Automated wind turbine maintenance scheduling

Au oma ed Wind Tu bine Main enance Scheduling Nu seda Y. Y¨u ¨u¸sena, Paul N. Rowleyb, Simon J. Wa sonc, Julio Mele oa,∗ aIns i u o Uni e si a io de In es igaci´on CIRCE (Uni e sidad de Za agoza - Fundaci´on CIRCE), C/ Ma iano Esquillo 15, 50018, Za agoza, Spain bCREST, Loughbo ough Uni e si y, Holywell Pa k, Loughbo ough, LE113TU, UK cDUWIND, Del Uni e si y o Technology, Kluy e weg 1, 2629 HS Del , Ne he lands Abs ac While many ope a ion and main enance (O&M) decision suppo sys ems (DSS) ha e been al eady p oposed, a se ious esea ch need s ill exis s o wind a m O&M scheduling. O&M planning is a challenging ask, as main enance eams mus ollow speci ic p ocedu es when pe o ming hei se ice, which equi es wo king a heigh in ad e se wea he condi ions. He e, an au oma ed main enance p og amming amewo k is p oposed based on eal case s udies conside ing a ailable wind speed and wind gus da a. The me hodology p oposed consis s on inding he op imal in e en ion ime and he mos e ec i e execu ion o de o main enance asks and was buil on in o ma ion om egula main enance isi asks and a co ec i e main enance isi . The objec i e is o ind possible schedules whe e all wo k o de s can be pe o med wi hou b eaks, and o ind ou when o s a in o de o minimise e enue losses (i.e. doing main enance when he e is leas wind). Fo he DSS, ou ine main enance asks a e g ouped using he indings o an agglome a i e nes ing analysis. Then, he ask execu ion windows a e sea ched wi hin p e-planned main enance day. Keywo ds:Wind Tu bine, O&M, Main enance, Scheduling 1. In oduc ion The cos o main enance is a majo con ibu o o he o al le elized cos o ene gy (LCOE) om wind a ms accoun ing o a sha e o a ound ∗Co esponding au ho : [email p o ec ed] P ep in submi ed o Reliabili y Enginee ing & Sys em Sa e y Ma ch 20, 2020 20%-25% [1]. Minimisa ion o he main enance cos s equi es as p ecise as possible main enance scheduling. Ex ended down ime, which can occu when main enance in e en ions a e delayed du ing poo wea he , incu s inancial cos s o wind a m owne s. Wind a m ope a ional scheduling has been ound in he li e a u e o be a unc ion o a ange o ac o s such as he ene gy demand [2], elec ici y ma ke p ice and wind speed [3]. When he cons ain s a e in es iga ed, wind a m accessibili y no mally comes in i s place in e ms o impo ance. Fa m accessibili y depends on he a iabili y o he wind speed and he as- socia ed heal h-sa e y and en i onmen egula ions (HSE) [4–9]. In o he wo ds, inding an app op ia e wea he window is a majo c i e ion o any ype o in e en ion o wind u bines and he e is a esea ch need closing he gap be ween academic models and applica ion in p ac ice [10]. Acco ding o he li e a u e, main enance wea he windows a e dependen on he wind speed o an onsho e wind a m, while he wa e heigh is also a decisi e ac o o o sho e wind a ms whe e accessibili y depends on he ype o main e- nance essel u ilized [11–14]. Fo o sho e wind a m ope a ions, he loca ion (dis ance o sho e and wa e dep h), me eo ological and oceanog aphic a i- ables in luence he si e accessibili y, i is highligh ed in he li e a u e ha he e is a end o mo ing om nea -sho e o deep wa e o o sho e wind a m ins alla ions, which esul s in lowe si e accessibili y and highe cos s o he execu ions o co ec i e main enance ac ions [15]. In addi ion o he measu ed mean wind speed, indus y p ac ice high- ligh s wind gus as an impo an pa ame e when conside ing access o a wind u bine [16]1. Howe e , hus a , his has no been e e ed o in he li e a u e conce ning scheduling s udies as a cons ain which a ec s ei he ope a ional scheduling o down ime. P e ious academic wo k in his ield has conside ed only wind speed and ou pu powe as he decisi e pa ame