Lifetime estimation of IGBT power modules
Author
Nicola, Laura
Term
4. term
Education
Publication year
2013
Submitted on
2013-08-09
Pages
63
Abstract
Effektmoduler er blandt de mindst pålidelige komponenter i elektriske systemer, der arbejder i barske omgivelser. Det er derfor vigtigt at kunne forudsige deres levetid. Denne afhandling undersøger strømforsyningen til et partikelacceleratorsystem, som bruges efter en uregelmæssig, applikationsspecifik profil (missionsprofil). Vi opbygger en elektro-termisk model i PLECS-software for at bestemme chipsenes interne knudetemperatur under drift. Derefter bruger vi rainflow-analyse, en standardmetode til at tælle cyklusser, til at finde middeltemperatur og temperatursving i hver cyklus. Med disse data estimerer vi levetid med en analytisk model baseret på Coffin-Manson-loven, der forbinder gentagne temperaturskift med materialeudmattelse, og vi kvantificerer den akkumulerede skade i transistoren med Palmgren-Miner-reglen, der summerer skader fra mange cyklusser. Til sidst giver vi levetidsforudsigelser for tre driftsscenarier udledt fra en del af missionsprofilen.
Power modules are among the least reliable parts in electrical systems that operate in harsh conditions, so predicting their lifetime is essential. This thesis examines the power supply of a particle accelerator system, which follows an irregular, application-specific pattern (a mission profile). We build an electro-thermal model in the PLECS software to determine the internal junction temperature of the chips during operation. We then apply rainflow analysis, a standard method for counting cycles, to extract the average temperature and the temperature swings in each cycle. Using these inputs, we estimate lifetime with an analytical model based on the Coffin-Manson law, which links repeated temperature cycling to material fatigue, and we quantify the accumulated damage in the transistor with the Palmgren-Miner rule, which sums the effects of many cycles. Finally, we provide lifetime predictions for three operating scenarios taken from part of the mission profile.
[This abstract was generated with the help of AI]
Keywords
Documents