Title: Towards reliable power converters for wind turbines: Field-data based identification of weak points and cost drivers

Transcription

1 Scence & Research Abstract for EWEA 2015, Pars, November 2015 Ttle: Towards relable power converters for wnd turbnes: Feld-data based dentfcaton of weak ponts and cost drvers Author: Dr.-ng. Katharna Fscher, Fraunhofer nsttute for Wnd Energy and Energy System Technology WES, Appelstr. 9a, Hannover, Germany; Conference topc: O&M and logstcs ntroducton Numerous studes have dentfed the power converter as a frequent source of falure n varable-speed wnd turbnes (cf. [1]-[4]). Ths s n lne wth the experences made by wnd-turbne operators worldwde, who state the lmted converter relablty to be a consderable drver of mantenance cost and downtme. However, the development of remedal measures s hndered by the fact that lttle s known about the causes and mechansms underlyng the converter falures. Comprehensve research has been carred out on the thermal- and power-cyclng nduced falure mechansms known to be lfe-lmtng n GBT-based converters n other applcatons (see e.g. [5]-[12]): the lft-off or fatgue-damage of the bond wres, and the fatgue of de-attach or baseplate solder jonts. However, the results of a frst study on the root causes of converter falure [13] suggest that those mechansms play a mnor role n wnd turbnes and emphasse the mportance of a feld-experence based approach to the problem. On ths background, a research cluster for enhancng power-converter relablty n wnd turbnes has been started n Germany [14]. n ths cluster, numerous companes jon forces wth Fraunhofer nsttutes and academa n order to move from suspected falure causes to clear evdence and, n the next step, to effectve countermeasures. The project consortum ncludes wnd turbne and converter manufacturers, converter-component supplers, wnd-turbne operators, mantenance servce provders, an nsurance and companes specalsed n measurng and montorng technology. Besdes an extensve root-cause analyss, whch s based on comprehensve feld-data analyss, drected measurement campagns and post-mortem analyss of faled converter components, the research subjects of the cluster nclude the condton montorng of converters and fault-tolerant generator-converter systems, wth the overall objectve to enhance the relablty and maxmse the avalablty of power converters n wnd turbnes. Ths presentaton / paper presents results of the statstcal feld-data analyss carred out wthn the research cluster descrbed above. Approach The objectve of the followng analyss s to dentfy the predomnantly falng components wthn the power-converter system as well as the man cost drvers, ncludng both the repar cost and the revenue losses resultng from converter unavalablty. n ths way, the work ams to provde a bass for drectng subsequent research to the most crtcal components of the converter system. For that purpose, the analyss makes use of a data set of mantenance and operatng data that ncludes repar-cost and downtme nformaton for each falure event. The results presented n the followng are based on a fleet of 103 wnd turbnes located n 11 onshore wnd parks n Germany, spannng n total months of wnd-turbne operaton durng All turbnes are equpped wth doubly-fed nducton generators (DFG) and partally-rated GBT-based low-voltage converters (two-level back-to-back

2 voltage source converters). The fleet conssts of turbnes of three manufacturers. The commssonng dates of the turbnes range from 1999 to 2007, and the turbnes have got rated capactes n the range of kW. n contrast to the precedng study descrbed n [13], the complete converter system s taken nto consderaton. Based on falure descrptons and nformaton on used spare parts contaned n the mantenance records, the falures of the converter are classfed accordng to the followng categores: - phase module (ncludng GBT modules and correspondng drver boards, DC-lnk capactors, busbars) - converter control board - crowbar - coolng system - semconductor fuse - man crcut breaker - grd-couplng contactor - other Note that wthn the scope of ths analyss, only faults requrng on-ste repar and the consumpton of materal / spare parts are consdered as falures (.e. faults remeded e.g. by a remote reset or by cleanng components are not ncluded). Because phase modules are typcally replaced as complete unts, the data does not allow a further localsaton of the defect nsde the phase modules. The average falure rate of each converter-component category s calculated accordng to f 1 1 N X T (1) wth N denotng the number of falures of the component n the tme nterval, X descrbng the number of wnd turbnes reportng to the database n tme nterval, and T beng the duraton of the tme nterval. All falure rates are gven n falures per wnd turbne and year. Note that calendar tme s used for the calculaton,.e. turbne downtme s not excluded. Correspondngly, the average repar cost c rep and average downtme t down arsng from each component category per turbne and year are calculated usng c 1 rep 1 C rep, X T (2) and t 1 down 1 tdown,. X T (3)