e s when gene a ing a easible main enance plan in onsho e [18, 19] and signi i- can wa e heigh , wa e peak pe iod and wind speed in o sho e [7, 15]. I is al eady no ed in bo h onsho e and o sho e c ane manuals and sa e wo king guides ha wo king heigh and wind gus speed in luence execu ions o c ane ope a ions [20, 21]. When a c ane ope a ion canno be pe o med, he co - 1In his s udy, wo majo wind u bine manu ac u e s’ O&M guidelines a e used. These wo companies a e also leading o iginal equipmen manu ac u e s in he wind sec o . Ac- co ding o 2017 s a is ics, he o iginal equipmen manu ac u e s o wind u bine ha e he highes ma ke sha e among he wind a m O&M se ice p o ide s [17]. 2 esponding wind u bine main enance ac ion can also no be pe o med as well and esul s in delay o O&M ac ions. This delay con ibu es o wea he ela ed down ime, which is esul ing om coa se main enance planning and insu icien accessibili y o bo h wind a m si e and wind u bine componen . In he p esen s udy, he au ho s conside bo h mean wind speed and wind gus as limi ing ac o s o accessibili y o an onsho e wind u bine and demons a e he applica ions o wind speed measu emen s in de e mining ask execu ion sequence whils minimising down ime due o ad e se wea he condi ions du ing pe iods o in ended main enance. The goal is au oma ed scheduling o asks o be pe o med wi hin a wo kday, such ha asks wi h s ic equi emen s a e scheduled when condi ions a e mos benign. The no - mal p ac ice depends on wo weeks ahead main enance se ice eam booking wi h a single call en ailing which ala m is ac i a ed o which u bine. These wo k o de s a e lacking de ailed planning o he main enance day and he asks o be pe o med. The e o e, he e a e coa se planning and wea he ela ed wai ing pe iods in he wind a m si e. The s uc u e o his pape is as ollows: in he ollowing sec ion, an o e iew o a ypical wind a m main enance policy is desc ibed. Gene al cha ac e is ics o mean wind speed, wind gus , main enance log books and ask comple ion du a ion da a a e p esen ed in Sec ion 3. The nex sec- ion desc ibes he me hodology, and he p oposed amewo k. Case-s udies a e hen p esen ed o show he alue o a p oposed main enance planning me hodology, which demons a es how an op imal sequence o main enance in e en ions can be de ised. Then, in Sec ion 6, p ac ical explana ion o he indings, he limi a ions and he assump ions a e p esen ed. The inal sec ion summa ises he main ou comes o his s udy. 2. Main enance plans & p oblem s a emen Wind u bine main enance can consis o bo h co ec i e and p e en i e ac ions. Long e m main enance policies mus co e bo h o hese. Co ec i e main enance is no mally ca ied ou once a aul has been de ec ed, whe eas, p e en i e main enance is gene ally pe o med acco ding o calenda -based p e-de e mined in e als such as biannual, annual, biennial and quinquennial pe iods [4, 5, 22–24]. The numbe o asks and he du a ion o a scheduled main enance ac ion a e di e en om one case o ano he and depend on speci ic sub-assembly, componen s, manu ac u e , model and capaci y o he wind u bine. Scien i ic li e a u e and manu ac u e s’ main enance guides 3 gi e igu es o he equi ed du a ion o a ange o main enance asks ha a y om lub ica ion which ypically akes ew hou s o o he mo e leng hy which las up o 18 hou s [23, 24] du ing a biannual main enance isi . In addi ion, wo king p ac ices may di e om one ope a o o ano he . These ac o s mus be aken in o accoun in e ms o de ining a comp ehensi e main enance s a egy and p o ide a challenge when de eloping a model o main enance op imisa ion. (a) 4 (b) Figu e 1: Main enance scheduling p ocedu e (a) p e en a i e policy and (b) co - ec i e in e en ion. P e en a i e main enance is usually planned a yea in ad ance on an annual