Comparison of analysis techniques to obtain modulus and phase angle from sinusoidal test data

Title: Comparison of analysis techniques to obtain modulus and phase angle from sinusoidal test data
Author(s): T. Pellinen, B. Crockford
Paper category : conference
Book title: Sixth International RILEM Symposium on Performance Testing and Evaluation of Bituminous Materials
Editor(s): M. N. Partl
Print-ISBN: 2-912143-35-7
e-ISBN: 2912143772
Publisher: RILEM Publications SARL
Publication year: 2003
Pages: 301 - 307
Total Pages: 7
Nb references: 4
Language: English

Abstract: This paper compares different analysis techniques of computing modulus and phase angle from cyclic sinusoidal data in compressive dynamic (complex) modulus testing. Analyzed data reduction techniques were 1) localized multiple linear regressions for peak picking to define amplitude and phase angle from the peaks averaged over a group of cycles, 2) sinusoidal full waveform curve fitting for a single estimate of amplitude and phase over a group of cycles, and 3) two-step analyses over a group of cycles involving regression across peaks (Spencer's 15-point filtering) for amplitude and phase determination from the central part of the waveform. The computed modulus values were statistically less sensitive to different analysis techniques than the phase angles. The first method was the most robust but predicted phase angles that appeared to be too high at intermediate temperatures. Spencer's 15-point filtering with central waveform bracketing turned out to be a slightly more robust method than sinusoidal regression to obtain stable parameter estimates from the imperfect raw data. However, sinusoidal regression gave parameter estimates almost identical to the estimates obtained from Fast Fourier Transformation (FFT), while the other methods deviated from the FFT estimates. To make conclusive recommendations, an additional study is underway to compare Master Curves constructed using raw data analyzed with different techniques. The performed analysis though suggests that there is a need to limit the deviations of the controlling load waveform from a perfect sine wave to guarantee good quality test data.

Online publication: 2003-03-02
Publication type : full_text
Public price (Euros): 0.00
doi: 10.1617/2912143772.037