multi-reference impact testing of frp bridge deck material

MULTI-REFERENCE IMPACT TESTING OF FRP BRIDGE DECK MATERIAL

Michael S. Lenett(1), Arthur J. Helmicki(2), Victor J. Hunt(1) (1) Department of Civil and Environmental Engineering, University of Cincinnati, Cincinnati, OH 45221 (2) Department of Electrical and Computer Engineering, University of Cincinnati, Cincinnati, OH 45221

ABSTRACT Over the past few years, fiber reinforced polymer (FRP) materials have been introduced into the infrastructure engineering field as potential bridge components. As part of this process, a continuous five-span reinforced concrete deck on steel-girder bridge over the Great Miami River in Dayton, Ohio has been slated to have it’s existing deck replaced with FRP decking material. However, prior to actual replacement, initial laboratory investigations were performed on samples of FRP deck material. These investigations involved small-scale multi-reference impact tests. The purpose of these tests was to determine whether modal experimental and analytical methods could be successfully applied to FRP material and if a proper and accurate measure of in-situ flexibility could be identified from acquired impact data. Following the completion of initial tests, FRP samples were subjected to cyclic loadings and eventually failed. Multireference impact tests were then performed on a damaged sample. Throughout this paper, the authors discuss the before and after damage impact tests performed on this particular FRP sample, present the parameters – natural frequencies, mode shapes, flexibility – identified from these respective tests, and provide a comparison of the before and after damage test results. These results are subsequently used to assess the potential of multi-reference impact testing and modal flexibility as tools for the healthmonitoring/condition assessment of FRP infrastructure systems. NOMENCLATURE [H(z)] [K] [f] {y}r {y}rT

FRF matrix stiffness matrix flexibility matrix rth mode shape transpose of rth mode

{y}r* {y}r*T kr MAr 1.

complex conjugate of rth mode Hermitian of rth mode rth eigenvalue (=rr + jzr) Modal A (scaling) for rth mode

INTRODUCTION

According to Federal Highway Administration inspection statistics, more than one-third of the nation’s 585,000 highway bridges are considered deficient or functionally obsolete [1]. Many deficient bridges are considered as such due to the poor physical condition of their respective deck [2]. Poor deck condition is typically caused by repeating cycles of freeze-thaw as well as rusting of internal reinforcing steel (the increase in steel bar diameter due to rusting introduces crack-inducing tensile stresses within the concrete which, if unchecked, lead to deck spall). Due to such deterioration mechanisms, reinforced concrete bridge decks often need to be replaced within 20 years of service life. This loss of serviceability translates into traffic tie-ups and expensive construction-related deck replacement methods. Recent investigations regarding fiber-reinforced polymer (FRP) composites, however, have revealed that FRP material is higher in strength, lighter in weight, and more corrosion resistant than reinforced concrete [3]. For these reasons, using FRP composite deck systems for the construction of new bridges and retrofit of existing bridges has gained acceptance within the bridge engineering community. Previous research involving reinforced concrete deck on steel-girder bridges has demonstrated that multi-reference impact test methods provide a proper and accurate measure of in-situ structural flexibility which has been used to effectively assess the condition of, and/or identify damage on, such bridges [4,5,6]. Can these same modal-based tools also be used on FRP structural systems for condition assessment/damage identification purposes? To provide preliminary answers to this question, multi-