basis o onsho e wind a ms [25]. A ypical low diag am o his ype o ad anced planning is shown in Figu e 1a whe e accoun needs o be aken o he wea he and elec ici y ma ke p ices as well as he a ailabili y o a main enance eam [16]. Requi emen s o p e en a i e main enance can also be seen e en when planning co ec i e ac ions as shown in Figu e 1b. Bo h, egula and co ec i e main enance in ol e unce ain ies, pa icula ly conce ning he wea he ela ed limi a ions. The ypical limi ing ac o o execu ing main enance ac ions is he wind speed. Regula ions and manu ac- u e s’ good p ac ices se he maximum alues o he wind speed which allow wo k a di e en loca ions on he u bine. This in o ma ion has been used in p e ious esea ch wo ks o de elop main enance amewo ks. Fo example, one s udy ixed he wind speed limi as 10 m/s o accessing he whole u - bine [5], while ano he based he sa e wo king limi on cu in wind speed, i.e., he u bine was only conside ed main ainable when he wind speed was lowe han cu -in [6]. Fu he mo e, cu en egula ions and main enance guides in- clude dynamic sa e y limi s aking in o accoun no only he mean 10-minu e wind speed alue bu also he gus alue, when a c ane usage is equi ed o such a case like majo componen eplacemen . The de ini ion o gus is a sho -du a ion (seconds) maximum o he luc ua ing wind speed [26]. The maximum pe missible wind gus speed o c ane usage depends on 5 a ious ac o s such as mean wind speed, in e en ion heigh and weigh o he load [20]. The e o e, co esponding wind gus es ic ion o any in e - en ion equi es imely and case based con ols. Mo eo e , high gus alues cause mo e es ic i e wind u bine componen speci ic accessibili y ules e- ducing he highes allowed mean wind speed. Taking in o accoun only wind speed limi s, he sa e wo king ules a e also di e en depending on u bine model and size. Fo example, in he case o MADE AE 46 u bines, p e en a i e main enance equi es wind speeds below 20 m/s a he nacelle, howe e changing he whole nacelle equi es he wind speed o be no mo e han 5 m/s. I we check he equi emen s o NEG Micon NM 52 u bines, wo king in he hub equi es wind speeds below 15 m/s while wo king in he nacelle oo is allowable un il 12 m/s and gene a o alignmen should no be pe o med o wind speeds abo e 10 m/s. Finally, o he Ves as V 90 3.0 MW model, gene a o alignmen in e en ion can’ be done o wind speeds abo e 8 m/s, changing pi ch angle equi es wind speed alues smalle han 6 m/s and wo king in he d i e ain is allowed up o 7 m/s [16]. Wi hin a wo k shi , a ious asks mus be comple ed on a wind u bine acco ding o he p e ailing ime and labou o ce es ic ions. As s a ed in [11], i is almos impossible o gene a e a lawless main enance plan in e ms o a oiding p oduc ion loss, since i is di icul o ind a pe iod whe e he u bine is no p oducing due o low wind speeds. Wha can be done in his sense is o schedule he main enance wi h an accep able unce ain y [27, 28]. 3. Da a sou ce, wind a m main enance p ocedu e and da a Main enance logs, se ice wo k o de s and SCADA da a we e ob ained om a Spanish wind a m. In his analysis, O&M se ice epo s, which co e a 3 yea s window, a e used o de ine he lis o ac ions and he needed du a ion o each ype o in e en ion and ac i i y in he s udied wind a m. Rega ding he me eo ological da a, 10-minu e wind speed and wind gus da a a e collec ed o yea 2019. In he inal analysis, accessibili y in es iga ions o 24 hou s windows a e p o ided o he example cases. Acco ding o he in o ma ion ga he ed, he a e age du a ion o he biannual, annual, biennial and quinquennial isi s a e app oxima ely 21, 26, 15 and 18 hou s espec- i ely. The o al numbe o di e en asks o be pe o med in main enance isi s is 169. Mos o hem, 117, a e included in he biannual isi ac ions while he o