Question About Testing Approach

My view is that you should use the same specimen.
To follow your own example, if a crack started on test X.
a) It's failed anyway.
b) It may not be noticed if it is microscopic in which case test Y making it propagate will be a good thing as it will make it visible.
Conversely if test X does no damage then proceeding with test Y won't be a problem.
Again this is just my furry opinion.
Del

No, that is at least misleading, you would be making up the results, because you will always have a specimen that has not been tested in at least two axes, and you just assigned to it, the results from another.

Just think: what if the DUT was about to fail on an axis that was not tested?

Yahlasit

Testing one component in all three axes simultaneously is significantly cheaper than a dead bird.

Thank you for the insight.

Designing an appropriate fixture would be a reasonable approach; but I was once told that it would be just pseudo-three-axis test and you need actual stimulation on all axes in order to test the actual condition. But there seems to be another problem; there is random vibration testing involved and if you do random test in three axes simultaneously, your resulting vector direction will constantly change because the magnitude in each axis is random...

It is random in the three axes but it is generated by a source and the source is usually coherent in the three axes. This means that the vector will not change very much its direction.

From an other point of view the need to verify in 3 axes depends on the fastening method between components and board. The board compliance is maximal (and the critical frequency minimal) in the direction normal to the board plane. In same direction the stiffness of the fasteners is maximal. In this direction the component can be considered as part of the board as a local added mass. In the other directions the fasteners have a higher compliance but the combination between component mass and stiffness leads normally to higher critical frequency.

Up to you how you want to make your tests but I would discuss with the specificator since a combined test can bring more insight. 3 combined tests (at different angles and frequencies spectra) can bring more than 3 tests with only one direction. So that you have same number of tests but more knowledge at same cost!

nick name is ofcourse absolutely right but in view that you have only a one axis shaker that might be out of the question. So, if you have been able to define a representative load spectrum separated in the three axes then you have to perform these consecutive tests on the same sample and preferrably do this on a set of samples. The test is a simulation (or representation) of actual load on the component and applies load cycles to it which contribute to the overall fatigue loading of the component. At the end of the test the component will have been fatigue damaged by these accumulated load cycles. This is ofcourse assuming that loads in all axes operate in the finite life regime of the component. If loads in a specific axis operate in the infinite life regime then you may get away with completely skipping the test for this axis.