reference impact tests were performed in a laboratory setting on an FRP deck panel in both an undamaged and damaged state. The major objectives of these tests were to (a) determine whether modal analysis theory, and thus impact test methods, could actually be applied to FRP structural systems, and (b) demonstrate that modal flexibility – the measure of structural flexibility acquired through impact test data – is sensitive to changes in the condition/health of FRP structural systems. 2. FRP DECK PANEL SPECIMEN Prior to a discussion of impact test methods and their application to an FRP system, a description is provided of the actual FRP laboratory specimen to which such methods were applied. The specimen was an FRP panel manufactured with a corrugated core sandwich system. Thin, polyester resin sheets were bonded together with epoxy to form the bottom and top flanges of the panel (Figure 1a). Internal corrugation served as web reinforcement. A cross-section of a similarly manufactured specimen (Figure 1b) revealed that the panel was essentially a single-tier sandwich panel and that the internal corrugation was bonded to both the bottom and top flanges with an epoxy adhesive. Width, length, and thickness dimensions associated with the panel specimen were 36.00 in., 132.00 in., and 9.38 in., respectively. Because this specimen had originally been allocated for load and failure evaluation (i.e., load testing), the intended impact tests had to coordinate with the load test schedule. Impact tests were therefore performed on the panel while it was positioned in the respective load test-frame. This test frame supported the panel in a simply-supported manner and consequently reduced the overall length of the panel to an effective simple span of 120.00 in. Figure 1c presents the panel specimen dimensions and the test-frame setup. 3. IMPACT TESTING OF FRP PANEL SPECIMEN Impact testing of the FRP deck panel specimen was divided into three phases – validation of basic modal assumptions, undamaged panel tests, and damaged panel tests. Figure 2a displays the accelerometer layout used for all three of these phases. This symmetric pattern was established so that (a) strain gages previously mounted to the top flange of the panel (for load test purposes) would not be interfered with, and (b) to avoid steel test-frame members located directly above the simple-supports. All accelerometers were mounted directly to the top flange of the FRP panel with hot glue. The impact hammer in Figure 2b was used to apply impulse (impact) force near the six accelerometer positions highlighted in Figure 2a. These six positions were chosen to ensure the excitation of pertinent modal characteristics, in particular natural frequencies and mode shapes. Through a VXI-based data acquisition system (Figure 2c), all impact force and acceleration response signals were acquired and processed into frequency response functions. The data

acquisition parameters used by the VXI system for all three phases of impact testing were established through a series of trial and error measurements. The pertinent parameters established in this manner are summarized in Table 1.

Frequency Bandwidth Frequency Resolution Coupling Voltage (ADC) Range Windows Number of Averages

0 – 400 Hz 0.125 Hz AC within ! 2V Force Window Width = 0.1 sec Exp. Time Constant, t = 2 sec Five

Table 1 – Data Acquisition Parameters for FRP panel impact tests

3.1. Verification of Modal Methods Prior to performing a rigorous impact test on the undamaged panel specimen, it must be verified that the panel, like any structure subjected to experimental and analytical modal methods, satisfies the basic modal assumptions of linearity, observability, and time-invariance. If the panel does not comply with these behavioral characteristics, modal methods, such as impact testing and modal flexibility, are inapplicable. 3.1.1. Linearity Maxwell-Betti’s principle of reciprocity may be used to verify system linearity. In terms of frequency response functions, this principle may be expressed as Hpq = Hqp. Figure 3 displays a typical example of panel specimen reciprocity. This and other similar examples of observed reciprocity sufficiently satisfy Maxwell-Betti’s principle for a linear system. 3.1.2. Observability A system is considered observable when input-output measurements can be used to define system parameters. Note that the frequency response functions obtained from impact testing of the panel are based on the measured impact forces and the corresponding measured acceleration responses of the panel. The peaks within any frequency response function (refer to Figure 3) are indicative of the damping, natural frequencies and mode shapes of the test structure or system. Consequently, a group of measured frequency response functions may be used to identify the modal parameters of the panel. Since the input-output measurements that are made throughout impact testing can be used to accomplish such identification, the FRP deck panel in its respective test setup (Figure 1c) may be interpreted as an observable system.

3.1.3. Time-Invariance Time-invariance, or stationarity, of the undamaged FRP deck panel specimen was evaluated regardless of the fact that the panel was located within an enclosed laboratory with uniform environmental and temperature conditions. As a means of verifying time-invariance, frequency response functions obtained from different tests, yet corresponding to the same measurement points, were superimposed. Figure 4 displays an example of a time-invariance check for the undamaged panel. This comparison, as well as comparisons involving other frequency response functions (frequency response functions associated with other measurement points), revealed that there were no significant shifts along either the frequency or magnitude (and phase) axes. In particular, there was no shifting in the vicinity of the peaks (refer to Figure 4). Consequently, time-invariance checks identified the undamaged FRP deck panel specimen, while positioned within the ambient friendly laboratory, as a timeinvariant, or stationary, system. Because the undamaged FRP panel was evaluated as a linear, observable, and timeinvariant structure, it was deemed permissible to model it as a modal system and subject it to modal methods such as impact testing. Furthermore, these results imply that it is possible to perform modal tests on other structures comprised of FRP composite materials. 3.2. Undamaged Panel Tests Multi-reference impact test data was used to identify mode shapes at natural frequencies ranging from 48.42 Hz to 370.89 Hz. Several of these modes are depicted in Figure 5. The identified mode shapes, their respective natural frequencies, damping, and scaling were subsequently substituted into the following equations to compute modal flexibility:

