22.12.2012 Views

Forgeabilité des aciers inoxydables austéno-ferritiques

Forgeabilité des aciers inoxydables austéno-ferritiques

Forgeabilité des aciers inoxydables austéno-ferritiques

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

tel-00672279, version 1 - 21 Feb 2012<br />

118 Chapter IV. STRAIN PARTITIONING<br />

IV.4.3.4 Per-phase average strains<br />

In Figure IV.21, two different polygonal lines joining microgrid intersections are drawn and represent<br />

two possible ways of defining the interface between ferrite and austenite, including or not part of the<br />

neighbour phase. The real interface is somewhere in between. Per-phase average deformation can be<br />

computed using the procedure <strong>des</strong>cribed in section IV.4.3.2 applied to the two different polygonal lines<br />

mentioned previously and defined in Figure IV.21. Different results are obtained if the line “with or<br />

without interface” is used. The average of both values provi<strong>des</strong> a good estimate of the exact values.<br />

20 μm<br />

δ-ferrite<br />

γ-austenite<br />

γ-austenite withinterface<br />

γ-austenite without interface<br />

Figure IV.21. Integration scheme for the average gradient over a phase<br />

IV.4.3.5 Errors on the computation of the strains<br />

There are several elements that induce some errors on the strain calculation. In this section, only a<br />

few of these elements are considered, more details can be found in [98] and [128].<br />

� To estimate the error related to the manual location of the microgrid intersections, the same<br />

region of interest has been analyzed two times and the results given in each case have been<br />

compared. The comparison of the results showed that the error related to the manual location<br />

of the microgrid intersections could be neglected.<br />

� The main source of errors is due to the bad approximation of the interface, as shown by the<br />

difference in the area of the integration domains in each case, which generally is close to 15%<br />

of the area of the characterized area. In fact, the computations “with interface” for the harder<br />

phase (austenite) takes into account this phase plus some domain of the weaker phase (fer-<br />

rite), more strained, and thus gives upper bounds for the average gradient components; it<br />

leads to lower bounds in case of the weaker phase. For a given phase, the error related to the<br />

bad location of the interface can be estimated by the half difference between the result “with<br />

interface” and without interface”. Note that the average of the result “with interface” of a given<br />

phase and the result “without interface” of the other phase, weighted with the corresponding<br />

areas, gives the overall value.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!