<%BANNER%>

Damage Evaluation of Glass Fiber Reinforced Polymer Decks Using Acoustic Emissions

Permanent Link: http://ufdc.ufl.edu/UFE0043840/00001

Material Information

Title: Damage Evaluation of Glass Fiber Reinforced Polymer Decks Using Acoustic Emissions
Physical Description: 1 online resource (75 p.)
Language: english
Creator: Asencio, Rafael
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2011

Subjects

Subjects / Keywords: acoustic -- bridge -- deck -- emissions -- fiber -- glass -- polymer -- reinforcement -- tetsting
Civil and Coastal Engineering -- Dissertations, Academic -- UF
Genre: Civil Engineering thesis, M.E.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Glass fiber-reinforced polymer (GFRP) decks are being used as a replacement for bridge decks, due to their light weight, fast installation time, and high strength. Several different types of deck systems are available for commercial use. The nature of the resin and glass materials of the decks results in a brittle and catastrophic failure mode. Due to this failure mode and the geometry of different deck systems, traditional structural evaluation methods such as deflection and strain may not detect possible damage in the material, which could lead to a failure. In this investigation a non-destructive detection method, acoustic emissions (AE) was used to develop a possible evaluation method for in-service GFRP bridge decks. Lab test setups were designed to replicate the positive and negative bending experienced by an in-service deck system. Three different deck systems were tested in both setups. A testing load procedure was developed based on similar testing of in-service GFRP tanks. The testing consisted of service-level loading of undamaged and damaged samples. Damage was induced on the samples by loading them to capacity in the test setups. AE data were collected during all the loadings as well as strain and deflection data. The strain and deflection data were analyzed and revealed linear behavior in all the samples up to failure. The AE data were analyzed using a structural evaluation method adapted from the Calm ratio and the load ratio, recovery ratio analysis (RRA). Calm ratio is the ratio of a selected feature AE activity during the unloading to the AE activity during the loading. The load ratio is the ratio of the load at the onset of AE activity to the previous maximum load experienced by the specimen. RRA was successful in the lab test and was then used on data collected during a bridge load test of the Hillsboro canal bridge.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Rafael Asencio.
Thesis: Thesis (M.E.)--University of Florida, 2011.
Local: Adviser: Hamilton, Homer R.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2011
System ID: UFE0043840:00001

Permanent Link: http://ufdc.ufl.edu/UFE0043840/00001

Material Information

Title: Damage Evaluation of Glass Fiber Reinforced Polymer Decks Using Acoustic Emissions
Physical Description: 1 online resource (75 p.)
Language: english
Creator: Asencio, Rafael
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2011

Subjects

Subjects / Keywords: acoustic -- bridge -- deck -- emissions -- fiber -- glass -- polymer -- reinforcement -- tetsting
Civil and Coastal Engineering -- Dissertations, Academic -- UF
Genre: Civil Engineering thesis, M.E.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Glass fiber-reinforced polymer (GFRP) decks are being used as a replacement for bridge decks, due to their light weight, fast installation time, and high strength. Several different types of deck systems are available for commercial use. The nature of the resin and glass materials of the decks results in a brittle and catastrophic failure mode. Due to this failure mode and the geometry of different deck systems, traditional structural evaluation methods such as deflection and strain may not detect possible damage in the material, which could lead to a failure. In this investigation a non-destructive detection method, acoustic emissions (AE) was used to develop a possible evaluation method for in-service GFRP bridge decks. Lab test setups were designed to replicate the positive and negative bending experienced by an in-service deck system. Three different deck systems were tested in both setups. A testing load procedure was developed based on similar testing of in-service GFRP tanks. The testing consisted of service-level loading of undamaged and damaged samples. Damage was induced on the samples by loading them to capacity in the test setups. AE data were collected during all the loadings as well as strain and deflection data. The strain and deflection data were analyzed and revealed linear behavior in all the samples up to failure. The AE data were analyzed using a structural evaluation method adapted from the Calm ratio and the load ratio, recovery ratio analysis (RRA). Calm ratio is the ratio of a selected feature AE activity during the unloading to the AE activity during the loading. The load ratio is the ratio of the load at the onset of AE activity to the previous maximum load experienced by the specimen. RRA was successful in the lab test and was then used on data collected during a bridge load test of the Hillsboro canal bridge.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Rafael Asencio.
Thesis: Thesis (M.E.)--University of Florida, 2011.
Local: Adviser: Hamilton, Homer R.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2011
System ID: UFE0043840:00001


This item has the following downloads:


Full Text

PAGE 1

DAMAGE EVALUATION OF GLASS FIBER REINFORCED POLYMER DECKS USING ACOUSTIC EMISSIONS By RAFAEL ASENCIO A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORID A IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ENGINEERING UNIVERSITY OF FLORIDA 2011 1

PAGE 2

2011 Rafael Asencio 2

PAGE 3

To my wife, Erin, my daughter, Clare and son, Rafi 3

PAGE 4

ACKNOWLEDGMENTS The author thanks Dr. Hamilton, my advisor and princi pal committee member, for his guidanc e and patience duri ng this process and committee members Dr. Consolazio and Dr. Cook. The author acknowledges and thanks the Flor ida Department of Transportation (FDOT) for prov iding funding for this projec t. Much of the laboratory work for this project was conduc ted at the FDOT Marcus H. Ansley Structures Research Center and the author would like to express appreciation to David Allen, Stephen Eudy, Sam Fallaha, Tony Hobbs, Will Potter, Paul Tighe, David Wagner, and Chris Weigly for their assistance and input. Dr. Plunkett and Kansas Structural Compos ites inc. is thanked for the donation of GFRP deck samples. 4

PAGE 5

TABLE OF CONTENTS page ACKNOWLEDG MENTS ..................................................................................................4 LIST OF TABLES............................................................................................................7 LIST OF FI GURES ..........................................................................................................8 ABSTRACT ................................................................................................................... 11 CHAPTER 1 INTRODUC TION ....................................................................................................13 2 BACKGRO UND ...................................................................................................... 15 GFRP Deck Design a nd Fabricat ion .......................................................................15 Pultruded deck panel (Deck A) .........................................................................15 Preform molds deck panel (De ck B) .................................................................16 Wet lay-up deck panel (Deck C) .......................................................................16 GFRP Bridge Deck Stru ctural B ehavior .................................................................. 17 Acoustic Emission ...................................................................................................17 AE fundament als ..............................................................................................18 Equipment and settings .................................................................................... 19 Evaluation methods ..........................................................................................19 Field test ing ......................................................................................................20 3 TEST SETUP AND SPECIMEN DESIGN AND CO NSTRUCTION ........................24 Span and Support Conf iguratio n .............................................................................24 Specimen Design and Construction ........................................................................25 4 INSTRUMEN TATION .............................................................................................29 Strain Gages ........................................................................................................... 29 Load and Displacem ent G ages ...............................................................................30 Acoustic Emissi on Sensor s .....................................................................................30 Disp 16 Sy stem .......................................................................................................31 5 TEST PR OCEDURE ...............................................................................................35 Service Loading (U ST and DS T) ............................................................................35 Ultimate Loading (ULT) ...........................................................................................36 6 ULTIMATE STRENGTH TEST -RESULTS AND DI SCUSSION ..............................38 Deck A Positive B ending (A _P) ..............................................................................38 5

PAGE 6

Deck A Negative Be nding (A _N).............................................................................40 Deck B Positive B ending (B _P) ..............................................................................41 Deck B Negative Be nding (B _N) .............................................................................42 Deck C Positive B ending (C _P) ..............................................................................43 Deck C Negative Be nding (C _N) ............................................................................44 7 ANALYSIS OF AE TEST DATA US ING RECOVERY RATIO ANALYSIS ..............57 8 BELLE GLAD E TEST ............................................................................................. 65 Sensor Loc ations ....................................................................................................65 Relative Ratio Analysis on Belle Glade Brid ge Data ..............................................65 9 SUMMARY AND CO NCLUSION S ..........................................................................70 LIST OF RE FERENCE ..................................................................................................72 BIOGRAPHICAL SKETCH ............................................................................................ 75 6

PAGE 7

LIST OF TABLES Table page 3-1 GFRP sample di mensions (in.). ..........................................................................28 4-1 Strain gage location dimensions positive bending tests (in.) ...........................33 4-2 Strain gage location dimensions negative bending tests (in.). ..........................34 4-3 AE sensor location dimensions positive bending tests (in.). .............................34 4-4 AE sensor location dimensions negative bending tests (in.). ............................34 4-5 AE recording parameter s. ...................................................................................34 5-1 Testing sequence an d rest time. .........................................................................37 5-2 Load profile char acterist ics. ................................................................................37 6-1 A_P sti ffness....................................................................................................... 55 6-2 A_N sti ffness. ..................................................................................................... 55 6-3 B_P sti ffness....................................................................................................... 55 6-4 B_N sti ffness. ..................................................................................................... 56 6-5 C_P sti ffness. ..................................................................................................... 56 6-6 C_N sti ffness. ..................................................................................................... 56 7-1 Average values from RRA and stiffness and strength rati o. ...............................64 8-1 AE sensor identificat ion and coordi nates. ...........................................................69 7