[H (ω = 0)] =

1 = [ f ] = flexibility [K ]

* *T  {ψ } {ψ }T { ψ }r {ψ }r r r [f ]= ∑ + * M Ar − λ*r r =1  M Ar (− λ r )  N

(

  

)

(1)

(2)

Equations (1) and (2) reveal that evaluating the mathematical expression for a frequency response function matrix at z = 0 yields the expression for modal flexibility. Modal flexibility defined in this manner may be interpreted as the in-situ flexibility of the tested structural region [6, 7]. The reliability of the resulting undamaged panel modal flexibility, as well as the reliability of the identified modal parameters and impact test data, was evaluated by comparing measured static load test displacements with simulated displacements. Simulated displacements were computed by multiplying the modal flexibility matrix [ f ] with a load vector {P} that is based on the magnitude, direction, and position of loads applied during actual load tests of the panel. Comparisons between measured and

simulated displacements are shown in Figure 6 for two different load cases. Note that both cases display good correlation between the measured and simulated profiles. Such correlation reveals that the modal flexibility obtained from the undamaged panel impact test data provides a reliable and structurally relevant measure of in-situ panel flexibility. 3.3. Damaged Panel Tests The FRP deck panel specimen was load tested to failure after completion of the undamaged panel impact tests. Damage induced by failure was essentially delamination at the bottom flange (Figure 7). While in a damaged, or failed, state, it was intended that the panel would be subjected to a second round of impact tests. Through modal linearity (by means of reciprocity), observability, and time-invariance checks, it was verified that the damaged panel could be modeled as a modal system and therefore tested with modal impact methods. The data acquired during impact testing of the damaged panel was used to identify mode shapes at natural frequencies ranging from 49.40 Hz to 241.19 Hz. Figure 8 displays several of these modes. The identified modal parameters were then used to define the modal, or in-situ structural, flexibility of the damaged panel. To determine whether modal flexibility is sensitive to the condition/health of an FRP structural system, displacement profiles based on modal flexibility were defined for both the undamaged and damaged panel states and subsequently compared. For purposes of comparison, undamaged and damaged displacement profiles along a sensor line were used. These profiles were obtained by multiplying the respective modal flexibility matrix with a load vector comprised of virtual downward loads of unit magnitude positioned at each measurement location along a particular sensor line (refer to Figure 9). Comparing the undamaged and damaged profiles obtained in this manner (Figure 9) revealed observable anomalous changes in the displacement profiles which were, at particular locations, accompanied by a 10% or greater change in deflection magnitude (near the vicinity of visible damage, this change was greater than 20%). Through these comparisons, it was consequently revealed that modal flexibility is sensitive to changes in the structural condition of FRP systems. 4. CONCLUSIONS Performing multi-reference impact tests on an FRP deck panel specimen prior to and after inducing damage revealed: (a) an FRP deck panel specimen could be modeled as a linear, observable, and time-invariant system, thus permitting application of impact test methods; (b) modal flexibility acquired through impact test methods provided a good measure of the FRP systems in-situ structural flexibility; and (c) displacements simulated with modal