3 Results and dscusson Fgure 1 shows the average falure rates n the dfferent converter-component categores as well as the overall rate of converter falure events. Wth a mean number of 0.21 falures per turbne and year, the phase modules have the hghest falure rate among the consdered component categores. On average, there were 0.51 converter falure events per year on every turbne. Due to the fact that n case of approx. 1/5 of the falure events, components from several categores were replaced to restore the functonalty, the sum of the component falure rates s hgher than the overall converter falure rate. Coolng system Semconductor fuse Man crcut breaker Grd-couplng contactor Converter (total) Falure rate [1/WT/year] Fg. 1: Average falure rates of the converter components Fgure 2 llustrates the percentage dstrbuton of faled components over the dfferent categores. Besdes the phase modules, the semconductor fuses connected to these, the man crcut breaker and the converter control board consttute the largest portons.

4 Dstrbuton of faled components over categores 4% 12% 21% 14% 6% 32% 10% Coolng system Semconductor fuse Man crcut breaker 1% Grd-couplng contactor Fg. 2: Dstrbuton of converter falures over component categores Fgure 3 shows the dstrbuton of repar cost over the component categores. Note that some uncertanty arses from the abovementoned falure of several components n one ncdent, as n those cases the repar cost s a bulk sum and the exact shares for repar of the dfferent defect components are unknown. n these cases, the repar cost correspondng to the respectve falure events s estmated based on the followng assumptons: f multple component categores are affected n the falure event, but the phase module remaned ntact, the repar cost s equally dstrbuted over the concerned categores. n case there are multple affected categores and these nclude a phase module, the cost s dvded at the rato of 90:10 (or 80:10:10 n case of three affected categores), as the cost for replacng a phase module by far exceeds the cost of replacng other components. A smlar procedure s used to estmate the downtme caused by falures n each category (see the downtme dstrbuton n Fgure 4). However, due to the fact that no systematc dfference n the downtmes related to phase-module and other converter falures could be observed n the data, the downtme s assgned to the affected component categores n equal portons for all multple-category falures.

5 Cost dstrbuton over categores 1% 3% 11% 2% 2% 8% 9% 64% Coolng system Semconductor fuse Man crcut breaker Grd-couplng contactor Fg. 3: Dstrbuton of converter repar cost over component categores Downtme dstrbuton over categores 4% 18% 30% 15% 9% 18% Coolng system Semconductor fuse Man crcut breaker 3% 3% Grd-couplng contactor Fg. 4: Dstrbuton of converter-related downtme over component categores n order to assess the economc mpact of the downtme, the average revenue loss resultng from converter unavalablty s estmated. Wth an average rated capacty of P r = 1.66 MW per turbne and assumng a sales prce of electrcty of C el = 85 /MWh as well as a capacty factor of cf = 0.18, the mean converter-related downtme of 23.1h translates nto an annual revenue loss of approx. 600 per wnd turbne and year accordng to c P cf t C. rev. loss r down el (4)

6 Ths can be compared wth the average repar cost due to converter falure per turbne and year of approx calculated usng Eq.(2), see Table 1. Table 1: Economc mpact of converter falure through downtme and repar cost Average downtme (t down) Assocated revenue loss (c rev.loss) Repar cost (c rep) 23.1 h/turb./a 588 /turb./a 3596 /turb./a Conclusons Wthn the man converter systems of the analysed fleet of wnd turbnes wth DFG and partally-rated converter, the phase modules stand out wth a hgh falure rate. Wth 0.21 falures per turbne and year, ths s n a smlar order of magntude as the value of falures/wt/a obtaned for GBT-module falures n the converters of DFG turbnes n [13]. Partcularly outstandng, however, s the economc mpact of phase-module falures due to ther hgh repar cost: Approx. 64% of the annual cost for converter repar s caused by falures of the phase modules. The economc mpact of the repar cost s found to be consderably hgher than that resultng from the turbne downtme. n summary, the phase modules can be concluded to be both the weak pont n terms of relablty and the man cost drver n the consdered converter systems. Ths suggests that future research should focus partcularly on clarfyng the root causes and developng relablty-enhancng solutons for ths component. Learnng objectves The man objectves of ths talk / paper are: - stressng the mportance of an feld-experence based approach for enhancng the relablty of power converters n wnd turbnes - steppng from subsystem-level relablty analyss to component-level analyss - dentfyng the weak pont and man cost drvers wthn the converter system, n order to focus subsequent research accordngly Abbrevatons DFG GBT WT Doubly-fed nducton generator nsulated gate bpolar transstor Wnd turbne