PAGE 8

LIST OF FIGURES Figure page 2-1 Pultruison process ( Ultrafibergla ss.com). ...........................................................20 2-2 Zellcomp deck panel. ..........................................................................................21 2-3 Preform sample beam used by Stru ctural Composites of Melbourne (performs .com)...................................................................................................21 2-4 Structural Com posites de ck panel. ..................................................................... 21 2-5 Wet lay-up glass fiber schedul e (Kalny et al. (2004) ..........................................22 2-6 Fiber reinforced polym er honeycomb deck panel ..............................................22 2-7 Acoustic emi ssions ev ent. ..................................................................................22 2-8 Feature data gathered from a single AE waveform............................................23 3-1 Positive bendi ng test se tup. ................................................................................26 3-2 Negative bending test se tup. ..............................................................................27 3-3 Stud-in-grout connection det ail. .......................................................................... 28 4-1 Strain gage locations for posit ive bending test soffit vi ew. ...............................32 4-2 Strain gage locations for negative bendi ng deck A. ............................................32 4-3 Strain gage locations negative bendi ng test wearing surface view. .................32 4-4 Strain gage locations negativ e bending tests soffit vie w..................................32 4-5 Displacement gage locations positive bending setup A) el evation and B) plan view. ................................................................................................................... 33 4-6 AE sensor locations for positiv e bending tests soffit view. ...............................33 4-7 AE sensor locations for negativ e bending tests so ffit view. .............................33 5-1 UST and DST l oad pr ofile ...................................................................................36 5-2 Ultimate loading test ( ULT) pr ofile. .....................................................................37 6-1 Load-displacement pl ot A_P tests. ..................................................................45 6-2 A_P top plate bu ckling under load. .....................................................................46 8

PAGE 9

6-3 A_P web failure..................................................................................................46 6-4 Load-strain plot A_P ULT. ................................................................................46 6-5 Load-deflection plot A_N te sts. ........................................................................47 6-6 A_N dama ged web. ............................................................................................ 47 6-7 Damaged web location. ......................................................................................48 6-8 A_N ULT load-st rain pl ot. ...................................................................................48 6-9 Load-displacement pl ot B_P tests. ..................................................................49 6-10 B_P test cra cking patte rn. ..................................................................................49 6-11 B_P cracks at failure load. ..................................................................................50 6-12 Load-strain plot B_P ULT. ...............................................................................50 6-13 Load-deflection pl ot B_N te sts. ...........................................................................51 6-14 Shear failure ov er support B_N. ...................................................................... 51 6-15 Shear failure location B_N. .............................................................................. 51 6-16 Load-strain plot B_N ULT. ..................................................................................52 6-17 Load-displacement plot C_P tests. ..................................................................... 52 6-18 Debonded z one C_P. ......................................................................................... 53 6-19 Load-strain plot C_P ULT. ..................................................................................53 6-20 Load-displacement plot C_N tests. ..................................................................... 54 6-21 Debonding fa ilure C_ N. ......................................................................................54 6-22 Load-strain plot C_N ULT. ..................................................................................55 7-2 Specimen B_P AE data unfilt ered. ..................................................................... 61 7-3 Specimen B_P AE f iltered AE da ta. ....................................................................62 7-4 Filtered data used in AE ra tio analysis calc ulations. ...........................................62 7-5 Average DST and UST values for A) positive and B) negative bending tests. ...63 7-6 RRA plot for positive bending tests all specimens ..........................................63 9

PAGE 10

7-7 RRA plot for negative bending tests all s pecimens ..........................................64 8-1 Bridge site A) aerial phot o and B) detailed site pl an. ..........................................67 8-2 AE sensor and strain gage lo cations on the deck panel s. ..................................68 8-3 AE and influence ratio calculati ons for LC 4 fr ont wheel ....................................68 8-4 AE data from br idge te st LC4. ............................................................................68 8-5 Sample RRA results for bridge load test. ............................................................69 10

PAGE 11

Abstract of Thesis Pres ented to the Graduate School of the University of Florida in Partial Fulf illment of the Requirements for the Ma ster of Engineering DAMAGE EVALUATION OF GLASS FIBER REINFO RCED POLYMER DECKS USING ACOUSTIC EMISSIONS By Rafael Asencio December 2011 Chair: Trey Hamilton Major: Civil Engineering Glass fiber-reinforced polymer (GFRP) dec k s are being used as a replacement for bridge decks, due to their light weight, fast in stallation time, and high strength. Several different types of deck systems are available for commercial use. The nature of the resin and glass materials of the decks results in a brittle and catastrophic failure mode. Due to this failure mode and the geometry of different deck systems, traditional structural evaluation methods such as def lection and strain may not detect possible damage in the material, which could lead to a failure. In this investigation a nondestructive detection method, acoustic emissi ons (AE) was used to develop a possible evaluation method for in-service GFRP bridge decks. Lab test setups were designed to replicate the positive and negative bending ex perienced by an in-service deck system. Three different deck systems were tested in bo th setups. A testing load procedure was developed based on similar testing of in-service GFRP tanks. The testing consisted of service-level loading of undamaged and damaged samples. Damage was induced on the samples by loading them to capacity in the test setups. AE data were collected during all the loadings as well as strain and deflection data. The strain and deflection data were analyzed and revealed linear behavior in all the samples up to failure. The 11

PAGE 12

12 AE data were analyzed using a structural evaluation method adapted from the Calm ratio and the load ratio, recovery ratio analysis (RRA). Calm ratio is the ratio of a selected feature AE activity during the unloading to the AE activity during the loading. The load ratio is the ratio of the load at the onset of AE activity to the previous maximum load experienced by the specimen. RRA was successful in the lab test and was then used on data collected during a bridge lo ad test of the Hills boro canal bridge.

PAGE 13

CHAPTER 1 INTRODUCTION Florida has the largest inventory of moveabl e bridges in the nation, with a total of 148, of which 91% are bascule, 7% are swing and 2% are lift bridges. Most employ open grid s teel decks as a riding surface for par t of their span (National Bridge Inventory 2008). Compared to solid brid ge decks, steel grid decks have several advantages: they can be assembled in the factory, they are li ght weight, and they are easy to install. Unfortunately, worn steel grid decks have hi gh maintenance costs and provide poor skid resistance, especially when wet. As a possibl e option to replace the worn steel grid decks, the Florida Department of Transportation (FDOT) is investigating the possibility of using glass-fiber reinforced polymer (GFRP) decks. GFRP deck panels can be designed and manufactured to meet weight and dimensional requirements of a bridge, allowing direct replacement of steel grid decks GFRP bridge decks are relatively new to the bridge industry. One concern regarding GFRP deck systems is their durab ility and field performance. Developing Non-Destructive Evaluation (NDE) methods that can be used to monitor the GFRP decks is important to ensure long-term per formance is monitor ed and documented. My work focuses on the use of acoustic emission (AE) to inspect GFRP decks. AE has been used extensively to assess the stru ctural integrity of GFRP pressure tanks and vessels. Testing of these vessels has been conducted using ASTM standards. These standards were used as a starting point in the development of AE test methods and evaluation criteria for GFRP bridge dec ks. Three commercially available GFRP bridge decks were tested in fl exure at service load levels and to ultimate strength. Serv ice level loads were used with AE inspection 13

PAGE 14

14 to determine if damage could be detected. Relative ratio analysis (RRA) was derived from a previously proven method for the evaluat ion of the AE data. In this thesis, the RRA method was derived and implemented firs t on laboratory test data and then on the AE data collected during a bridge load test of a recently installed GFRP deck.

PAGE 15

CHAPTER 2 BACKGROUND Over the past 15 years, more than 100 bridges in the United States have been rehabilitated using FRP composites (Reev e 2010). The materials light weight, corrosion resistance and fast installati on time are the main advant ages over conventional deck systems. In 1996, the firs t glass fiber reinforced polymer (GFRP) bridge was installed over a creek in Kansas (OConnor 2008). Since then, GFRP has been used in a variety of vehicular bridge decks. GFRP Deck Design and Fabrication GFRP deck designs and fabrication processes are manufacturer specific. Consequently, the details of components and production are usually proprietary. Samples used in this investigation were manufactured using the following assembly methods: pultrusion, wet layup, and preform molds. Pultruded deck panel (Deck A) In a typical pultrusion process, flexible glass fiber reinforcement is drawn in varying orientations and sequences with a manufacturer specific vinyl/ester resin through a die to construct a continuous elem ent of constant cross section ( Figure 2-1 ). Longitudinal reinforcement is pro vided by the glass fibers in the roving racks; the mats can have various fiber orientations as needed for the particular design. Zellcomp deck is the pultruded deck used in this invest igation and will be referenced as deck A. Zellcomp deck is pultruded in two separate piec es and is available in 5, 7, and 9 in. depths. ( Figure 2-2 ). The depth of the dec k used in this research is 5 in. The bottom panel is composed of a bottom pl ate and four I-shaped sections (hereinafter referred to as webs) that were pultruded as a single unit. The top plate was a pultruded sheet 15

