4. Practical advice

Before carrying out a recalibration procedure, it is essential to carefully choose the parameters to be adjusted. This requires reflection on the part of the user in order to enter the right parameters that make physical sense in relation to the study to be carried out.

It is also necessary to perform a sensitivity study of the functional to be minimized in relation to the parameters to be adjusted. Indeed, it is useless to adjust a parameter that does not make the function vary. If the parameter is essential for the study but the functional one is insensitive to this parameter, then it is necessary to find a new functional that is much more suitable.

It is also necessary to limit the number of parameters to be adjusted.

Two main categories of methods can be distinguished.

The first category consists in exploiting the measured quantities directly. This concerns sensitivity-type methods.

The second category consists in having the characteristic properties of the eigenmodes verified on the numerical model, by substituting the calculated eigenmodes by the identified eigenmodes. An expansion of the identified modal deformation is carried out in order to obtain a quantity defined on the numerical model.

The advantage of the first category lies in the fact that the measured data is directly exploited. The disadvantage is that it is necessary to perform a pairing between the measured mode and the calculated mode analogous to each iteration of the readjustment procedure. This technique is therefore not appropriate when the modal density is high.

The advantage of the second category lies in the fact that neither a modal calculation is made on the numerical model during the calculation iterations nor a pairing between the calculated mode and the measured mode. The disadvantage is that it is necessary to perform an expansion of the measurement on the numerical model. This technique is therefore not appropriate when the number of measurement points is reduced. A second iteration of readjustment after updating the support numerical model may be necessary in order to refine the results. Indeed, the updating of the support model makes it possible to improve the quality of modal expansion.

Overall, the following conclusion can be drawn:

• If the modal density is high, exploiting the properties of modal matrices generally gives better results. However, this requires a good distribution of observation points in order to be able to achieve a better expansion of the modal deformation.

• If the modes of the structure are isolated modes, and if the number of observation points is limited, then the sensitivity technique is much more appropriate for adjusting the initial numerical model.