7 References [1] F. Spnato, P.J. Tavner, G.J.W. van Bussel, and E. Koutoulakos, "Relablty of wnd turbne subassembles," ET Renewable Power Generaton, vol. 3, no. 4, pp , 2009, DO: /et-rpg [2] J. Bueno Gayo, Fnal Publshable Summary of Results of Project RelaWnd, Project number , 2011, onlne: [3] K. Fscher, F. Besnard, L. Bertlng, "Relablty-Centered Mantenance for Wnd Turbnes Based on Statstcal Analyss and Practcal Experence," EEE Transactons on Energy Converson, vol.27, no.1, pp , 2012, DO: /TEC [4] J. Carroll, A. McDonald, D. McMllan, Relablty Comparson of Wnd Turbnes Wth DFG and PMG Drve Trans, EEE Transactons on Energy Converson, DO /TEC [5] M. Bartram,. von Bloh, R.W. De Doncker, Doubly-fed machnes n wnd-turbne systems: s ths applcaton lmtng the lfetme of GBT frequency-converters?, Proc. of the 35th Annual EEE Power Electroncs Specalsts Conference, pp , Aachen, Germany, June 20-25, [6] F. Fuchs, A. Mertens, "Steady state lfetme estmaton of the power semconductors n the rotor sde converter of a 2 MW DFG wnd turbne va power cyclng capablty analyss," Proc. of the EPE th Europ. Conf. on Power Electroncs and Applcatons, Brmngham, UK, Aug Sept. 1, [7] M. Musallam and C.M. Johnson, mpact of dfferent control schemes on the lfe consumpton of power electronc modules for varable speed wnd turbnes, n Proc. of the EPE th Europ. Conf. on Power Electroncs and Applcatons, Brmngham, UK, Aug Sept. 1, [8] L. We, R.J. Kerkman, R.A. Lukaszewsk, H. Lu, and Z. Yuan, Analyss of GBT power cyclng capabltes used n doubly fed nducton generator wnd power system, EEE Transactons on ndustry Applcatons, vol. 47, no. 4, 2011, DO: /TA [9] H. Wang, K. Ma, and F. Blaabjerg, Desgn for relablty of power electronc systems, n Proc. of the ECON th Annual Conference on EEE ndustral Electroncs Socety, EEE, pp , Montreal, Canada, Oct , 2012, DO: /ECON [10] E.E. Kostandyan and K. Ma, Relablty estmaton wth uncertantes consderaton for hgh power GBTs n 2.3MW wnd turbne converter system, Mcroelectroncs Relablty, vol. 52, no. 9 10, pp , Sep. 2012, DO: /j.mcrorel [11] S. D Arco, T.M. Undeland, M. Bohlländer, J. Lutz (2012): A Smplfed Algorthm for Predctng Power Cyclng Lfetme n Drect Drve Wnd Power Systems, th nternatonal Mult-Conference on Systems, Sgnals and Devces (SSD), Chemntz, Germany, March 2012, DO: /SSD [12] Ke Ma; M. Lserre, F. Blaabjerg, T. Kerekes (2015): "Thermal Loadng and Lfetme Estmaton for Power Devce Consderng Msson Profles n Wnd Power Converter," EEE Transactons on Power Electroncs, vol.30, no.2, pp , Feb. 2015, DO: /TPEL [13] K. Fscher, T. Staln, H. Ramberg, J. Wenske, G. Wetter, R. Karlsson, T. Thrnger, Feld-experence based rootcause analyss of power-converter falure n wnd turbnes, EEE Transactons on Power Electroncs, vol. 30, no.5, pp , 2015, DO: /TPEL [14] Webste of the Fraunhofer nnovatonscluster Lestungselektronk für regeneratve Energeversorgung (nnovaton cluster on power electroncs for renewables),