PAGE 16

that provides the wearing surface and transfers wheel loads to the webs. Top plates were attached to the bottom panel using self -tapping countersunk scr ews; the top plate is typically attached during the installation of the bridge deck. Preform molds deck panel (Deck B) Stay-in-place preforms are used in the wet lay-up assemb ly process to support the wet mats during the curing process. T hey are made from polyurethane foam molded into the desired shape to creat e the cross-section. This method was developed for the boating industry ( Figure 2-3 ). The assembly process is si milar to th e wet layup process but utilizes foam forms, which are made of 2 pound per cubic foot floatation-grade polyurethane foam, to hold the shape of the cross-section until the resin (preforms.com 2011). The deck is manufactured with threaded sleeves that are used to attach the deck to the bridge superstructure with clips. Deck B was manufactured by Structural Co mposites, Inc. and used trapezoidal preforms to manufactu re its GFR P deck ( Figure 2-4 ). The preform was used to hold the wet lay-up fiber in plac e during the cure period. The depth of the deck used for this investigation was 5 in. Wet lay-up deck panel (Deck C) Wet lay-up is a fabrication te chnique that arranges layers of the glass fiber mats and resin in a prescribed sequence. The type of glass reinforcement and the orientation of the fibers are selected by the manufacturer ( Figure 2-5 ). One type of deck that is constructed using wet lay-up is the GFRP honeycomb sandwich deck. The deck consists of a honeycomb core bonded between two flat sheets of GFRP using vinyl ester resin ( Figure 2-6 ). (Kalny et al. 2004) Deck panels are typically manufactured with pockets to accommodate stu d connec tors. The fiber-reinforced polymer 16

PAGE 17

honeycom b panels (FRPH) used in this inve stigation were manufactured by Kansas Structural Composites, Inc. GFRP Bridge Deck Structural Behavior A number of GFRP deck systems struct ural behaviors have been investigated. Prachasaree et al. (2009) investigat ed in-plane and out-of-plane shear behavior of GFRP specimens and the load transfer efficiency of the joints between the adjacent specimens. Keller and Gurtler (2005) examined the bending behavior of pultruded GFRP deck in a simply supported test se t-up. Alagusundaramoorthy et al. (2006) investigated the deflection behavior of GRFP in a similar set-up. Camatta and Shing (2010) also tested FRPH deck and found the de ck failure to occur in the bond between the core and the face panels. Brown and Berm an (2010), Cousins et al. (2009), Vyas et al. (2009), Kalny et al. (2004) and Chen and Davalos (2010), while testing different GFRP decks, reported similar behavior when approaching the specimens capacity; the tested deck systems failed suddenly, us ually preceded by audible cracking. Hong and Hastak (2006) indicates that t he main problems found in FRP dec ks are debonding, delamination, and cracki ng of the wearing surface. No specific inspection, repair, or maintenance spec ifications or recommendati ons, however, are currently available. Telang et al. (2006) recommends t he use of visual and t ap-test methods as a suitable inspection method. Acoustic Emission Acoustic emissions (AE) evaluation is a widely used non-destructive evalu ation (NDE) method for detecting damage in materials such as FRP liquid storage tanks (Fowler et al. 1989). In addition, a number of researchers have begun to use AE to evaluate GFRP bridge decks (Kalny et al. (2 004), Gostautas et al. (2005), and Cole et 17

PAGE 18

al. (2006)). The principle beh ind AE is that fi ber rupture and resin crackingindicators of structural damageemit stress waves in the structure that can be detected by sensors that are resonant in the fr equency range in which these stre ss waves occur. Features of these stress waves can be used to distinguish them as genuine emissions from damage rather than background noise or other s purious events. The following sections provide a brief description of AE fundamentals along with details of recent AE testing that was conducted on GFRP bridge decks. AE fundamentals Acoustic emissions waves are elastic wa ves produced by the release of stored strain energy caused by fracture in the ma terial. The release of energy causes an elastic wav e in the material detectable by a piezoelectric sensor, which converts the physical motion into an electrical signal. Figure 2-7 illustrates the AE detection process in whic h the wave is detected by the sens or and analyzed by the in strument. Features of the wave shown in Figure 2-8 are recorded by the AE system. Threshold is the minimum amplitude that must be reached by t he event for it to be analyzed. Threshold values are typically set based on the materi al and the testing conditions, events that do not exceed the threshold are ignored. Energy (also signal strength) refers to the area under the AE wave form. Amplitude is the peak voltage of t he event and is expressed in decibels rather than vo ltage (Grosse and Ohtsu 2008). Acoustic emisions are caused by the forma tion of cracks in the material created during the release of strain. The Kaiser e ffect sta tes that if a sample is loaded, unloaded and then reloaded under the same conditions, AE events should not be detected until the previous load peak is reached. This condition is typically satisfied in most materials if no permanent damage, the presence of signi ficant AE at a lower load 18

PAGE 19

than previously experienced by the specimen has occurred in the specimen. Felic ity effect is the breakdown of the Kaiser in which significant AE occurs at a load level below the previous maximum. The Felicity ratio (FR) is a measure of the significance of this effect and related damage to the sample; it is t he ratio of load at the onset of AE activity to the previously achieved load. Therefore, an FR of 1.0 or greater is viewed as no damage; an FR less than 1.0 may be indicative of cumulative permanent damage (Gostautas et al. 2005). Equipment and settings In previous tests of GFRP decks, Cole et al. (2006), Gostautas et al. (2005), and Kalny et al. (2004) used 150 kHz sens ors and broad-band sensors with a response range of 100 to 2100 kHz; both were pre-amplif ied for a 40 dB gain. AE monitoring was conducted in this research using a Ph ysical Acoustic Corp. (PAC) system. T he threshold was set to 45 dB. PAC data acquisition (DAQ) system with MISTRAS AEDAQ software was used to collect and record t he data. Each sensor was coupled to the surface using hot melt glue, which is a typical attachment method for AE sensors. Evaluation methods Calm ratio, comparison analysis, and intensit y analysis have been used to evaluate the AE data from different tests. The calm ratio, t he ratio of a selected feature AE activity during the unloading to the AE acti vity during the loading, is evaluated by plotting it against the load ratio, the ratio of load at the onset of AE activity to the maximum load experienced by the structure. Kalny et al. (2004) used AE to monitor the FRPH sample during testing. Comparison analysis compares the data to a previous loading and is based on Felicity ratio and Ka iser Effect. Intensity analysis has been 19

PAGE 20

used in the pressure vessel indus try and has been investigated in some deck monitoring. (Gostautas et al. 2005). Field testing AE testing on GFRP bridge components has been limited primarily to laboratory tests. A small number of AE tests have been conducted in the field on both GFRP and concrete structures. In a fi eld test of a GFR P deck by Joeng et al. (2006), the deck was tested using a load truck. Turner et al. (2004) performed a load test on a GFRP deck using load trucks near HS 25 load level. Luo et al. (2004) used AE to evaluate the structural integrity of an in-service concrete bridge pier on a train bridge. Calm ratio was used in the analysis of the structure and a new index was developed to replace the load ratio, ratio of Repeated Train load at the onset of AE activity to Relative maximum load for Inspection period (RTRI). The introduc tion of this factor makes it possible to use the calm ratio analysis when the historic maximum load on the structure was not known. Figure 2-1. Pultruison proc ess (Ultrafiberglass.com). 20

PAGE 21

3 sp @ 8" = 2'-0" 3.5" 3.5" 4.5" 0.5" self-tapping fasteners 4" 4" 0.5" Figure 2-2. Zellcomp deck panel. Figure 2-3. Preform sample beam used by Structural Compos ites of Melbourne (performs .com). Figure 2-4. Structural Composites deck panel. 21

PAGE 22

Figure 2-5. Wet lay-up glass fi ber schedule (Kalny et al. (2004). Figure 2-6. Fiber reinforced polymer honeycomb deck panel. Figure 2-7. Acoustic emissions event. 22

PAGE 23

23 Threshold Time Time Rise Time Volts Amplitude Energy Counts Duration Threshold Crossing Figure 2-8. Feature data gather ed from a single AE wav eform.