he s a e dis ibu ed o e he o he isi s. Howe e , no all main- 6 enance ac ions a e ca ied ou du ing each planned isi as some o hem ha e p io i y based on he indings o p e ious se ice isi s and he needs o he wind u bine. Table 1: Execu ed asks o he scheduled isi Tu b. Wo k Zone Sub Sys em Task Numbe s A-G ound Towe 1 o 2 A-G ound Elec ical Pa s 3 A-G ound Ro o -Blades 4 B-Pla o m Elec ical Pa s 5 o 7 C-Towe Yaw Sys em 8 o 14 D-Nacelle Main Sha and Bea ing 15 o 17 D-Nacelle Gea box 18 o 27 D-Nacelle Gene a o 28 D-Nacelle Base S uc u e and Co e 29 o 31 D-Nacelle Elec ical Pa s 32 E-Hub Ro o 33 o 34 F-Ou side o Nacelle Senso s 35 o 36 Figu e 2: Example o u bine wo king zones Figu e 2 shows he conside ed u bine wo king zones, while he ask num- be s associa ed o hese zones a e lis ed in Table 1. In his wo k, o he sake 7 o simplici y, only he asks numbe s lis ed in Table 1 a e used o de ine a case s udy conside ing a egula se ice isi . A second case s udy is based on a majo in e en ion, which equi es a c ane usage. Mo e speci ically, a gene a o eplacemen is s udied and mo e in o ma ion will be p o ided e- ga ding he co esponding ask. To explain he wo king en i onmen o he se ice pe sonal o pe o ming ei he a egula se ice o a majo in e en- ion, he seasonal and gene al cha ac e is ics o he subjec wind a m a e shown in Figu es 3 and 4. In Figu e 3, he wind speed seasonal his og ams om he case Spanish wind a m a e p esen ed. The annual his og am is included in each g aphic o highligh he seasonal con ibu ion. I can be seen ha he majo i y o wind speed obse a ions lie be ween 0 and 10 m/s in summe mon hs. Then, summe looks he bes season o main enance ac ions, bu he e a e s ill a signi ican numbe o wind speed obse a ions wi h alues highe han 15 m/s. Figu e 3: Annual e sus seasonal wind speed his og ams using 10-minu e a e aged mean wind speeds Figu e 3 shows he seasonal cha ac e is ics o he nacelle wind speed ob ained om he analysed wind a m. I is known ha he seasonal wind 8 speed beha iou is dependen on he loca ion o he wind a m. The annual main enance plan mus be p epa ed conside ing he seasonal wind beha iou and he elec ici y ma ke p ices o he coun y whe e he analysed wind a m is loca ed. Then, he seasonal wind beha iou is an impo an ac o o long e m scheduling, which is no he aim o his s udy. The esul ing p og am om he annual main enance plan is an inpu o decision making suppo ool. The e o e, his inpu mus be modi ied, when he analysed wind a m is changed. Figu e 4 illus a es he diu nal beha iou o he wind speed o each season du ing 2019 compa ing he maximums eco ded in hou ly da a pe seasons. I can be seen ha he day shi (08:00 o 18:00) in summe , wi h wind speed maximums lowe han 20 m/s, indica es ela i ely easonable wind a m accessibili y o pe o m a main enance isi . Figu e 4: Seasonal wind speed ends as hou ly maximums. This igu e is ob ained calcula ing he maximums pe hou o each day o e a season in 2019. The window, which is shaded in yellow ep esen s he day shi om 08:00 o 18:00. The majo i y o scheduled main enance in e en ions a e planned in sum- me and au umn mon hs in he case s udy main enance log. Fo his eason, 9 In dend og am isualisa ion, heigh ep esen s he alue o he Euclidean dis ance be ween clus e s. To es ima e his dis ance, inpu da a mus be scaled. As an example, o an inpu consis ing o 100 ows and 2 columns, he i s column indica es he wo king zone and he second one s ands o he co esponding wind speed es ic ion. A e scaling he inpu da a each obse a ion is i s ly assigned o a empo a y clus e . Following his p oce- du e, in he i s s ep he e exis 100 clus e s (100 asks) and, o ins ance, he Euclidean dis ance be ween Clus e 1 and Clus e 2 can be ob ained as; Eucdis =q(HSE C1−HSE C2)2+ (WZC1−WZC2)2(5) whe e HSE ep esen s he wind speed es ic ion and W Z ep esen s he wo king zone. The same calcula ion is epea ed o all 100 clus e s. A e - wa ds, he Wa d algo i hm g oups hese clus e s acco ding o he minimisa- ion p inciple o Euclidean dis ances. 