flexibility were sensitive to damage within the FRP structural system. These results imply that structures utilizing and comprised of FRP composite materials, such as bridges, can have their behavior and condition evaluated/monitored through application of impact test techniques. This concept is currently being evaluated on a five-span bridge crossing the Great Miami River in Dayton, Ohio. This bridge, in its original state, consisted of built-up steel girders and a reinforced concrete deck. At the time of this writing, baseline impact and truck-load tests have been performed on the original bridge. The reinforced concrete deck was removed after completion of these tests and FRP deck systems are currently being installed. Once all FRP deck systems are installed, follow-up impact and truck-load tests will be performed. In order to evaluate how FRP deck systems influence/alter the behavior of multi-span, multigirder (steel girder) bridges, the results of the baseline and follow-up (or retrofit) tests will be compared. Furthermore, periodic tests will be performed on the retrofitted bridge in order to monitor and evaluate possible changes in FRP deck and overall bridge behavior. The results of the baseline, retrofit, and periodic bridge tests will be presented at a future date.

Hunt, V., and Yao, J., Objective Global Condition Assessment, Proceedings of the 15th International Modal Analysis Conference, February 1997. 6.

Lenett, M., Hunt, V., Helmicki, A., Brown, D., Catbas, F.N., and Aktan, A.E., Condition Assessment of Commissioned Infrastructure Using Modal Analysis and Flexibility, Proceedings of the 17th International Modal Analysis Conference, February 1999.

7.

Catbas, F.N., Lenett, M., Brown, D.L., Doebling, S.W., Farrar, C.R., and Turer, A., Modal Analysis of Multi-Reference Impact Test Data for Steel Stringer Bridges, Proceedings of the 15th International Modal Analysis Conference, February 1997.

Polyester Resin Sheet

Epoxy

5. ACKNOWLEDGEMENTS

Epoxy

This research was supported by the Ohio Department of Transportation, Research and Development. The encouragement and support extended by Messrs. Fagrell, Morton, Dalal, and Green of the Ohio DOT is gratefully acknowledged. Access to FRP panel specimens and successful coordination of impact test and load test schedules was made possible by the cooperation of Dr. Harik of the University of Kentucky. For his time and effort, the authors are deeply appreciative.

Figure 1a – FRP deck panel flange construction

Corrugated ICI - CCTICore

6. REFERENCES 1.

Parsons Brinckerhoff, Bridge Rehabilitation, A Practical Guide. Sons, Inc., New York, 1993.

Inspection and John Wiley and

2.

Dunker, K.F., and Rabbat, B.G., Assessing Infrastructure Deficiencies: The Case of Highway Bridges, Journal of Infrastructure Systems, Vol. 1, No. 2, pp. 100-119, 1995.

3.

Civil Engineering Research Foundation (CERF). Composites Durability Assessment Research. http://www.cenet.org/research/summary/composit.htm. Accessed July 1999.

4.

Lenett, M., Global Condition Assessment Using Modal Analysis and Flexibility, Doctoral Dissertation, University of Cincinnati, 1998.

5.

Aktan, A.E., Brown, D., Farrar, C., Helmicki, A.,

Figure 1b – Cross-section of FRP deck panel

FRP Deck Panel

A

A

36.00 in.

120.00 in. 132.00 in.

9.38 in. FRP Deck Panel

Figure 2c – VXI data acquisition system

View A-A

Figure 1c – Panel dimensions and test-frame setup

8.00 in.

0 .0 16

. in

0 .0 24

. in

. . . in in in 0 0 0 .0 2.0 .0 2 4 1 1 2

0 .0 16

in

1

2

3

4

5

6

7

8

9

10

11

12

13

14

8.00 in.

= Accelerometer Position

12.00 in.

= Accelerometer and Impact Position

36.00 in.

12.00 in. 12.00 in.

Observable peaks

= Accelerometer Position = Accelerometer and Impact Position 120.00 in. Magnitude (g/lbf)

1E-1

132.00 in.

1E-2 1E-3 1E-4

H13,2

H2,13

1E-5 1E-6 0

50

100

150

200

250

300

350

400

Frequency (Hz)

Figure 3 – Undamaged panel reciprocity and observability

Figure 2a – Sensor Layout

Magnitude (g/lbf)

1E-1 1E-2 1E-3 1E-4

H3,2 (Trial 1)

H3,2 (Trial 2)

1E-5 1E-6 0

50

100 Frequency (Hz)

150

200

Figure 4 – Time-invariance check for undamaged panel

Figure 2b – Impact hammer used throughout panel tests

Mode 1 - 48.42 Hz

0

Distance (in.)

