3. Modeling A

3.1. Characteristics of the mesh

Number of knots: 149

Number of meshes and types: 40 elements QUAD8

3.2. Tested sizes and results

Identification of a \(m\) common to the three experimental bases and a \({\sigma }_{u}\) per base.

Temperature \([°C]\)	Reference		Calculated values
			MAXI_VRAI		REGR_LINE
	\(m\)	\({\sigma }_{u}\) \([\mathrm{MPa}]\)	\(m\)	\({\sigma }_{u}\) \([\mathit{MPa}]\)	\(m\)	\({\sigma }_{u}\) \([\mathit{MPa}]\)
—50	24	2800	26.7	2536	16.1		4983
—100	24	2700	26.7	2428	16.1		4642
—150	24	2600	26.7	2372	16.1		4555

3.3. notes

Using the maximum likelihood method, although the difference between the values of (\(m\), \({\sigma }_{u}\)) obtained with RECA_WEIBULL and their reference values remains not negligible, it is in accordance with the result sought considering the relatively small number of samples used for the calibration (15 per temperature). To obtain the reference values it would be necessary to considerably increase the number of samples per temperature (\(N>1000\)). However, the difference observed remains reasonable (of the order of 10%). In addition, the growth of \({\sigma }_{u}\) as a function of temperature is respected.

With the linear regression method, the difference is greater.