Implementation of Digital Image Correlation

for Structural Health Monitoring of Bridges

C. NONIS, C. NIEZRECKI, T.-Y. YU, S. AHMED, C.-F. SU

and T. SCHMIDT

ABSTRACT

As the civil infrastructure (e.g. bridges) of the global highway s ystem ages,
there exists a critical need for structural health monitoring over large areas that is
robust, inexpensive and easily interpreted. According to the American Society of
Civil Engineers, in 2008, 161,892 bridges were structurally deficient or obsolete. In
the next fifteen years, nearly half of Americas bridges will exceed the t ypical 50-
year design life. Thorough inspection of these aging bridges is important to ensure
public safety. Currentl y, bridge health is assessed primarily using qualitative visual
inspection, which is subject to variability due to an inspectors interpretation.
Instruments such as strain gages, accelerometers, fiber optic sensors, displacement
transducers, etc. are becoming more common for monitoring bridge health. These
sensors require external power, cabling/antenna for data transmission, high data
acquisition channel counts and only measure at a discrete point or along a line
(making it difficult to detect damage outside the sensors proximity). To rectify
these drawbacks, this paper presents research results of using three-dimensional
(3D) digital image correlation (DIC) as a new approach for quantitative bridge
structural health monitoring.
3D DIC is a non-contact, full field, optical measuring technique that uses two
digital cameras to measure surface geometry, displacement, and strain. Long term
monitoring with DIC can be accomplished by imaging the bridge periodically and
computing strain, displacement and surface geometry from
images recorded at different dates. In this paper, DIC is shown to
1) quantify spalling by comparing subsequent surface geometry measurements,
2) monitor crack width using extensometers established between photogrammetric
targets and 3) successfully locate non-visible cracks using full-field strain and
displacement measurements. These techniques are first confirmed in laboratory
tests.
Next field measurements are made on three operational full-scale bridges. This
paper also discusses the challenges and solutions to effectively implementing DIC
_____________
Christopher Nonis, Christopher Niezrecki, Tzu-Yang Yu, Shafique Ahm ed, Che-Fu Su,
University of Massachusetts Lowell, 1 University Avenue, Lowell, MA 01854, U.S.A.
Tim Schmidt, Trilion Quality System s, 500 Davis Drive, Plymouth Meeting, PA

679
on large structures in a field test environment. The results of this study demonstrate
DIC is an effective tool for structural health monitoring of large scale civil
infrastructure.

INTRODUCTION

As the civil infrastructure (e.g. bridges) of the global highway system ages,
there exists a need to perform structural health monitoring and inspection over large
areas, on a wide scale that is robust, inexpensive, and easy to interpret. In the next
fifteen years, nearly half of America's bridges will exceed the 50-year design life
[1]. As a result, in 2008, 161,892 bridges were classified as being structurally
deficient or obsolete [2].
Thorough inspection of bridges is paramount to preventing bridge failures and
ensuring public safety. Currently, bridge health is assessed primarily using visual
inspection which is subject to significant variability due to inspector interpretation
[3]. Also, some types of damage are difficult to detect or quantify using visual
inspection.
As a result, more advanced non-destructive test methods have been
implemented to help assess bridge health. Instruments such as strain gages,
accelerometers, fiber optic sensors, displacement transducers, etc. are becoming
more common in structural health monitoring [4, 5]. These types of sensors
generally possess several drawbacks such as: requiring external power,
cabling/antenna for data transmission, high data acquisition channel counts, and
they only measure at discrete points or along a line. Also, if the damage is outside
the proximity of the sensor, the damage may be difficult to detect.
Three-dimensional digital image correlation (3D DIC) is an evolving
measurement technique that has only very recently been proposed to enhance
bridge inspection. 3D DIC is a full field, non contact optical measuring technique
that uses two digital cameras to measure surface geometry, displacement, and
strain. All DIC analysis in this work was performed using ARAMIS software and
hardware (GOM mbH). To perform these measurements, a stochastic pattern is
applied to the surface of interest and a series of photographs (stages) are taken by
both cameras as the surface deforms. Strain and displacement are computed by
comparing the stochastic pattern of the deformed surface to the initial reference
measurement of the pattern. Three-dimensional information is extracted using the
principles of stereophotogrammetry. DIC can be used for structural health
monitoring by comparing current surface geometry, displacement, and strain
measurements to baseline measurements made on previous dates. Monitoring these
measurements over time will allow inspectors to quantify spalling, monitor crack
widths and measure full-field strain and displacement.
A very limited number of papers have been published involving field tests of
bridges using point tracking and DIC. For discrete displacement measurements,
points of interest were marked with photogrammetric targets and the targets
displacement was recorded while loading the bridge [6-11]. Full field displacement
was measured by tracking the natural surface pattern of concrete [12] or an applied
pattern [9, 13]. In general, it was concluded the DIC system accuracy is
comparable to existing displacement measurement techniques and DIC is an easier
way to measure displacement of multiple points at once.

