How well can we predict cluster fragmentation inside a mass spectrometer?