Line 2

Line 1

Line 2

Line 1

0

Distance (in.)

Distance (in.)

Distance (in.)

00 0. 12 00 8. 10 0 .0 96 0 .0 84 0 .0 72 0 .0 60 0 .0 48 0 .0 36 0 .0 24 0 .0 12

0

00 0. 12 00 8. 10 0 .0 96 0 .0 84 0 .0 72 0 .0 60 0 .0 48 0 .0 36 0 .0 24 0 .0 12

0

Distance (in.)

00 0. 12 00 8. 10 0 .0 96 0 .0 84 0 .0 72 0 .0 60 0 .0 48 0 .0 36 0 .0 24 0 .0 12

Distance (in.)

Mode 4 - 164.9946 Hz

Mode 3 - 100.37 Hz

Mode 1 - 49.40 Hz

00 0. 12 00 8. 10 0 .0 96 0 .0 84 0 .0 72 0 .0 60 0 .0 48 0 .0 36 0 .0 24 0 .0 12

0

00 0. 12 00 8. 10 0 .0 96 0 .0 84 0 .0 72 0 .0 60 0 .0 48 0 .0 36 0 .0 24 0 .0 12

00 0. 12 00 8. 10 0 .0 96 0 .0 84 0 .0 72 0 .0 60 0 .0 48 0 .0 36 0 .0 24 0 .0 12

0

Mode 6 - 155.6501 Hz

Mode 4 - 79.75 Hz

Line 1

Line 1

Line 2

Line 2

Visible Damage 120.00 in.

120.00 in.

Figure 8 – Damaged panel mode shapes

Figure 5 – Undamaged panel mode shapes

Unit Loads (1 lb.) at each measurement point on Line 1

Distance (in.)

00 0. 12 00 8. 10 0 .0 96

Displacement (in.)

Displacement (in.)

0 .0 84

120.00 in.

00 0. 12 00 8. 10 0 .0 96

120.00 in.

0 .0 84

-0.14

0 .0 72

-0.12

0 .0 60 0 .0 48

-0.10

0 .0 36

Line 2

0 .0 24

Line 1

-0.08

0

1 lb./point

0 -0.02 -0.04 -0.06

Distance (in.) 0

0 .0 12

00 0. 12

00 8. 10 0 .0 96

0 .0 84

0 .0 72

Line 2

0 .0 60

Line 1

0 .0 48

6,018.5 lb.

0 .0 36 0 .0 24

0

0 .0 12

Total Loading = 12,037 lb.

-0.00001 -0.00002 -0.00003 -0.00004 -0.00005 -0.00006 -0.00007

14% variation Simulated displacements for Line 1 Simulated displacements for Line 2 Actual measured displacements (along centerline of panel)

Undamaged Panel Damaged Panel

Unit Loads (1 lb.) at each measurement point on Line 2

Distance (in.)

0 .0 72

0 .0 36

Line 2

-0.30 -0.35

120.00 in.

Displacement (in.)

Line 1

-0.15 -0.25

0 .0 60 0 .0 48

0

-0.05 -0.10 -0.20

0 .0 24

Displacement (in.)

Distance (in.) 0

1 lb./point

0

0 .0 12

00 0. 12

0 .0 84

0 .0 60

120.00 in.

0 .0 72

Line 2

0 .0 48

Line 1

0 .0 36 0 .0 24

13,500 lb.

0 .0 12

0

00 8. 10 0 .0 96

Total Loading = 27,000 lb.

-0.00001 -0.00002 -0.00003 -0.00004 -0.00005 -0.00006 -0.00007

> 20% variation

Figure 6 – Comparison between measured and simulated displacements for undamaged panel

120.00 in.

Line 1 Line 2

1 2

3 4 5

8 9

10 11 12

6

7

Delamination

13 14

Region of Visible Damage (Delamination at Bottom Flange)

ICI 1

Figure 7 – Visible panel damage

Figure 9 – Sensitivity of modal flexibility to panel damage