Abstract
Automated in-plant diagnostic testing of prestressed concrete railroad crossties is now within reach due to recent progress in robust surface strain measurement techniques. The newly developed non-contact Laser Speckle Imaging (LSI) technique has been shown to provide rapid and accurate surface strain profile measurement, which is a key requirement for rapid transfer length assessment. Accurate determination of transfer length is critical for maintaining continuous production quality in the modern manufacture of prestressed concrete railroad crossties.
Conventional assessment of transfer length generally presumes the underlying existence of a bilinear prestressing force distribution and a corresponding bilinear surface strain profile. Furthermore, it is well-known that this bilinear profile is smoothed due to the effects of finite gauge length during the process of measuring surface strain. In addition, recent extensive crosstie measurements in concrete railroad tie plants have shown significant departures from this simple bilinear profile, which bring to question the general validity and reliability of the traditional 95% AMS (95% Average Maximum Strain) method. Deviations from the simple bilinear profile shape were shown to be partially due to the non-prismatic shape of typical concrete railroad ties. In addition, extensive comparisons between predicted and measured surface strain profiles on numerous crossties suggest that the underlying strain distribution for crossties is best represented by an exponential strain profile, with an asymptotic approach to the fully-developed compressive strain. This is in contrast with extensive testing of prisms with fixed cross-section and fixed prestressing wire eccentricity, for which the surface strain appears to be best represented by the simple bilinear strain profile.
Clearly, departures from non-prismatic behavior have added complexity to transfer length measurement. If accurate and reliable measurements of this important quality control parameter are to be realized, these issues of transfer length uncertainty need to be addressed. This paper provides an experimental comparison of several possible alternative transfer length assessment procedures, in an attempt to answer important uncertainty questions which need to be addressed if rapid real-time transfer length is to be achieved. It is shown that in spite of considerable differences in the transfer length processing methods, and significant departures from prismatic behavior, the averaged results are in many cases consistent with the simple bilinear underlying strain profile assumption. Bias in the measurement of crosstie transfer length due to non-prismatic behavior will also be investigated in this paper.