PAGE 24

CHAPTER 3 TEST SETUP AND SPECIMEN DESIGN A ND CONSTRUCTION The primary focus of this research wa s to develop an AE evaluation that can be applied to in-service GFRP decks Bec ause GFRP deck panels are typically designed to span continuously over several supports, capturing both the positive and negative bending behavior of each deck type during structural testing was necessary to ensure that the damaged deck condition was as close to actual field conditions as possible. In addition, concentrated loading was used to simula te wheel loading that typically controls the design of a bridge deck. Span and Support Configuration The test setup was designed to simulate the actions caused by wheel loads on a GFRP dec k such as the Hillsboro canal crossing bridge in Belle Glade, FL (Singh et al. 2010). In lieu of testing a single conti nuous panel, separate tests to independently evaluate the positive and negative bending behavior and associated AE activity were conducted. Figure 3-1 shows the positive bending setup. In this configuratio n the panels were tested in the upright position to simulate positive bending in the typical installation. Wheel loads were approximat ed by applying the load through a 10-in. by 20-in. bearing pad. Figure 3-2 shows the negative bending test se tup. The panels were inverted in the test setup to simulate negative bending Before testing, the deck-to-girder connection as was typical for t hat system. Wheel loads in th is setup were simulated by placing two 10 in. by 10 in. bearing pads between the deck and supports. The load was applied through the spreader beam which was sim ilar in width to the bridge girders an were stiffened by welding plates connecting the webs and flanges in four locations 24

PAGE 25

along the beam. Neoprene bearing pads we re used at support and load application point to reduce ambient and mechanical noise from contact with the supports and the hydraulic actuator; this reduced spurious AE activity. Specimen Design and Construction Three different types of commercially available GFRP deck, designed and fabricated by Zellcom p, Structural Compos ites of Melbourne, and Kansas Structural Composites Inc. were selected for testi ng. Each deck was tested independently in positive and negative bending test setups to capture the individual behavior of each action. The specimen depth was 5 inches and the dimensions for each deck type were the same for positive and negative bending specimens and are listed in Table 3-1 The positive bending specimen was A_P. To simulate negative bending (flexural tension in the top plate), the loading frame was fitted with a shear stud. This stud was grouted into specimen A_N as shown in Figure 3-3 The positive bending sample (B_P) had a wearing surface applied by the manufacturer. The negative bending sample (B_N) had a signific ant amount of cracking on the wearing surface; this was present prior to testing. The debonding of the wearing surface during the negative bending testing of the sample resulted in inaccurate strain data from the sensors on the wearing surface. For B_N the attachment mechanism to the girder was a series of bolts and clips th at secured the girder to the deck. These were placed at the wrong location, away fr om the girder connection, however and were not used in the testing. Positive bending specimen (C_P) was a solid section of FRPH. The negative bending specimen (C_N) used a single attachment point composed of a recessed steel flange into the deck which fastens the deck to the girder using a welded stud to the 25

PAGE 26

girder. Additional information on the connections can be found in Asencio and Hamilton (2011). Figure 3-1. Positive bending test setup. 26

PAGE 27

Figure 3-2. Negative bending test setup. 27

PAGE 28

Non-shrink grout Shear stud FRP sample Cardboard block off Top surface of deck Spreader beam Figure 3-3. Stud-in-g rout connection detail. Table 3-1. GFRP sample dimensions (in.). Specimen Length Width A 55 33 B 55 36 C 60 36 28

PAGE 29

CHAPTER 4 INSTRUMENTATION GFRP deck specimens were instrumented wi th strain gages, displacement gages, and load c ells to measure the deck response to service and ultimate loads. Acoustic emission (AE) testing must be conducted duri ng loading. Consequently, AE specimens had sensors in place during the loading and unloa ding process. Strain Gages Strain was measured using quarter-bridge, 120-ohm resistance, 6 mm long gages. This gage size had been used successfully on recent tests of simi lar GFRP deck systems and on recent bridge load tests on a GFRP bridge deck. Using the same gages in the lab tests helped maintain consis tency between the bridge and the lab tests results. Strain measurement focused on measuring fl exural tensile strains on the extreme fiber of the section. Because the decks we re tested in either positive or negative bending, tensile strains occu rred on opposit e faces of the deck depending on the setup used. Positive bending test setup and strain gage locations are presented in Figure 4-1 and Table 4-1 Gages were arranged to capture the va riation in s train over the width of the section that was expected to occur due the concentrated nat ure of the load. In the negative bending test set-up, the st rain gages were placed on the wearing surface of the specimens. Deck A has been s hown to have very little composite action between the top plate and bottom panel (Cousi ns et al. 2009), therefore the strain gages S1S3 for Deck A were placed on the flanges of the bottom panel as shown in Figure 4-2 in order to measure tensile strains under negative bending. 29

PAGE 30

The top surface strain gage locati ons for Decks B and C are shown in Figure 4-3 Strain gage locations were selected for each of these systems to accommodate the varying web configurations. Gages S1, S2, and S3 were placed on the top surface of the sample. The other five gages were placed on the soffit of the deck ( Figure 4-4 and Table 4-2 ). Load and Displacement Gages Displacement gages were arranged for the positive bending setup as shown in Figure 4-5 For Decks B and C, the gages were placed along the center line of the samples; the gages were offset from the c enter line for Deck A bec ause the top plates had been observed to separate sli ghtly from the bottom panel under high loads. The offset displacement gage location allows for the measurements to be based only on the bottom panel deflections. The load in the configuration is measur ed with an inline load cell placed between the load ac tuator and the specimen. Acoustic Emission Sensors Acoustic emission sensors were placed only in locations that would typically be accessible in a field applicati on. Sensors were placed on th e soffi t of the samples under and away from the load and under the di fferent features of each sample ( Figure 4-6 and Table 4-3 ). Damage was expected to occur in the area surrounding the loading girder in the negative bending test setup. Consequently AE sensor locations for the negative bending tests were similar to those of t he positive bending test, but adjusted to accommodate the placement of t he girder. As in the positiv e test setup, the sensor locations were chosen to capture the AE activity between the load point and the supports while being attentive to different features in each deck design. Figure 4-7 30

PAGE 31

shows the negative bending setup and the specif ic loc ations related to the different decks are listed in Table 4-4 Using hot glue as the couplant, the sensors were attached in accordance with ASTM E 650-97. Prior to loading, as per ASTM E 976-99, the sensor placement and data acquis ition were evaluated using the pencil lead break method. Disp 16 System The AE data acquisition system used wa s the DISP-16BT, manufactured by Physical Acoustics Corporation. This system used the software AEWin E3.11 to acquire, record, and replay eight channels of feature data. The system used integral preamp piezoelectric sensors model number R 15I-AST. The sensors had the capability to drive long cables without the need of a separate preamplifier. The sensors were designed for circumstances where background noi se can be high with a built in filter and a 40dB preamplifier in the sensor (P hysical Acoustic Corporation 2008). Table 4-5 lists the adjustable parameters val ues used in this investigation. Peak Definition Time (PDT) ensures the correct id entification of the signal peak for risetime and peak amplitude measurement; Hit Definiti on Time (HDT) ensures that each AE signal corresponds to only one hit, and Hit Lockout Time (HLT) avoids recording spurious measurements during th e signal decay. (Physical Acou stics Corporation 2004) The threshold was selected after several l ab tests measuring am bient noise. The sample rate was set to the maximum allow ed by the machine. PDT, HDT, HLT, and Max duration settings recommended by the equipment supplier for GFRP materials were used. 31

PAGE 32

Figure 4-1. Strain gage locations fo r positive bending test soffit vi ew. Figure 4-2. Strain gage locations for negative bending deck A. Figure 4-3. Strain gage locations negative bending test wearing surface view. Figure 4-4. Strain gage locations negative bending tests soffit view. 32

PAGE 33

A B Figure 4-5. Displacement gage locations pos itive bending setup A) elevation and B) plan view. Figure 4-6. AE sensor locations fo r positive bending tests soffit view. Figure 4-7. AE sensor locations fo r negative bending tests soffit view. Table 4-1. Strain gage location dim ensions positive bending tests (in.) Specimen a b c d e f A 12 8 4 4 4 4 B 13 9.5 5 5 4.5 4.5 C 13 9 5 5 6.5 7 33

PAGE 34

34 Table 4-2. Strain gage location dim ensions negative bending tests (in.). Deck Type a b c d e f g A N/A N/A N/A 3.2 9.5 3.5 12 B 0.0 4.5 9.0 7.0 10.25 3.5 11 C 4.0 9.0 13 3.5 7.0 6.0 3.5 Table 4-3. AE sensor location dimens ions positive bending tests (in.). Deck a b A 4 4 B 5 6 C 5 6.5 Table 4-4. AE sensor location dim ensions negative bending tests (in.). Deck a b A 4 4 B 5 9 C 5 6.5 Table 4-5. AE recording parameters. Threshold 45 dB Analog filter 10 kHz-2MHz Sample rate 10MSPS Peak Definition Time (PDT) 35 s Hit Definition Time (HDT) 150 s Hit Lockout Time (HLT) 300 s Max duration 1000 ms

PAGE 35

CHAPTER 5 TEST PROCEDURE Multiple load tests were conducted on each specimen using three different loading profiles. The sequence of the loading profiles is shown in Table 5-1 First, the specimen was subjected to tw o undamaged service tests (UST) composed of three load holds. The specimen was then subjected to an ultimate load test (ULT), in which the specimen was loaded to its ultimate capacity. This was done to cause damage without completely destroying the specimen. Fina lly, stepwise service loads (DST) were again imposed on the damaged deck to determine how well AE could detect the damage that had occurred during the ultimate strength te st. Both positive and negative bending specimens were tested using the same procedures. Load steps were separated by a brief load reduction to allow observation of Ka iser and Felicity effects during reloading. Details of the loading profile ar e provided in the next section. The loading rest time r equirements for each loading wer e designed to duplicate the field conditions experienced by a deck. The initial UST was followed by a rest period of 24 hours. A second UST was perform ed in order to collect data from a known undamaged previously loaded deck, which should have considerably less AE events than the initial loading. Service Loading (UST and DST) The setup designed for the laboratory test s for both positive and negative bending test allow the sections to have different loads per deck system. Details of the development of these load steps can be found in Asencio and Ha milton (2011). The purpose of intermediate load holds was to check for Kaiser and Felicity effects, as prescribed in ASTM 1067. Figure 5-1 shows the service loading profile used for both 35