4.4. P oposed amewo k The p oposed me hodology is g aphically explained in Figu e 7. The ini ial s ep is o p o ide in o ma ion on he ype o he in e en ion, ini ial sa e wo king ules and wind o ecas s. Then, i is equi ed o decide i wind gus measu emen s and es ima ions a e needed as decision a iables. The co esponding answe depends on he speci ic equi emen s o he planned in e en ion, such ha in e en ion may equi e a c ane usage. 16 Figu e 7: Flowcha o p oposed solu ion, HSE: Heal h-Sa e y and En i onmen egula ions In he p oposed me hodology iis he use de ined limi o ini ia ing he agglome a i e nes ing p ocess, as shown in Figu e 7. He e, we assumed ha a main enance ask can be done wi hin a minimum o ou s ages such as: 17 access o wo king a ea, access o ailed componen , emo e ailed componen and placemen o he new componen . Then, o a case ha each s age e- qui es a unique sa e wo king ule, he minimum o al numbe o sa e wo king ules is 4. The e o e, p ede e mined compa ison alue, i, is se o 4 . Fo an in e en ion consis ing o mo e han ou asks o equi ing he ul ilmen o mo e han ou sa e y ules ela ed o wind speed, o ecas s mus be used along wi h he ou comes o he agglome a i e nes ing as inpu in he sea ch p ocess. The gus o ecas s a e necessa y i he in e en ion is pe o med using a c ane, which equi es educ ion o he wind speed limi s due o he high gus alues. Las ly, he sea ch p ocess scans he a ailable ime windows du ing he in ended main enance day o ind he op imal ime window o he wo k o ake place. I he main enance in e en ion can be execu ed du ing he p e-planned day, op imal execu ion ime and o de o he asks a e de e mined. I no , a change in he p e-planned day is sugges ed. This me hodology can also be used o o sho e applica ions, bu i is e y impo an o upda e HSE equi emen s conside ing wa e heigh and o sho e ope a ions speci ic ules. Mo eo e , in e en ion ype, equi ed du- a ion, ou pu s o annual main enance, e c. mus be upda ed conside ing he echnology ype and he wo king en i onmen . 5. Resul s The ials wi h he p oposed app oach o wo dis inc main enance isi s a e epo ed in his sec ion. Case 1 is an applica ion es o a ou ine main- enance isi , whe eas Case 2 ocuses on a majo componen eplacemen . 5.1. Case 1: Rou ine Main enance 5.1.1. Clus e ing The p oblem o planning a high numbe o asks is simpli ied by applying he agglome a i e nes ing me hodology o he pool o 36 asks. Clus e ing was pe o med using he Euclidean dis ance as simila i y c i e ion. I was calcula ed using he wind speed limi and he co esponding u bine wo king zone o each ope a ion. Figu e 8 shows how he asks a e g ouped o ming a o al numbe o 4 clus e s ( ep esen ed wi h di e en colo s) as a unc ion o he es ic ions, wind u bine wo king zone and wind speed. 18 Figu e 8: G aphical ep esen a ion o he clus e ing p ocess. Di e en colou s ep esen he di e en clus e s o asks. (The dend og am needs o be egene a ed o di e en echnologies conside ing main enance in e en ion lis s.). A summa y o he clus e ing esul s is gi en in Table 2 whe e he clus e du a ion and i s wind speed limi a e shown. As main enance asks a e usually accomplished by wo echnicians, which will equi e hal he ime, and he equi ed esolu ion o he planning schedule is based on 10-minu e s eps, he ounded du a ion pe pe son on a 10-minu e scale is also p o ided. Table 2: Clus e ing esul s Clus e Du a ion (min) Pe pe son (10 mins) lim (m/s) 1 66 4 20 2 106 6 15 3 491 25 12 4 50 3 10 5.1.2. 