Geosynthetics 2007 January 2007, Washington, D.C. 1 Image Analysis for QC/QA of Geotextile Deformation during Wide Width Tensile Testing by Ahmet H. Aydilek1, M. Emin Kutay2, Members ASCE, Rita Sparacino1, and Hiruy Dafla1 Abstract A laboratory testing program was undertaken to evaluate the feasibility of using digital image analysis technique to track deformation of geosynthetics during tensile testing. The techniques developed in this study offer great potential to identify localized deformations in geosynthetics. The technique can be used in understanding the grip efficiency, quality control of manufactured products or determination of boundary conditions in geosynthetics under tensile loads, which may further be used in development of constitutive models. Introduction and Experimental Program Geosynthetics are expected to offer certain mechanical properties that will provide satisfactory performance when exposed to field conditions. Among various properties, strength is one of the most important ones in reinforcement applications, and wide-width tensile test is commonly used to determine strength (Koerner 1997). Primarily, the stress-strain behavior and strength properties determined from this test are defined at a particular strain or elongation level and strains are usually calculated on average basis for the entire specimen. The accurate determination of the deformation (therefore strain) zones is necessary particularly for quality control/quality assurance (QC/QA) evaluation of these materials. Due to limitations in the current test methodologies, these zones usually remain undetected in the wide-with tensile testing which results in incomplete characterization of mechanical performance. There are various methods used to measure strains, i.e., extensometers, strain gages, laser beam and infrared sensors, but with certain shortcomings as they cannot accurately define complete strain fields. Disruption of yarns or filaments on a geotextile by use of extensometers and strain gages is common, while all devices including the laser beam and infrared sensors, give only average strain along a selected gage length on specimen surface. Image-based methods, often termed as optical flow or particle tracking techniques, have the potential to define the strain zones in a geosynthetic during the tensile testing. As these techniques are non-invasive, they do not suffer from the 1 Assistant Professor, and former Undergraduate Research Assistant, respectively, 1163 Glenn Martin Hall, University of Maryland College Park, Maryland, 20742. E-mail: [email protected] 2 Senior Laboratory Manager, Turner-Fairbank Highway Research Center-FHWA, 6300 Georgetown Pike Rm. F210, McLean, VA 22101.Geosynthetics 2007 January 2007, Washington, D.C. 2shortcomings of the existing strain determination methods and do not disturb the specimen. Moreover, once the image-based method is developed, the model has the potential to be used as a QC/QA tool during geotextile manufacturing and beyond. A laboratory test program was employed in the current study to make automated image-based measurements of strain in geosynthetics during wide width tensile testing. Specimens were tested using both roller and pneumatic grips to identify the effects of clamping. The testing plan included one nonwoven, two low strength woven, and five high strength woven geotextiles. The woven geotextiles had a range of manufacturing styles including monofilament, multifilament and yarn filaments (Table 1). The dimensions of the specimens were selected in accordance with the ASTM D 4595. The selected strain rate was 11%/min for specimens tested in the hydraulic grips. A strain rate of 12%/min was utilized for the specimens, when roller grips were used for clamping. During testing, a monochrome camera was mounted apart from the test setup and simultaneously captured digital pictures of the test specimens at 2 or 5 seconds intervals depending on the rate of displacement applied during testing. The image frames Table 1. Properties of geosynthetic specimens tested Wide Width Tensile Strength (kN/m) Type Geotextile ID Polymer StructureMass per unit area (g/m2) At 5% strain Ultimate Nonwoven Geotextile NW NP, PP 278 NA NA W1 MF, PP 190 11.4 39.5 Low Strength Woven Geotextiles W2 MF, PP 250 13.2 47.4 HW1 FY, PP 284 19.8 47.3 HW2 MU, PET 290 15.8 70 HW3 FY, PP 490 35.1 70.2 HW4 FY, PP 578 43.9 105.1 High Strength Woven Geotextiles HW5 MU, PET 1907 316.0 632 Note: NW = Nonwoven geotextile, W = Low strength woven geotextile, HW = High strength woven geotextile, NP= Needle-punched, PP = polypropylene, PET = polyester, MF = monofilament, MU = multifilament, FY = fibrillated yarn, NA = Not analyzed.Geosynthetics 2007 January 2007, Washington, D.C. 3were saved onto the hard disk and subsequently used for image analysis of strain distributions developed within the specimens. Two image-based particle tracking methods, BMAD and normalized cross-correlation (NCC), were employed to define the time-dependent axial and lateral strains in geosynthetics. Kutay et al. (2006) describes the detailing of the testing procedure and image analysis methodology. Results Preliminary investigations indicated that both algorithms provided highly comparable results with cross-head displacements (Figure 1). This was highly encouraging. As the displacements measured by both algorithms were also comparable, BMAD was used for remainder of the tests. Figure 1 Image-based versus cross-head displacements The image based analysis successfully identified localized strains in geosynthetics. One primary advantage of the image-based methodology was to define the axial as well as lateral strains. This is obviously not available through cross-head displacement determination method generally used in wide-width testing. Figure 2 presents the lateral versus axial strains in selected geotextiles. The clear advantage is that an estimate of lateral strain in a geotextile at a given axial strain (e.g. design strain) can be made through these plots. These lateral strains can be significant in testing of some geotextile, such as nonwovens. Figure 3 presents the observed strains as well as displacements for the nonwoven tested in this study. 020406080100120020406080100120BMADNCCImage-based displacement (mm)Cross-head displacement (mm)Line of equalityHydraulic gripsGeosynthetics 2007