PAGE 36

UST and DST. Load hold magnitudes and duration are listed in Table 5-2 To detect either Kais er or Felicity effects and to develop undamaged AE data of the samples, two UST were performed on each sample before the ultimate loading. Samples were then loaded to ultimate capacity imposing signific ant damage. Following the ultimate loading test, the damaged samples were then retested (DST) using same load profile as was used for UST. Ultimate Loading (ULT) The ultimate load test (ULT) was design ed to provide AE data for each type of deck tested up to failure and to take the specimen to ultimate capacity to ensure signific ant damage had been imposed on the s pecimen before the follow-up DST. Figure 5-2 shows the load profile used in the UL T. The duration of each load hold was a minimum of 2 min. and was adjusted based on the live AE data observed during the loading (ASTM 1067-96). Load hold lev els st arted at 45 kip and increased by 15 kip until capacity was reached. Figure 5-1. UST and DST load profile. 36

PAGE 37

37 Figure 5-2. Ultimate loading test (ULT) profile. Table 5-1. Testing sequence and rest time. Loading Sequence Post-Test Rest time Initial UST 24 hr (min) UST 30 min. ULT 30 min. DST 30 min. Table 5-2. Load profil e characteristics. Load (kip) Load Hold A B&C Minimum Duration (sec.) A 5 10 120 B 9 15 120 C 12 20 120 D 2.5 5 30 E 5 10 30

PAGE 38

CHAPTER 6 ULTIMATE STRENGTH TEST -RESUL TS AND DISCUSSION Positive and negative bending tests were conducted on each deck type. The focus of the tests was to load the specimens beyond their ultimate capacity and cause damage that might be associated with an overload in th e field. This was done primarily to evaluate the effectiveness of AE in detecting the damage caused by overload. Recall that service loadings were applied both before and after each ultimate load test to assist in evaluating damage detection procedures. In general, failure modes differed between positive and negative bending tests. The following sections describe the behav ior and observations of the positive and negative bending tests conducted on the respective deck type. Initially, the specimens were loaded using the service load profile to establish the behavior of the undamaged specimens; these will be referred to as undamaged serviceload tests (UST). The ultimate loading te st (ULT) was then used to impose damage and determine the ultimate load capac ity. T he damaged samples were then reloaded using service level loads, these tests will be referred to as damaged service load test (DST). The following sections describe the behavio r during each of the three load tests including the failure mode noted during ULT. Strain data are also presented and evaluated. Deck A Positive Bending (A_P) Figure 6-1 shows the load-displacement re sponse of specimen A_P tests and include UST, SLT and DST plots. Two USTs were initially performed on the sample. During the first test, sharp audible cracking was noted as the load increased. During the service load hold, however, the deck re mained quiet. Audible cracking was less 38

PAGE 39

notable dur ing the second UST and was quiet during the load hold. The second load cycle is shown in the plot and is linear, indi cating that the deck remained linear-elastic during this test. ULT load-deflection was linear up to a peak load of 72 kip. Audible cracking was noted as load was increased. In addition, w hen service load levels were exceeded, the top plate of the deck system was observed to deform ou t-of-plane between fasteners ( Figure 6-2 ), this was an indication that the de ck system was not acting compositely. Ultimate capacity was signified by both a s harp audible crack and partial loss of load, but not complete collapse. A residual capaci ty of 43 kip remained after the specimen load reached its ultimate limit state. The unloading portion of the ultimate loading plot is not shown below the 43 kip mark to more cl early show the service load level loadings. Visual inspection indicated that the loss of load was caused by the crushing and buckling of the webs adjacent to the loading pad ( Figure 6-3 ). DST shown in Figure 6-1 indicated that the web dama ge reduced the overall deck stiffness. Table 6-1 shows the stiffness of the specimen for each test regime as determined by linear regressi on of the last curve of each test. As i ndicated in Table 6-1 the measured stiffness was reduced by 27% from the undamaged s tate. Figure 6-4 shows the ULT strain values for A_P. The relative magnitude of the strain values at any given load indicates transverse distribution of the concentrated load. The strain under the load point (S3) was approximately twice that of the strains under the outside web (S1) which indic ates t hat the interior webs are carrying about twice the load of the outside web. This signifi cant difference in load is indicative of the relative flexibility of the panels connecting the webs and the marginal amount of load 39

PAGE 40

distributed to adjacent webs. Linearity of th e strains in S1 and S2 indic ates that the system remained linear elastic up to ultimate capacity. Abrupt loss in load with little indication of softening demonstrates t he brittle nature of the material. Deck A Negative Bending (A_N) Figure 6-5 shows the load-deflection curves for the UST, ULT, and DST load test. It was anticipated that the grout pocket may have exhibited additional audible cracking during UST No increase in audible cracki ng, however, was noted over that of the positive bending test. ULT load-deflection was initially linear with audible cracking noted at about 50 kip, which was thought to be the gr out cracking. The grout wa s thought to provide some support due to the large deformations noted in ot her tests. At appr oximately 65 kip ULT load-deflec tion curve changes slope, which suggests the sample sustained a large amount of damage at that time but the specimen supported additional load up to 76 kip when the sudden failure occurred. The residual load after reaching ultimate strength was 63 kip. Post ULT inspection indicated that a si ngle web directly above the supports was found to be damaged ( Figure 6-6 ). Figure 6-7 shows the location of the damaged web in relation to the test setup. This localized failure is similar in nature and location to the failure enc ountered in the positive bending ULT. In this test, ho wever, the failure resulted in little loss of global stiffness due to its localized nature. Consequently, the specimen retained its original stiffness at th e lower service load levels, as indicated by the slope of the DST loaddisplacement curve. Table 6-2 shows the calculated deck stiffnes s for each test. As previously mentioned only two of the webs were di rectly in contact with the supports ( Figure 6-7 ) 40

PAGE 41

one of whic h was crushed during ULT. The stra in differential of the two webs directly under the load could not be conf irmed with the strain data since one of the strain gages was damaged during assembly. In contrast the positive bending sp ecimen lost 24% of its original stiffness after ULT and had dam age to both webs under the applied load. Figure 6-8 shows the load-strain data. The specimen initially had three strain gages, but S3, which was mounted under the web that failed, was damaged when the top plate was being replaced. S1 strain val ues were 35% of the S2 strains confirming transverse flexibility. The strain curves remained linear up to capacity. Deck B Positive Bending (B_P) Figure 6-9 shows the load displacement response of test B_P and includes UST, SLT, and DST plots. Specim en B_P was qualitatively noted to have significantly less audible cracking than A_P, which was thought to be due to damping e ffects of the foam core. ULT of the sample reached ultimate c apacity of 112 kip, which was accompanied by a sharp audible cracking. A residual load of 92 kip was observed prior to the unloading of the sample. During the ultimate strength test, cracks formed around the load point in the top panel (highlighted in Figure 6-10 ). As load was increased, the cracks grew parallel to the span and eventually merged over the supports when the ultimate capacity was reached Figure 6-11 Post ULT inspection of the specimen indicat ed that the bottom soffit of the sample was intact and that the top surface re tained a bowl-shaped permanent deformation, indicating that the interior webs were also permanently deformed. 41

PAGE 42

The difference in slope between the UST and the DST curves demonstrates the loss of stiffness after the ULT. The stiff ness and stiffness ratios were calculated using linear regression for the three loadings and are presented in Table 6-3 The stiffness of the section was decreased by 30% after ULT procedure. Figure 6-12 shows the strain values for the UL T of the B _P. Strain gages plots were grouped by their respective positions in relation to the load pad. S1 and S2, the gages positioned away from the load, reported values 50% that of the gages directly under the pad, S3 and S4. The transverse stiff ness of the deck is small and similar to Deck A. Deck B Negative Bending (B_N) Figure 6-13 shows B_N results of UST, ULT and DST load-deflection plots. Specimen B_N setup had no mechanical co nnection to the girder, the deck was insulated by a in. neoprene pad, this c ontributed to the lack of audib le cracking emitted by the specimen. A large crack formed over one of the supports at approximately 112 kip as indicate d by the discontinuity in the ULT plot. However the specimen did not lose load and the loading conti nued to an ultimate capacity of 134 kip. The specimen failed abruptly and violent ly with a residual load of 69 kip. Figure 6-14 shows punching shear failure over the support. The crack followed the edge of the interior web exten ded past the support into the deck. Figure 6-15 highlights the approximate loca tion of the failure plane. The deck wearing surface delaminated from the deck su rface and piec es can be seen on the steel support beam in Figure 6-14 Table 6-4 shows the stiffness and the stiffnes s ratio for the negativ e bending test. The difference in stiffness between the UST and the ULT was 14% this was the only 42