24 hou s e alua ion o execu able/no execu able windows Now by applying he p ocedu e, explained in Sec ion 4.2, wi h measu ed wind speed da a o es days ( he summe day was 27 h June 2019 and he au umn day, 08 h No embe 2019) execu able and no execu able pe iods o he main enance clus e s a e de e mined. Figu e 9 and Figu e 10 show he allowed in e en ion s a ing imes o each o he clus e s ound in he p e ious sec ion. 19 Execu ion o he main enance se ice is only possible, i he s a ing ime o he in e en ion is wi hin he g een do s. He e g een do s ep esen alid pe iods o bo h wind speed sa e wo king limi and he a ailabili y o a win- dow o accomplish he ask wi hin i s minimum equi ed comple ion du a- ion. In hese igu es, lim ep esen s wind speed limi and Du s ands o he equi ed du a ion o he execu ion o he co esponding clus e . Figu es 9 and 10 a e gi en in o de o display he complexi y o p og am- ming wi h dynamic wea he es ic ions. The decision make mus conside all he in e en ion speci ic accessibili y windows and gene a e a main enance p og am combining hem. Figu e 9: Rou ine main enance e alua ion wi h ac ual inpu da a o he summe day (a) clus e 1, (b) clus e 2 (c) clus e 3 (d) clus e 4 20 Figu e 10: Rou ine main enance e alua ion wi h ac ual inpu da a o he au umn day (a) clus e 1, (b) clus e 2 (c) clus e 3 (d) clus e 4 The esul s o each o he clus e s we e: Clus e 1: Tasks a e execu able du ing bo h analysed days, since he co esponding wind speed es ic ion is e y lexible and i s du a ion is ela- i ely low, see Figu es 9a and 10a. Clus e 2: Tasks a e mos ly execu able o bo h days, see Figu es 9b and 10b. Al hough, he e a e sho non-execu able windows in he au umn day, see Figu e 9b. Clus e 3: Tasks a e execu able o he calme summe day and asks a e non-execu able o he windie au umn day, see Figu es 9c and 10c. Clus e 3 asks a e he mos challenging g oup, because hey equi e a longe ime wi h majo wind speed es ic ions. Clus e 4: The execu ion o Clus e 4 asks depends mos ly on he mos es ic i e wind speed limi . Ne e heless, i can be seen ha he e exis some execu able windows, since he execu ion o his clus e equi es he lowes du a ion, see Figu es 9d and 10d. These p elimina y analysis shows ha in he summe day all asks can be pe o med, whe eas in he au umn day he e is no sui able ime window o pe o m Clus e 3 asks. The e o e, in he nex analysis only he esul s ob ained om he summe day a e p esen ed. Conside ing he hou ly elec ici y ma ke p ice, i is possible o combine 21 i wi h he ene gy losses o each plan, es ima ed om measu ed wind speed alues and manu ac u e ’s powe cu e, o ob ain he e enue p io i ised decision pools. Figu e 11 shows he day-ahead elec ici y ma ke p ices o 27 h June 2019 and 08 h No embe 2019 [35]. I can be clea ly seen ha in Summe , 27 h June 2019, he elec ici y ma ke p ices highe han 50 EUR/MWh, while in Au um, 08 h No embe 2019, mos o he p ices a e be ween 30 and 40 EUR/MWh, ollowing he common end obse ed in he spanish ma ke [35]. Figu e 11: Day-ahead hou ly elec ici y ma ke p ice Figu e 12 shows he co esponding e enue losses o he p io i ised main- enance plans, labelled as “low” when hey a e below he mean o he e enue losses es ima ed o he day unde conside a ion. When hey a e lesse han he hi d qua ile and g ea e han he mean, he label is “medium”. Las ly, o he plans wi h he e enue losses g ea e han he hi d qua ile, he label is “high”. This DSS is p epa ed as a compu a ional ool and he isualisa ion o he epo ing module is gi en in he Appendix 1, whe e he al e na i e plans and he e enue e alua ion p ocedu e a e exempli ied. 22 Figu e 12: Decision pool o ou ine main enance isi scheduling o he summe day, he second y axis s ands o he g ouping acco ding o he e enue losses, yellow shaded window shows he day shi . In Figu e 12, only selec ed al e na i e main enance plans a e plo ed when a cos -wise clea sepa a ion can be obse ed among he 1093 al e na i es (14 o 1093) o he analysed summe day. Each al e na i e ep esen s a p og am, which con i ms ha wea he ela ed down ime is minimised. Acco ding o Figu e 12, he ea ly hou s o he day a e mo e p e e able in o de o pe o m p e en a i e in e en ion conside ing he e enue losses. Al hough elec ici y ma ke p ices a e high du ing hese hou s, he limi ed wind esou ce a ail- abili y, educes powe p oduc ion losses and co esponding e enue losses. 