The design of observations for monitoring Horizontal deformations of engineering structures and the analysis of subsequent data

Abstract

The outstanding problems that pose a challenge to the successful monitoring of the deformations of engineering structures are discussed. A new approach to the design of observations leading to a detailed analysis of the data is developed to monitor fine horizontal movements. First, the troublesome aspects of deformation monitoring when fine movements are sought, are pointed out. It is shown that disparities and limitations in the relevant geometrical and physical concepts pertaining to deformation monitoring work against the idea of integrated analysis. The message that comes out at this point is that different sources of deformation data should be analyzed separately in order to remove the statistical misgivings associated with the analysis of hybrid systems. Second, the level of accuracy desired in structural deformation monitoring and how that accuracy is exposed to damage by the presence of undetected systematic errors are examined. Since such errors may easily attain magnitudes matching those of the point displacements sought, the consequence of the give-and-take process of least squares adjustment on the required point displacements is discussed. The discussion reveals that there is reason to doubt the appropriateness of the use of the concept of networks, coordinates and least squares adjustment in the extraction of fine movements. Such a doubt provides justification for adopting a different perspective of investigating structural deformation phenomena. Third, the establishment of a different approach begins with the formulation of deformation hypotheses. This is followed by the presentation of the principles of design and analysis of observations. The design of observations takes into account the fact that the displacements sought in structural deformation monitoring are at the margin of measuring errors and undetected systematic errors. Separation of the displacements from such errors requires a considerable amount of statistical control. In an attempt to assert this statistical control. a randomization technique has been incorporated. The technique, which basically involves carrying out the observations in a random order, guarantees inferential validity in the presence of the fluctuations in the atmospheric factors during the observation period and enhances the reliability of the estimates of errors in the resultant point displacements. The fine movements are revealed by verifying specific hypotheses regarding the deformation of a structure. Towards this end, the magnitude of a structural displacement at any point is determined from the field observations involving that point only - thereby eliminating the influence of undetected systematic errors in other observations relating to other points. Evidence is provided to the effect that the give and-take process of least squares adjustment may lead to incorrect estimates of small structural displacements in the presence of measuring errors and undetected systematic errors of about the same magnitude. The proposed approach localizes the influence of such errors and, therefore, yields better estimates of point displacements under such circumstances.