PAGE 43

deck system to lose si gnificant stiffness bet ween UST. The DST stiffness was 76% less than the UST and 62% less than the ULT. The punching shear failure caused a large loss of stiffness relative to the other deck tests. Strain gages were placed on the road surface of the sample to capture the largest magnitude of strain during l oading. During ULT, however the road wearing s urface applied by the manufacturer debonded from the deck damaging two of the strain gages. The ULT load-strain plot for the remain ing undamaged strain gages is presented in Figure 6-16 The strain magnitudes were signifi cantly less than the positive bending test strain values of the same deck type, which were nearly 10000 ( ) at 112 kip. The reported strain of the negative bending test was expected to be similar in magnitude to the positive bending strains. Recorded strain values, however, were only 1200 ( at 135 kip. This suggest s that the debonded wearing surface affe cted the remaining strain gage and the data are not accurate. Deck C Positive Bending (C_P) UST, ULT and the DST results for C_P are presented Figure 6-17 Initially two USTs were performed on specimen C_P follo wed by the ULT. During the service loadings audible emissions were notably less than those from the ot her two deck types. During the ULT sharp audible cracking began at 70 kip load level and continued up to the ultimate capacity of 100 ki p. Residual load of 80 kip wa s supported by the sample; the unloading portion of the curve was not shown to avoid obscuring the UST and DST results. The deck system is designed so t hat the compression flange delaminates from the honeycomb core, which was the observed failure mode. Delamination, however, 43

PAGE 44

was not visible sp coin tap testing was us ed to estimate extent of the delamination (outlined in blue in Figure 6-18 ). The stiffness of each loading was determined by linear regression ( Table 6-5 ). The samples stiffness was the greatest of t he three tested. The samples stiffness was reduced by 30% after the ULT. Figure 6-19 shows the ULT strain values. The transverse stiffness of the sample was very large in comparison the other samples. The strain value near the outside edge of thee deck (S1) had the smallest stra in magnitude and it was 75% of the peak value under the concentrated load (S4). Deck C Negative Bending (C_N) The load-deflection plots UST, ULT, and DST are presented in Figure 6-20 During the initial UST no audible cracking was noticed. The ULT was also quiet up to the 65 kip level when audible cracking was noted, increasing up to the sudden and violent failure at 79 kip load level. Immediat ely after the ultimate load level was reached the system lost almost all of its strength the damage was extensive and the system only had 14 kip of residual load immediately after the failure. The sample failed by delamination of the t from the center honeycomb section out to the support seen in Figure 6-21 The sample was designed to fail by delamination of the compression zone of the deck; however in this case the large load and the short span caused the compression zone to reach the support and the entire sheet delaminated. The failure seen in the samp le would not be seen on the field because the deck would be a continuous slab and the load would never be placed that close to an unsupported edge; however the deck behavior of a delaminated sample was still of interest to the project. 44

PAGE 45

The stiffness and the stiffness ratio to t he UST stiffness of each of the loadings above were calculated by linear regression and listed in Table 6-6 This deck was very stiff in comparison to the other decks test ed, but due to the failure mode the damaged loading had the largest decrease in stiffness of 86% after the ULT. The strain gage data of the deck along the top of the girder are presented in Figure 6-22 In this configuration the decks transverse stiffness was not as large as in the positive bending s etup. S1 only had 20 % of the peak strain recorded during the ULT near the center of the specimen (S2). Deflection (in.)Load (kip) 0 0.25 0.5 0.751 0 10 20 30 40 50 60 70 80 DST UST ULT Figure 6-1. Load-displacement plot A_P tests. 45

PAGE 46

Figure 6-2. A_P top plate buckling under load. Figure 6-3. A_P web failure. Figure 6-4. Load-strain plot A_P ULT. 46

PAGE 47

Deflection (in.)Load (kip) 0 0.25 0.5 0.751 0 10 20 30 40 50 60 70 80 ULT UST DST Figure 6-5. Load-deflecti on plot A_N tests. Figure 6-6. A_N damaged web. 47

PAGE 48

Figure 6-7. Damaged web location. Figure 6-8. A_N ULT load-strain plot. 48

PAGE 49

Deflection (in.)Load (kip) 0 0.25 0.5 0.75 11 0 .25 10 20 30 40 50 60 70 80 90 100 110 120 ULT DST UST Figure 6-9. Load-displacement plot B_P tests. Figure 6-10. B_P test cracking pattern. 49

PAGE 50

Figure 6-11. B_P crac ks at failure load. Figure 6-12. Load-strain plot B_P ULT. 50

PAGE 51

Deflection (in.)Load (kip) 0 0.25 0.5 0.75 11 .2 5 0 20 40 60 80 100 120 140 ULT DST UST Figure 6-13. Load-deflect ion plot B_N tests. Figure 6-14. Shear fail ure over support B_N. Figure 6-15. Shear fail ure location B_N. 51

PAGE 52

Figure 6-16. Load-strain plot B_N ULT. Deflection (in.)Load (kip) 0 0.25 0.5 0.751 0 10 20 30 40 50 60 70 80 90 100 110 ULT UST DST Figure 6-17. Load-displace ment plot C_P tests. 52

PAGE 53

Figure 6-18. Debonded zone C_P. Figure 6-19. Load-strain plot C_P ULT. 53

PAGE 54

Deflection (in.)Load (kip) 0 0.25 0.5 0.75 11 0 .25 10 20 30 40 50 60 70 80 90 ULT UST DST Figure 6-20. Load-displace ment plot C_N tests. Figure 6-21. Debonding failure C_N. 54

PAGE 55

Figure 6-22. Load-strain plot C_N ULT. Table 6-1. A_P stiffness. Deck Loading Stiffness (kip/in) Stiffness Ratio UST 114 1 ULT 111 0.97 DST 83 0.73 Table 6-2. A_N stiffness. Deck Loading Stiffness (kip/in) Stiffness Ratio UST 114 1 ULT 113 0.99 DST 110 0.96 Table 6-3. B_P stiffness. Deck Loading Stiffness (kip/in) Stiffness Ratio UST 117 1 ULT 116 0.99 DST 82 0.70 55

PAGE 56

56 Table 6-4. B_N stiffness. Deck Loading Stiffness (kip/in) Stiffness Ratio UST 140 1 ULT 120 0.86 DST 34 0.24 Table 6-5. C_P stiffness. Deck Loading Stiffness (kip/in) Stiffness Ratio UST 181 1 ULT 184 1.01 DST 128 0.70 Table 6-6. C_N stiffness. Deck Loading Stiffness (kip/in) Stiffness Ratio UST 169 1 ULT 172 1.01 DST 24 0.14

PAGE 57

CHAPTER 7 ANALYSIS OF AE TEST DATA USING RECOVERY RATIO ANALYSIS This chapter covers the analysis of AE data gathered during structural testing of the deck specimens. Either the adaptation of an existing method or develop a new evaluation procedure was needed so that the AE data could be us ed to evaluate the structural integrity of a GF RP deck. Intensity analysis was developed for and is still in use by the railroad tank industry for inspecting steel rail cars; it uses the cumulative signal strength and changes in signal strength to determine stru ctural integrity. This method was used initially to evaluate the AE data, but wit h limited success. Recovery ratio analysis was developed for this project by adapting calm ratio, which has been used to evaluate reinforced concrete structures (Grosse and Ohtsu 2008). Luo et al. (2004) and Ohtsu et al. (2 002) both used the calm ratio to evaluate cracking in concrete structures. The calm ratio is the ratio of a selected feature AE activity during the unloading to the AE acti vity during the loading. This approach requires that the strain be meas ured at or near the AE sensor to ensure that the local peak strain is identified; loading and unloading AE dat a are then apportioned appropriately based on the peak strain. Ohtsu et al. (20 02) used load in laboratory testing of prestressed concrete beams to index the key points needed for the analysis. Other strain related parameter s such as load or displacement have also been used to index the data. Luo et al. (2004) applied the calm ratio to a load test on an in-situ bridge pier using displacement as the indexing factor. Structural evaluation using the calm ra tio also requires that the load ratio be determined, which is based on the Felicity effect and is the ratio of the load at the onset of AE activity to the maximu m load experienced by the specimen. Ohtsu used the load 57

PAGE 58

ratio in his investigation because it is appr opriate for laboratory settings where the load is known. Load histor y for in-service br idge components, however, is unknown. To circumvent load measurement, Luo proposed using the RTRI ratio (ratio of the Repeated Train load at the onset of AE activity to Relative maximum load for Inspection period) be used in place of the load ratio; it uses the ratio of disp lacement at the onset of AE activity to the peak displacement seen in the current load cycle. This differs from the load ratio in that a relative value is used instead of the maximum load ever experienced. To evaluate the level of damage in t he structure using AE, Ohtsu suggested plotting the calm ratio along the ordinate and load ratio along the abscissa. From empirical data the rati os are divid ed into damaged and undamaged groups. These values are plotted as vertical and horizontal lines on the plot, dividing the plot into quadrants corresponding to the severity of t he damage. The top right quadrant would have the most significant damage and the botto m left quadrant would define the least amount of damage. Leaving the other two quadrants as the intermediate damage level. A high calm ratio combined with a low load ratio is indicative of heavy damage. A large amount of AE data during unloading (h igh calm ratio) is typically indicative of significant damage; this combined with AE activity at low l oads (low load ratio) is a strong indicator of damage. The goal of the present research was to determine if AE could be used to detect damage in GFRP dec k panels in the laborator y and then eventually in the field during bridge load tests. To that end recovery ratio analysis was developed by adapting procedures by Ohtsu 2002 and Luo 2004 for use in the laboratory deck. The relative 58