5.2. Case 2: Gene a o eplacemen 5.2.1. 24 hou s e alua ion o execu able/no execu able windows In his case s udy, he gene a o eplacemen is in es iga ed o he p o- posed scheduling p ocess. To eplace he gene a o , a c ane mus be used. 23 Fi s ly, he nacelle co e mus be emo ed and hen he ailed gene a o mus be aken ou . These emo als a e ollowed by ins alla ion o he new gen- e a o and e-ins alla ion o he o iginal nacelle co e . In o he wo ds, his in e en ion equi es wo ypes o li ing /unloading asks. Sa e y equi e- men s wi h ega ds o wind speed a y due o he gus alues. The mean wind speed limi o sa e wo king has o be dec eased by 2 m/s when he wind speed gus is abo e 5 m/s o ope a ion equi ing a c ane usage [16], om 10 m/s, o a gus lowe han 5 m/s, o 8 m/s o a gus highe han 5 m/s in he case o nacelle co e and om 8 m/s o 6 m/s, o he same gus alues, in he case o he gene a o . I is wo h men ioning he e ha he gus limi , o he au ho s knowledge, has ne e been conside ed in p e ious scien i ic s udies. Ano he di e ence, ega ding ou ine main enance plan, is he equi emen o ollow a ixed ask o de , as ob iously, i would no be possible o pe o m emo al o old gene a o be o e emo ing he nacelle co e . The e o e, he main enance execu ion o de is ixed o his p oblem. The ob ained esul s o a co ec i e main enance isi in he p e iously selec ed Summe day a e shown in Figu e 13. Execu able (g een) and no execu able ( ed) ime windows a e shown o he ou main asks o a co ec- i e in e en ion. He e, a 120 minu es window is sea ched o he emo al o he old gene a o and ano he 120 minu es window o placemen o he new one. In hese sea ches, he pe missible wind speed educes om 8 m/s o 6 m/s, when he wind gus alue exceeds 5 m/s. The emaining asks equi e a 90 minu es window sea ch o he emo al o he nacelle co e and ano he 90 minu es o he placemen . In hese sea ches, he pe missible wind speed educes om 10 m/s o 8 m/s, when he wind gus alue exceeds 5 m/s. 24 Figu e 13: E alua ion o gene a o eplacemen when conside ing dynamic sa e wo king limi s due o gus (Summe day). I is a he easy o highligh he impac o he gus a iable wi h a simple compa ison be ween Figu e 13 and Figu e 10. Due o he gus ela ed es ic- ions, a co ec i e in e en ion canno be pe o med in his case, al hough i was possible o pe o m a p e en a i e main enance in e en ion. 6. Discussion and Limi a ions When a decision make uses only he mean wind speed cha ac e is ics, any day om he summe season is a good candida e in o de o p epa e he main enance plans. This s udy p esen ed ha each candida e day mus be analysed p o oundly. Because, while he powe losses esul ing om he main enance in e en ions could be limi ed, he e enue losses could be se- e e due o he elec ici y ma ke p ices and ice e sa. As i is shown in his s udy, no only he mean wind speeds, bu also he wind gus s a e he limi ing ac o s o pe o ming some majo main enance ac i i ies. The im- plemen a ion o o he en i onmen al limi ing ac o s ( og, ain, e c.) was no possible due o da a una ailabili y. The p ac icali y o such a DSS highly depends on he inpu da a. Unce - ain ies in ega ds o du a ion o asks and in ela ion o wea he o ecas s a e no conside ed in he p esen s udy. I mus be no ed ha in o de o 25