Gas turbine blade failure is a crucial issue, not regarding blades themselves, but concerning the collateral effects. Actually, broken pieces are brought into the machine by the working fluid flow and they impact against the other blades, both of rotor and stator, causing a severe damage to the whole system. This kind of machine damage can easily reach billion of dollars, which the losses due to the lost energy production must be added to. Another issue regards the cost of a single blade, typically thousands of Euros each. Thus, it is convenient to take advantage of the whole blade lifetime. For this reason, inspection campaigns are performed for detecting crack and assessing whether the component is able to continue its work or it is necessary to replace it. Whenever a critical crack is detected inside a turbine blade, it is hypothesized to remove the region of material surrounding it and to replace the removed material with another one deposited by welding, in order to improve the lifetime of said turbine blade. The base material, which blades are made of, is René80, a nickel-based superalloy characterised by a coarse grain structure, whereas the welded material would be Inconel625, a nickel-based superalloy suitable for undergoing welding processes. Regarding the issues explained about gas turbine blades, the thesis work here discussed presents models capable of describing crack propagation within René80 and Inconel625, making possible to estimate the residual working time of the blade when a crack of a certain length is detected within. In order to define such analytical models, the characterisation of Inconel 625 is performed, and an alternative definition for Delta J is presented to describe accurately the variable behaviour of René80. Concerning about René80, the analytical propagation model is compared with the experimental tests that were run for material characterisation previously. The cyclic model developed for Inconel625 is compared with the experimental results and a FEM model is also developed; this possibility is given by the fine grain structure that makes the isotropic material hypothesis valid for Inconel. Then, a distribution of initial crack defects is identified in order to estimate the typical flaws present in both Inconel and René. In addition to this, analytical equivalent initial flaws in Inconel625 are compared to experimental evidences visible on specimen fracture surface to validate the built analytical models, whenever possible. Eventually, starting from the estimated flaws within each material, the built model is applied to foresee the expected life of dummy components supposed to contain the extracted flaw populations.
La rottura delle palette di turbine a gas durante l'esercizio è un problema rilevante, non tanto per palette stesse, ma per i danni collaterali. Infatti, i pezzi staccatisi dalla paletta vengono trasportati all'interno della macchina dal flusso del fluido di lavoro e, impattando sulle altre palette, statoriche e rotoriche, provocano un danneggiamento globale alla macchina. Incidenti simili provocano danni per milioni di euro, a cui va aggiunta la perdita per mancata produzione di energia. Un ulteriore problema riguarda la paletta stessa, poichè il suo costo raggiunge facilmente le migliaia di euro. E' quindi conveniente utilizzare la paletta il più a lungo possibile. Perciò cambiare una data paletta nel momento in cui viene rilevata la presenza di una cricca al suo interno richiede un importante sforzo economico, ma attraverso una serie di campagne di ispezione è possibile monitorare lo sviluppo della cricca stessa e valutare le condizioni operative della suddetta paletta. Per incrementare la vita utile della paletta, con conseguente risparmio economico, si ipotizza di rimuovere la regione della paletta interessata dalla cricca e successivamete ripristinare la regione stessa mediante saldatura. Il materiale di base delle palette è René80, una superlega di Nickel, caratterizzata da una granulometria grossolana, mentre il materiale d'apporto per la saldatura è Inconel625, anch'esso una superlega di Nickel, che presenta buona saldabilità. Per ovviare ai problemi descritti in precedenza riguardo le palette di turbine a gas, questa tesi propone dei modelli per la valutazione della propagazione di cricche per entrambi i materiali, rendendo possibile una stima della vita residua di un componente soggetto a cricche. Per lo sviluppo di questi modelli, è stato necessario effettuare la caratterizzazione dell'Inconel625 e ridefinire la formulazione del Delta J per tenere conto della variabilità nel comportamento del René80. Riguardo al René80, il modello di propagazione analitico viene comparato con i risultati sperimentali. Per quanto riguarda l'Inconel625, invece, il modello analitico viene confrontato anch'esso con i risultati sperimentali e, inoltre viene sviluppato un modello FEM, la cui fattibilità è data dalla granulometria fine dell'Inconel che rende possibile trattare il suddetto materiale come macroscopicamente isotropo. Successivamente, viene stimata una distribuzione di difetti iniziale per definire una grandezza caratteristica delle difettosità presenti in ciascun materiale. Nell'Inconel625 la distribuzione ottenuta viene confrontata con i difetti effettivamente osservabili sulla superficie di frattura per la validazione del modello di propagaione qui sviluppato. Infine, partendo dai difetti stimati in ciascun materiale, i modelli di propagazione definiti vengono applicati per la stima di vita di componenti fittizi che presentino al loro interno le suddette distribuzioni di difetti.
LCF crack propagation models for repaired turbine blades
SANGALLI, FILIPPO MARIA;MONTI, STEFANO
2015/2016
Abstract
Gas turbine blade failure is a crucial issue, not regarding blades themselves, but concerning the collateral effects. Actually, broken pieces are brought into the machine by the working fluid flow and they impact against the other blades, both of rotor and stator, causing a severe damage to the whole system. This kind of machine damage can easily reach billion of dollars, which the losses due to the lost energy production must be added to. Another issue regards the cost of a single blade, typically thousands of Euros each. Thus, it is convenient to take advantage of the whole blade lifetime. For this reason, inspection campaigns are performed for detecting crack and assessing whether the component is able to continue its work or it is necessary to replace it. Whenever a critical crack is detected inside a turbine blade, it is hypothesized to remove the region of material surrounding it and to replace the removed material with another one deposited by welding, in order to improve the lifetime of said turbine blade. The base material, which blades are made of, is René80, a nickel-based superalloy characterised by a coarse grain structure, whereas the welded material would be Inconel625, a nickel-based superalloy suitable for undergoing welding processes. Regarding the issues explained about gas turbine blades, the thesis work here discussed presents models capable of describing crack propagation within René80 and Inconel625, making possible to estimate the residual working time of the blade when a crack of a certain length is detected within. In order to define such analytical models, the characterisation of Inconel 625 is performed, and an alternative definition for Delta J is presented to describe accurately the variable behaviour of René80. Concerning about René80, the analytical propagation model is compared with the experimental tests that were run for material characterisation previously. The cyclic model developed for Inconel625 is compared with the experimental results and a FEM model is also developed; this possibility is given by the fine grain structure that makes the isotropic material hypothesis valid for Inconel. Then, a distribution of initial crack defects is identified in order to estimate the typical flaws present in both Inconel and René. In addition to this, analytical equivalent initial flaws in Inconel625 are compared to experimental evidences visible on specimen fracture surface to validate the built analytical models, whenever possible. Eventually, starting from the estimated flaws within each material, the built model is applied to foresee the expected life of dummy components supposed to contain the extracted flaw populations.File | Dimensione | Formato | |
---|---|---|---|
2016_12_Monti_Sangalli.pdf
non accessibile
Descrizione: Testo della tesi
Dimensione
90.6 MB
Formato
Adobe PDF
|
90.6 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/131309