Processing method to improve the reliability of data recorded by an industrial press in describing powder behavior during uniaxial cold compaction

In previous work the data recorded by an industrial press (forces and displacements) were extensively used to describe powder behavior during uniaxial cold compaction. Satisfactory models describing densification were proposed, as the result of the axial and radial stresses acting on the powder column. The recorded data are very precise from an industrial perspective, mainly in the last stage of compaction, directly related to final green density. Nevertheless, they are slightly scattered in the first stage, which is interesting from a scientific perspective, as the step where rearrangement occurs. To overcome this limit, this work proposes a methodology to analyze the signal of forces in the frequency domain to increase the signal to noise ratio. The signal was decomposed through the Fourier transformation, and the noise was smoothed by a low-pass filter specifically designed for the press. Reliability of the data and effectiveness of the derived relationships result significantly improved.