PAGE 59

load ratio (RL) ( (71 ) is defined as the ratio of the l oad at onset of AE activity to the maximum load during each loading cycle: MLD ILD RL (71) where ILD is the load at which significant AE activity is detected, and MLD is the maximum load level of the load cycle. The AE ratio (AEr) is a hit-based parameter and can be calc ulated as follows: LDH ULH AEr (72) where ULH is the number of hits during the unloading of the sample and LDH is the number of hits duri ng the loading. Eight channels of AE data were collect ed during testing of both undamaged (UST) and damaged (DST) specimens. Recall that se rvice loads were used, which simulate the type of loading that would be used to generate AE in the field. Figure 7-2 (a) shows the hits recorded during DST on specimen B. To remove spurious hits, the modified Swansong 2 filter was used ( Figure 7-3 ). This is typically used in pressure tank industry to remove known mechanical noises from th e data (Association of American Rail Roads 1998). Any hit with a duration linger than 3. 5 milliseconds and amplit ude within 10 dB of the threshold is ignored along with all hits occurring within sec of the selected hit. This approach removes long-duration hits, whic h are typically indicative of rubbing and not resin cracking or fiber rupt ure. In this investigation only the hits known to be the rubbing are removed and the one second window around the hit was not removed. All hits with zero energy were also removed fr om the data set. In the example shown in 59

PAGE 60

Figure 7-2 filtering removed approximat ely 60% of the hits. Fr om the filtered data the AE ratio and the relative load ratio were then calculated for all the tests. Figure 7-4 illustrates the parameters used in the RRA and the method used to calculate AEr and RL. The analy sis is c onducted for each channel; each analysis produces a single set of values that can be used to evaluate whether damage has occurred. Average AEr results for DST for both pos itiv e and negative bending test for all samples are plotted in Figure 7-5 A & B. The UST AEr average values for most of the samples were considerably less than the DS T values, indic ating that RRA evaluation technique can consistently distinguis h between damaged and unda maged specimens. Only one sample (B_N UST) did not have the same trend as all the other samples. It is not clear why the undamaged specimen generated more AE than the damaged specimen. That particular sample only had one DST which was valid for use in the evaluation. In general, however, the method provided excellent di scrimination between damaged and undamaged labor atory specimens. Figure 7-6 shows the results from RRA for pos itive bending test. Respective damaged (hollow markers) and undamaged (solid markers) values are plotted. As noted previously, the results from all thr ee decks have distinctive differences between results from damaged and undamaged dec ks. T he various failure mechanisms and the varying severity of damage explain the large range of DST results. The UST AEr results were exclusively grouped in the lowe r range of the plot, this was the pattern used to develop the proposed pass-fa il line, highlighted in blue in Figure 7-6 Specimens A and B appear to have some difference in the relative loading ratio, but 60

PAGE 61

specimen C does not. More testi ng should be performed in the varying damage levels for the development of a pass fail criterion. Figure 7-7 shows the negative bending UST and DST results. The negative bending appeared to have good discr imination for both positi ve and negative bending. Consequently, two sets of damage regions were proposed based on the results and are depicted by the blue lines. UST RRA result s were grouped below an AEr of 0.2 with one outlier at 0.6 for sample A. This configur ation did present a pattern in relation to the relative load axis. The DST results all plott ed below 0.1 relative loads. A larger sample group will need to be considered in order to confirm these findings. Table 7-1 shows the average AE evaluation result s with the stiffness and strength ratios. These ratios provide a metric fo r comparing the damage levels sustained by each of the decks; lower ratios indicate more severe damage. The damage data do not appear to be correlated with the AE evaluation results. Time ( sec. ) L o a d ( k ip) 0100 200 300 400 500 600 700800 0 5 10 15 20 25 Figure 7-2. Specimen B_P AE data unfiltered. 61

PAGE 62

Time ( sec. ) L o a d ( k ip) 0100 200 300 400 500 600 70080 0 0 5 10 15 20 25 Figure 7-3. Specimen B_P AE filt ered AE data. Figure 7-4. Filtered data used in AE ratio analysis calculations. 62

PAGE 63

A B Figure 7-5. Average DST and UST values for A) positive and B) negat ive bending tests. Figure 7-6. RRA plot for posit ive bending tests all specimens. 63

PAGE 64

Figure 7-7. RRA plot for negative bending tests all specimens. Table 7-1. Average values from RRA and stiffness and strength rati o. AEr (avg.) RL (avg.) Specimen UST DST UST DST Stiffness ratio Strength Ratio A_P 0.05 0.53 0.15 0.24 0.73 0.60 A_N 0.02 0.43 0.20 0.09 0.96 0.83 B_P 0.05 0.91 0.50 0.17 0.70 0.82 B_N 0.58 0.37 0.47 0.06 0.24 0.51 C_P 0.04 0.29 0.16. 0.09 0.70 0.80 C_N 0.07 0.25 0.20 0.03 0.14 0.18 64

PAGE 65

CHAPTER 8 BELLE GLADE TEST In a related FDOT research project, a steel deck of a bri dge over the Hillsboro Canal in Belle Glade Florida was replaced with Deck A (McCall et al. 2011). The deck was load tested using strain and displacement gages; during this load test AE data were also recorded. Trucks typically use lanes 1 and 2 to deliver sugarcane to the refineries north of the bridge during the harvest ( Figure 8-1 ). These lanes were tested during the load test because of the large loads experi enced compared to the South bound lanes. Strain and displacement gages were plac ed along the north bound lanes. AE sensors were located near strai n gages. Details of the test procedures can be found in McCall et al. (2011). Sensor Locations Figure 8-2 is a diagram of the bridge deck pl acement and the location of the AE sensors on the bridge relative to the deck panel s. B9 and B10 are the panels monitored during the load test. The number ed lines running along the direction of travel represent the steel girders of the bridge. Table 8-1 lists the AE sensors identifications and the coordinates corresponding to the location of each sensor on the soffit of the deck. Relative Ratio Analysis on Belle Glade Bridge Data. The Relative Ratio analysis procedure developed to analyze the AE results from the laboratory testing requi res that the peak load be known so that RL can be calculated, which may not be known during a bridge test. Another parameter that may be used to mark peak load is wheel position. The influence of the l oad on the area of interest can be obtained from the combination of GPS a nd strain gage data at the AE 65

PAGE 66

sensor location. This approach is particula rly suited to bridge te sting using GPS truck positioning in whic h the position of the wheel relative to the sensor is known. The best approach would be to use the wheel position to calculate the Relative Load ratio. Unfortunately, the wheel location relative to the AE sensor position was not recorded. Figure 8-3 illustrates the adjustment made to the Relativ e Ratio analysis procedure to accommodate the constraints of the bridge test. RRA was adjusted so that the strain at the sensor caused by the rolling load ra ther than the position of the rolling load was used. The Influence ratio is defined as the ratio of the strain at the onset of AE activity to the relati ve maximum strain of the loading. Each wheel load passing the AE sensors could be used in the RRA because each represents the loading and unloading of the deck. Figure 8-4 show the hits recorded during the load test for one pa ss of the truck over a single AE sen sor. Each grouping represents the load imposed by one of the five truck axles. Although RRA could be used for any of the wheel loads, only the front and rear axles were selected for analysis. The front wheel was chosen because the data for this load was not influenced by any adjacent loads. The rear axle was also chosen because it transmits the largest wheel load. Figure 8-5 shows a summary of the RRA results for the sampled loadings for truck position one; the AE ratio was calculated es t imating the point of peak load from the AE data. Even though all the channels were used in the collection of the data, only channels 1-4 registered hits during the loading due to the attenuation of the emissions across the deck. 66

PAGE 67

AE ratios results for all the load levels ranged from 0.3 to 0. 8. These results demonstrate the potential of the analysis to be used in the evaluation of bridge decks. The exact wheel position with reference to the sensor was not known for the load test, due to the large number of hits near what was believed to be the peak of the load the ratios may shift. The front axle results we re above the 0.5 line and all the rear axle results were less than 0.5. These results suggest that the proxim ity of other wheels may have an effect on the calculated AE ratio. It is also possible that grout cracking may have influenced the AE recorded during the bridge test. A B Figure 8-1. Bridge site A) aeria l photo and B) detailed site plan. 67

PAGE 68

Figure 8-2. AE sensor and strain gage locations on the deck panels. Figure 8-3. AE and influence ratio ca lculations for LC 4 front wheel. Figure 8-4. AE data from bridge test LC4. 68