680
 Researchers have also measured localized full field strain and displacements on
bridges with 2D DIC. Sas et al. [14] recorded the strain on a concrete girder during
a full scale bridge failure test and Kntz et al. [15] measured the displacement field
on a cracked concrete girder during a bridge loading test. In both cases DIC was
able to locate cracks. Both papers demonstrate the advantage of inspecting concrete
bridges with DIC.
The prior DIC bridge tests recorded strain and displacement changes in
response to an applied load over a relatively short time period (several hours) and
primarily using 2D DIC. To the authors knowledge, this is the first research effort
to explore using 3D DIC to monitor strain and displacement of bridges due to
damage or degradation over several months or longer. Also, for the first time,
concrete spalling is monitored by comparing surface geometries measured with 3D
DIC using a projected pattern. These damage detection capabilities are first
validated on small scale laboratory tests and then utilized on real bridges in the
field. The results of these experiments are discussed as well as some of the
limitations and challenges of using DIC for bridge monitoring.

LABORATORY TESTS OF DIC DAMAGE DETECTION CAPABILITIES

Initial testing of the DIC damage detection capabilities were conducted in a

laboratory setting. Specifically, the ability of DIC to measure strain and
displacement, quantify spalling, and locate cracks was investigated.

Laboratory Reinforced Concrete Beam Test:

A four point bend test on a reinforced concrete (RC) beam was used to evaluate
the strain and displacement measurement and crack detection capabilities of DIC.
The dimensions of the concrete beam are 35 in x 6 in x 6in with two 0.5in diameter
steel reinforcing bars near the bottom surface.
Figure 1 shows axial strain contour plot of the RC beam subjected to four point
bending at 33% of the failure load. The DIC strain gage length is ~11mm.

Figure 1. Axial strain contour plot of RC beam subjected to 4 point loading at 33% of failure load

The beam bending test results (see Figure 1) show that it is possible to detect
cracks based on areas of amplified strain. The beam was visually inspected, at this
loading level and the cracks were not visible to the human eye. When processing

681
DIC data to detect cracks, the DIC strain gage length should be relatively short to
yield better spatial resolution.

Laboratory Spalling Quantification Tests:

Spalling quantification was tested on a small, 30cm x 30cm x 30cm, concrete

block. The surface geometry was measured before and after concrete was chipped
away in two locations. To quantify the spalling, the deviation between the initial
surface and damaged surface was computed and is plotted on left side of Figure 2.
The deviation between the two surfaces was computed in GOM TM SView. The
software performs an automated best-fit alignment of the two geometries and
calculates the deviation between the two surfaces.
The traditional paint patterning technique for DIC is not feasible for spalling
quantification, because the pattern would fall off with the spalled concrete. To
circumvent this issue, a pattern was projected on the surface with an LCD projector.
Note that projected patterns cannot measure strain, because the pattern does not
deform with the surface.

Surface Deviation Contour Plot Photo of Damaged Surface

Figure 2. Contour plot of deviation between damaged and undamaged surfaces, and photograph of
damaged surface; red circles denote the locations of induced damage

From the contour plot, the locations and extent of material loss can be
determined. Surface geometry data is missing for some areas of the surface due to
cavities on the surface where pattern matching was difficult to perform.
The results of these laboratory tests confirm DIC can measure strain and
displacement, locate cracks and quantify spalling. These laboratory tests serve as
the basis for the field tests.

