Integrating multi-camera DIC at MBDA

By Floriane Soulas, R&D Engineer at EikoSim

As part of its increasing competence in cutting-edge measurement techniques, MBDA has launched a process of testing and internal auditing of measurement resources by multi-camera DIC (Digital Image Correlation). After testing these measurement tools on coupon-type test campaigns, a new stage has been launched with the testing of this technology on more complex structures using multi-camera systems.

Compressive loading test were carried out on a part named “hood”, made of composite material. For this test, two distinct zones of interest were determined by MBDA. The first is located on the tongue of the cap and the second on the outer face of the arms, respectively called zone 1 and zone 2 (see Fig. 1). The hood is maintained in position by a special tooling which comes to constrain it in a precise position.

multi-camera DIC — Fig. 1: Presentation of the two zones of interest on which the measurements by stereo-correlation of digital images will be performed

The aim of this test is to validate the use of digital image correlation on a structural compressive loading test, with two camera systems to monitor several zones of interest and to compare the results obtained with the numerical simulations carried out beforehand. A second interest of this test is the study of the deployment of the measurement technique in a crowded test environment. This test was also an opportunity to train the MBDA teams in good test practices related to multi-camera DIC and the use of EikoTwin DIC software. In this study, all the data processing on EikoTwin DIC was carried out by the MBDA teams with the support of EikoSim.

Displacement tracking by multi-camera (or multi-view) DIC

Pre study

The test machine on which the hood is tested being in a relatively crowded laboratory, and the cover being itself embedded in a special tooling, the room to operate and place the physical sensors (LVDT, strain gauges) is restricted. These restrictions also reduce the field of view available for the cameras, as well as the space to place them around the part. In order to overcome this problem, EikoSim has carried out a virtual pre-study of the test with the EikoTwin Virtual tool. This tool allows a virtualization of the test scene to choose the position of the cameras before the compressive loading test for the two areas of interest (Fig. 2)

stereo-correlation of digital images — Fig. 2: Position of camera systems with EikoTwin VIRTUAL

Thanks to EikoTwin Virtual, it is thus possible to predict the positioning of the cameras, as well as to determine the adequate size of the speckle pattern relative to the test specifics (taking into account the characteristics of the cameras, the distance between the cameras and the hood as well as the refinement of the mesh). The software provides virtual images as close as possible to those that will be taken for the real test (Fig. 3).

This virtualization then allows, thanks to the virtual images of the test, to determine the minimal uncertainty values that can be expected for the displacement measurement in this configuration. This is the measurement noise of the test, that is calculated under ideal conditions and below which no displacement can be measured. The minimum expected measurement uncertainty for this test, for both areas of interest, is presented in Table 1. These data are interesting because they allow us to predict if the expected measurement precision for a given test is compatible with the expected displacement.

Measurement	Uncertainty (µm)
Average according to X	5.8
Average according to Y	9
Average according to Z	9.8

Table 1: average uncertainty obtained by the pre-study

Performance of the test

Following the test configuration defined during the pre-study, we positioned two pairs of static cameras (1 image/second, 4112×3008 pixels) observing respectively zones 1 & 2.

Fig. 4 : test setup with a multi-camera systems

Thanks to the use of the image acquisition software EikoTwin Vision, images of the two zones of interest were taken simultaneously during the compressive loading test of the hood using the multi-camera DIC system (see Fig. 5). Several types of pressure loadings were carried out by MBDA on the hood.

Results and perspectives

The finite element model of the structure being extremely complex and refined, only the of interest are selected to carry out the measurement. It is on these two zones only (in green on Fig. 6) that the displacements will be measured and that the comparison between test and simulation results will be made.

From the test images, the EikoTwin DIC software measures the displacement field directly at the nodes of the finite element model provided by MBDA, for the two areas of interest. The results of image correlation can thus be compared directly on the mesh with the predictions of the finite element model and the possible physical sensors present during the test.

Fig. 7 shows a comparison of the displacement fields (according to the normal to the surface) obtained between the image correlation (on the left) and the simulation field projected on the part for the study area 1 on the right. It can be seen that the two fields obtained have strong similarities. The measured field appears to be more symmetrical in the displacement distribution around the center of the part, while the simulation predicted stronger and more spread out displacements from the center to the hood tab.

Fig. 7: Normal displacement fields for image correlation (left) and numerical simulation (right) for zone 1

Fig. 8: Z-displacement fields for zone 2

Fig. 8, above, shows the measured displacement field for zone 2 along the main load axis Z. With EikoTwin DIC, it is also possible to directly compare simulation and measurement with the difference fields, which show the deviation between the measured and simulated field in the area of interest.

For zone 2, this result is shown in Fig. 9. It is interesting to note that the difference field here is very noisy, indicating that there are no significant differences between the two fields, or a displacement pattern that would be found in one of the two fields and not in the other. If discrepancies were noted by MBDA between the measurements of the sensors, the multi-view DIC and the simulation, these are in no way significant. The displacement fields will then allow an efficient, precise and more extensive calibration of the simulation than is possible with the physical sensors.

Conclusion

MBDA has invested in the acquisition of cameras to generalize the use of multi-view DIC in its panel of measurement means. In order to facilitate the integration of DIC in their process, EikoSim intervened as a companion to the increase in competence of MBDA’s test services and design office. To do so, the DIC tools (hardware and software) have been tested on different tests of increasing complexity, from samples and specimens to a complete structure tested in multi-camera during a compressive loading test.

A pre-study was carried out in order to prepare the test, the positioning of the cameras and the realization of the speckle pattern, thanks to the EikoTwin Virtual software. This pre-study also allowed to know the average uncertainty of the measurement. Test data were collected on two distinct areas of the hood, which served as a study structure, using a multi-camera system and the EikoTwin Vision software. The images obtained were processed internally by MBDA via EikoTwin DIC, after training of the teams.
These tests are encouraging and provide qualitative and quantitative results in areas of study on which previously only a few physical sensors were applied, allowing them to enrich and optimize their digital model. Today, the method of measurement by multi-camera DIC is implemented at MBDA and applied to various test campaigns conducted internally to validate material characterizations or tests on mechanical structures.