A mechan­i­cal fail­ure is often the result of a mate­r­i­al fatigue. The nat­ur­al age­ing of parts, notably those made from met­als and alloys, leads to the for­ma­tions of cracks that may prop­a­gate and cause the part to break. And because this may have seri­ous con­se­quences, it is impor­tant to under­stand how cracks form ad evolve in time. 

Fol­low­ing a “4 week Eras­mus stay that final­ly last­ed 4 years”, Chrysan­thi Papamichail, with her engi­neer­ing diplo­ma from the Ecole Poly­tech­nique in Athens (grad­u­at­ed in elec­tri­cal engi­neer­ing and com­put­er sci­ence) and a Master’s degree in applied math­e­mat­ics is the author of a the­sis on cat­a­stroph­ic mechan­i­cal fail­ure she defend­ed at the UTC-LMAC Lab­o­ra­to­ry in a col­lab­o­ra­tive pro­gramme with the UTC-Rober­val Lab­o­ra­to­ry. The the­sis sub­ject she chose was the “Esti­ma­tion of dynam­ic sys­tems with appli­ca­tions in mechan­i­cal engi­neer­ing sci­ences”.

As Chrysan­thi detailed for us, “we asso­ciate sto­chas­tics process­es with dig­i­tal tech­niques to mod­el a crack and deter­mine its prop­a­ga­tion char­ac­ter­is­tics”. By apply­ing a math­e­mat­i­cal for­mu­la, the ran­dom phe­nom­e­non can be recre­at­ed vir­tu­al­ly. “The equa­tions enable us to esti­mate the prop­a­ga­tion of the cracks over a set time inter­val, to iden­ti­fy the moment of cat­a­stroph­ic fail­ure (i.e., rup­ture) and thus to pro­pose an oper­a­tional life expectan­cy for mate­ri­als”. The results obtained then allow us to make com­par­isons with real data and to improve the algo­rithm and also to con­comi­tant­ly reduce the resource needs in terms of com­pu­ta­tion­al power. 

There are some impor­tant appli­ca­tions areas, in numer­ous fields: “My equa­tion is suf­fi­cient­ly reli­able and pow­er­ful per se to be used in any sec­tor that employs met­al parts that could prove dan­ger­ous. The range of pos­si­ble appli­ca­tions runs from build­ing con­struc­tion to aero­space, from rail­roads to health con­cerns about implants. But it can also be applied to nat­ur­al events such as earth­quakes or bio­log­i­cal phe­nom­e­na”, she details. 

For the time being, the accent is on appli­ca­tions in aero­space where acci­dents can lead to death. Nuclear pow­er pro­duc­tion also is an area where there have been sig­nif­i­cant progress, for rea­sons you can well imag­ine “but only a few coun­tries are con­cerned and the data is often shroud­ed in secre­cy” adds Chrysanthi.