Several issues can be identified:
1 A "frozen process" is only effective if all of the parameters remain virtually constant. The "frozen process" parameters should not have been established using new shot. Shot properties change with use. These include a reduction in average shot diameter and 'conditioning' of the particles. The process parameters should be re-established using a shot load that is 'stable'. A stable shot load is one that is screened regularly and has regular, small, additions of new shot trickled in to replenish shot losses. The new shot additions may, or may not, require to be pre-conditioned.
2 There is a lack of information presented about how the 'intensity' changes are being measured. Are the stated changes for complete saturation curves or are they simply changes in the arc height for a fixed peening time? If the changes are detected using complete saturation curves then which method is employed to determine the 'peening intensity' point from the set of data?
