with the aim to preventpremature engagement, which could lead to earlyfailures. As can be seen in Figure 22, for a givenflank modification, the optimum noise and vibrationlevel can only be reached at a certain level of partialloading. Th
42、is point of partial loading requiresthorough investigation for the following reasons:S In partial loading conditions gearbox tonal noisehas a larger impact on overall wind turbine noisebehavior than under full loading conditions. Thisis due to lower rotor noise level at lower blade tipspeed.S In par
43、tial loading conditions effective values ofgear mesh parameters, such as differ fromnominal values due to changes in length ofcontact, caused by flank modifications, toler-ances and the effect of loading. This might havea positive or negative impact regarding noisebehavior.For the noise and vibratio
44、n analysis, as described inthis paper, the program DZP was used to analyzegear set dynamic behavior also for partial loadingconditions. Although calculated parameters usedfor evaluation of tonal noise do not match measuredresults exactly (as can be seen in Figure 9 -Figure 14), trends do match to a
45、certain extent. Asmentioned earlier, the program DZP calculatesparameters for prediction of overall noise andvibration, rather than focusing on tonal noisebehavior specifically. Calculated parameters suchas transmission error, total contact ratio under loadand dynamic tooth forces (including FFT ana
46、lysis)can be used for the evaluation of the noise excita-tion level of the gear stage. From experience, it canbe said that high gear set excitation levels will alsohave their impact on tonal behavior. The trends,visible in Figure 9 and Figure 10 (33% loading con-dition) show a clear link between low
47、 level of TE andlow level of measured tonal noise.Figure 22. Principle development of noise behavior in function of gearbox power illustratedusing DZP parameter tooth force level18Gear set design is a complex task in which a trade-off must be made between load carrying capacityand noise and vibratio
48、n behavior. State-of-the-artsoftware, such as Rikor or LVR - for load and loaddistribution calculation - and DZP - for dynamicanalysis- can support gear designers in finding anoptimum solution.Based on the findings, described in this paper a2-step gear design approach is recommended:S Step 1: Defini
49、tion of the nominal gear geometrywith focus on load carrying capacity, including afirst evaluation of tonal noise parameters as perTable 2.S Step 2: Dynamic analysis with focus on noiseand vibration behavior, using advanced soft-ware such as DZP. Dynamic models should in-clude all parameters affecting the gear mesh,such as bearing stiffness, shaft geometry andflank modifications. Results from this analysisshould be evaluated against known referencegear designs.Concl