PAGE 69

Figure 8-5. Sample RRA results for bridge load test. Table 8-1. AE sensor i dentification and coordinates. Coordinate (in.) Sensor Sensor part number Panel ID X-coordinate Y-coordinate AE 1 BD 45 B10 36 188 AE 2 BD 32 B10 70 206 AE 3 BD 36 B10 104 224 AE 4 BD 48 B10 152 250 AE 5 BD 34 B9 152 282 AE 6 BD 42 B9 200 307 AE 7 DM 09 B9 234 325 AE 8 BD 43 B9 268 344 69

PAGE 70

CHAPTER 9 SUMMARY AND CONCLUSIONS Three commercially available GFRP bri dge decks were tested in b oth positive and negative bending in a three-point loading setup. Each specimen was subjected to sequential load tests having the following characteristics. First, the specimen was subjected to two undamaged service tests (UST ) composed of three load holds. The specimen was then subjected to an ultimate load test (ULT) in which the specimen was loaded to its ultimate capacity to caus e damage without completely destroying the specimen. Finally, stepwise service loads were again imposed on the damaged deck to determine how well AE could detect the da mage that had been imposed by the ultimate strength test. This test is referred to as the damaged service test (DST). Both positive and negative bending specimens were tested using the same procedures and load steps were separated by a brief load reduction to allow observation of Kaiser and felicity effects during reloading. During load testing AE, load, strain, and displacement were monitored. Strain and displa cement data were evaluated. Two AE analysis methods were used to ev aluate the data. Intensity analysis, which is routinely used in the testing of pressure vessels and has been tested on GFRP decks, was used initially to analyze the AE data fr om th e load tests. In addition, relative ratio analysis (RRA) was developed and applied to the laboratory results. RRA was also used to analyze selected bridge load test data from Belle Glade. Three repair procedures were developed and applied to Deck A. Repair A1 was to place GFRP bars in the cavities between the w ebs and then fill this cavity with concrete. A2 was to apply wet layup GFRP to the dam aged we bs. A3 was to fill the cavities with grout and apply GFRP to the soffit of the deck. 70

PAGE 71

71 From the previously described testing the following conclusions are drawn: Flexural capacity and damage characteristics of each sample wer e determined for both positive and negative bending in each of the three deck types. The ultimate capacity was found to be well beyond the maximum service wheel loads and ranged from 3.8 to 6.7. Deck A retained 60% and 83% of its positive and negative bending capacity, respectively. Deck B retained 82% and 51% of its positive and negative bending capacity, respectively. Deck C retai ned 80% and 14% of its positive and negative bending capacity, respectively. Recovery ratio analysis provided signific ant discrimination between AE data taken from unda maged and damaged decks. For positive bending, damaged decks generally exhibited an AE rati o of 0.1 or greater, wh ile undamaged results were below 0.1. For negative bending, those specimens with an AE ratio greater than 0.2 and a Relative Loading less than 0.1 were damaged, while nearly all those with an AE ratio less t han 0.2 were undamaged. Recovery ratio analysis was applied to AE data gathered from a bridge load test on a bridge with Deck A. The results, w hen compared to the criteri a developed for the laboratory tests indica ted damage. No deck dam age was noted. Refinement in the method may be needed or cracking of the grout pad used to support the GFRP deck system may be leading to false positive readings.

PAGE 72

LIST OF REFERENCE Alagusundaramoorthy, P. Harik, I. E., and C hoo, C. C. (2006). Structural Behavior of FRP Composite Bridge Deck Panels, J. Bridge Eng., 11 (4), 384-393. Arnold, R. E, (2003). Acoustic Emission Evaluation of FRP Comp osite Specimens in Tension and Bending. Morgantown, W. Va.: West Virginia University College of Engineering and Mineral Resources West Virginia University. Asencio, R. and Hamilton, H. R., (2011) D amage Detection and Repair Methods for FRP Bridge Decks. Vol. 1, FDOT Research Report BDK-75-977-17. Association of American Railroads (1998). Procedure for Acoustic emission ev aluation of tank cars and IM101tanks Issue 6. Operations and ma intenance departmentmechanical division. Brown, D. L., and Berman, J. W. (2010). F atigue and Strength Evaluation of Two Glass Fiber-Reinforced Poly mer Bridge Decks. J. Bridge Eng., 15 (3), 290-301. Camatta, G., and Shing, P. B. (2010). Static and fatigue load performance of a gfrp honeycom b bridge deck. Composites: Part B 41, 299-307 Chen, A., and Davalos, J. F. (2010). Strengt h evaluations of sinusoidal core for FRP sandwich bridge deck panels. Composite Structures 92, 1561-1573 Cole, T. A., Lopez, M., and Ziehl, P. H. ( 2006). Fatigue behavior and nondest ructive evaluation of full-scale FRP honeycomb bridge specimen. J. Bridge Eng., 11 (4), 420-429. Cousins, T. E., Lesko, J. J. Mjumdar, P. K., and Liu, Z. (2009). Rapid replacement of Tangier Island bridges including lightweight and durabl e fiber-reinforced polymer deck systems. FHWA/ VTRC 1 0-CR3 Fowler, T.J., Blessing, J.A. and Conlisk, P. J. (1989). New directions in testing. In: Proc. 3rd International Sy mposium on Acoustic Emission from Composite Materials Paris, France Gostautas, R., Ramirez, G., Peterman, R. J. and Meggers, D. (2005). Acousti c Emission Monitoring and Analysis of Gla ss Fiber-Reinforced Co mposites Bridge Decks. J. Bridge Eng. 10 (6), 713-721. Grosse, C. U., & Ohts u, M. (2008). Acoustic Emissions Testing. Springer-Verlag Berlin Heidelberg Hong, T., & Hastak, M. (2006) Constructi on, Inspection, and Maintenance of FRP Deck Panels. Journal of Composites for Constructi on 10 (6), 561-572. 72

PAGE 73

Kalny, O., Peterman, R. J., and Ramirez, G., (2004). Perfo rmance evaluation for repair technique for damaged fiber-reinforced polymer honeycomb bridge deck panels. J. Bridge Eng., 9(1), 75-86. Keller, T., and Gurtler, H. (2005). Quasi-static and fatigue performance of a cellular FRP bridge deck adhesively bonded to steel girders. C ompos. Struct., 70, 484 496. Liu, Z., Cousins, T. E; Lesko, J.J., & Sote lino E. D. (2008). Design Recommendations for a FRP Bridge Deck Support ed on Steel Superstructure. Journal of Composites for Construction, 12 (4), 660-668. Luo, X., Haya, H., Inaba, T., Shiotani, T., and Nakanishi, Y. ( 2004). Damage evaluation of railway structures by using train-induced AE. Construction and Building Materials 18, 215. Majumdar, P. K., Lesko, J. J., Cousins, T. E., and Liu, Z. (2009). C onformable tire patch loading for FRP composite bridge deck. J. Compos. Constr., 13 (6), 575-581. McCall, J. L., Peng, X., Singh, A. P., and Hamilton, H. R., (2011) Hillsboro Canal Bridge Monitoring., FDOT Research Report BDK-75-977-16. OConnor, J. S. (June 2008), GR P bridge decks and superst ructures in the USA. Reinforced Plastics, 26-31. Ohtsu, M, Uchida, M; and Okamoto, T (2002). Damage assessment of reinforced concrete beams qualif ied by acoustic emission. ACI Structural Journal, 99, 411417. Park, K. T., Hwang, K. T., Lee, Y. H., and Kim, S. M (2007). Perform ance verification of a new pultruded GFRP bridge deck-to-girder connection system. Compos. Struct 81 (2007) 114. Physical Acoustics Corporation (2004), AE win software users manual. Rev. 1.90, Physical Acoustic Corporation, Prin ceton Junction., New Jersey, USA. Physical Acoustics Corporation (2008), R15I-AST sensor internal preamplifier acoustic emissions sensor. R ev. 10/05 #81-05, MI STRAS Group Inc., Princeton Junction, New Jersey. Prachasaree, W., Gangarao, H. V. S., and Shekar, V. (2009). Performance Evaluation of FRP Bridge Deck Under Shear Loads. J. Com pos. Materials, 43(4), 377-395 Reeve, S., (2010), FRP br idges-14 years and counting. Reinforced Plastics, 40-44. 73

PAGE 74

74 Telang, N. M., Dumlao, C., Mehr abi, A. B., Ciolko, A. T., and Gutierrez, J. (2006). Field inspection on in-service FR P bridge decks. (No. 564). Transportation Rese arch Board. Turner, M. K., Harries K. A., Petrou, M. F., Rizos, D.(2004). I n situ structural evaluation of a GFRP bridge dec k system. Compos. Struct. 65, 157. Vyas, J. S., Zhao, L., Ansley, M. H., and Xia, J. (2009). Characteriza tion of a low-profile fiber-reinforced polymer deck system for mov eable bridges. J. Bridge Eng., 14 (1), 55-65.

PAGE 75

BIOGRAPHICAL SKETCH Rafael Asencio received an Associate of Science from Santa Fe Community College in Automotive Technologi es, in 2001. Rafael received a Bachelor of Science in Civil Engineering in 2009 and a Master of E ngineering in Civ il Engi neering in 2011 from the Civil and Coastal Engineer ing Department at the University of Florida. 75