Session: 08-06-01 Internal Flows & FIV

Paper Number: 105119

105119 - Dynamic Small-Scale Riser Model Experiments: A Physics-Based Algorithm to Recover Lost Measures From Optical Tracking Systems 

One of the main aspects of structural monitoring concerns data acquisition quality assurance. In a recent experimental campaign, the displacements of reflective targets placed along the lengths of three small-scale riser models were acquired with the use of an underwater optical tracking system. The experiments were specially designed to directly assess the dynamics of intake riser models subjected to external flow, internal flow and motion imposed at their top.

Most of the targets were correctly tracked by the optical system during the experimental runs. However, some targets were lost by the optical tracking system, when excessively large motions took place. Moreover, clearly spurious measures were obtained from a few targets, during short intervals of time.

Aiming at providing a better set of data for post-processing analysis, a methodology was developed to recover the lost signals by using the acquired data. This could have been made by means of various approaches, for example, performing a simple linear interpolation from the closest measures. However, just adopting a simple interpolation technique, without considering the underlying physics behind the origin of the data, would cause significant errors, which might lead to erroneous conclusions. To avoid that, physical arguments should be incorporated to define acceptable interpolations.

The obvious underlying physics behind the experimental measures is the mechanical behaviour of the structure as a flexible body. Considering that the response of the structure along its length may be simply seen as a weighted sum of its modes of vibrations, with the weights varying with time, the objective of the recovery algorithm is to define the values of such weights at each time instant using the uncorrupted measures.

To obtain the weights, the Least-Square Method (LSM) is then used to fit the trial function composition to the measured data, at each instant of time. However, the presence of the spurious measures can affect the weights obtained with the LSM. An automatic correction scheme, a weighted LSM (WLSM), is then applied, weighting the data to be retrieved proportionally to the inverse of the difference between the measured values and the reconstruction made with the first application of the LSM, in a recursive process. This approach “filters” the effects of spurious data, minimizing them. The proposed algorithm was tested for the experimental campaign, providing very good data recovery, aligned with the expected physical behaviour of the structure and with the observed behaviour. The algorithm also proved itself to be computationally fast, being able to generate the reconstructed data with small processing time.

Although the basic theory involved in the algorithm is well-known and detailed in the literature, to the best of the authors' knowledge there is no methodology established for data recovering or automatic removal of the influence of spurious measures obtained from optical tracking systems, with proper physical guidance, as posed in this paper. Another gain from the proposed algorithm is that the conceived procedure allows for an enhancement of the obtained experimental data. In fact, its output is a continuous function, physically representative of the obtained measurements, and generates smoother spatial representations of the phenomena, simplifying further analyses.

Presenting Author: Celso Pesce University of São Paulo

Presenting Author Biography: Professor of Mechanical Sciences.

Authors:

Guilherme J. Vernizzi Universidade de São Paulo
Vitor Schwenck Franco Maciel University of São Paulo
Wagner A. Defensor Filho University of São Paulo
Renato M. M. Orsino University of São Paulo
Guilherme R. Franzini University of São Paulo
Celso P. Pesce University of São Paulo