BRIDGE MONITORING USING DIGITAL IMAGE CORRELATION

Fifteen areas on three bridges near Lowell, Massachusetts were monitored to

demonstrate spalling quantification, crack monitoring using photogrammetric
targets, and long term strain and displacement monitoring. Only results from four
representative areas are presented in this paper.

682
Field Spalling Quantification Tests:

Spalling of a bridge abutment was monitored over 77 days using DIC with a
projected pattern. Spalling is quantified by computing the deviation between
surface geometries measured at different dates. Figure 3 shows a photograph of the
spalling area and surface deviation contour plot.

Surface deviation contour plot of spalling abutment Photo of spalling abutment

Figure 3. Photograph (right) and corresponding surface deviation contour plot (left) of spalling test
on abutment. Largest areas of spalling are indicated by the red ovals.

The regions of large negative deviation specify the location and extent of the
spalling. These results confirm that DIC can be used as a long term monitoring tool
to quantify spalling.

Field Photogrammetric Target Extensometer Tests:

Photogrammetric targets were used to monitor cracks in a bridge abutment and

relative motion between the abutment and a retaining wall. Simple point to point
extensometers are used for monitoring. Figure 4 shows a photograph of the
measurement area with the points and extensometers labeled. The abutment was
patterned for strain measurements, but was not used for this analysis.

Figure 4. Photograph of abutment with nine photogrammetric targets denoted by blue Xs and
extensometers between targets indicated by color coded lines.

683
 The abutment monitoring area was measured on four dates, September 14,
2012, October 11, 2012, February 5, 2013 and April 24th, 2013. On each date, each
of the extensometers were measured ten times and averaged to compensate for
noise. Figure 5 plots the average deviation from the initial length of each
extensometer, for the four dates.

Figure 5. Length deviation of photogrammetric target extensometers monitoring joint and cracks
widths on a bridge abutment

The solid lines represent length deviations of extensometers measuring across

the joint between the abutment and retaining wall, and the dashed lines represent
measurements across cracks. Figure 5 shows the joint between the abutment and
retaining wall opened to a maximum of approximately 2.5mm in February and at
last measurement (4/24/2013) was approximately 1.5mm wider than the initial joint
width measured in September. The opening of the joint is likely caused by thermal
contraction, not damage. Both cracks have opened approximately 0.5mm and are
currently fluctuating around that width. The crack widths do not appear to correlate
with temperature and therefore could indicate some permanent deformation from
damage has occurred. These results show that long term monitoring of relative
displacements can be performed using photogrammetric targets.

Long Term Full-field Displacement and Strain Monitoring:

Long term full-field strain and displacement is computed by comparing

measurements of the monitoring area from different dates. The facets from the
initial/reference image are found in the subsequent images from different dates.
Displacement and strain are calculated based on the deformation of the facet field
from the initial/reference facet field. Full-field strain and displacement was
monitored over 145 days on a retaining wall with a preexisting vertical crack, see
Figure 6.

684
 Horizontal Displacement Horizontal Normalized Strain

Figure 6. Full-field horizontal displacement and strain of retaining wall, monitoring period
September 13th, 2012 to February 5th, 2013

The displacement plot shows the two sides of the crack have moved apart
approximately 1.4mm. The X strain contour plot also indicates the crack has
expanded, as evident by the intense regions of positive strain surrounding the crack.
The X strain was normalized because the strain values were unrealistically high due
to short strain gage lengths (~70mm), minimal spatial filtering and high noise
levels. High noise levels result from errors matching facets in images from
different dates. Nevertheless, the strain distribution appears accurate and clearly
indicates the crack locations.
In this area, the presence of a crack was obvious visually, but in other areas new
cracks were discovered using the full-field displacement and strain. A crack was
detected in a bridge pier by monitoring the full-field strain and displacement over
223 days, see Figure 7. The DIC strain gage length is ~47mm.

Horizontal Displacement Normalized Horizontal Strain

Figure 7. Full-field horizontal (X Direction) displacement and normalized strain contour plots of
bridge pier. Monitoring period September 13, 2012 to April 24, 2013

685
 The crack in the pier is detected by the discontinuity in the full-field
displacement or by the region of amplified strain in the full-field strain. The crack
width is on the order of 0.1mm and was not detected when the area was visually
inspected. The results of these two areas demonstrate the advantage of using 3D
DIC; the growth of existing cracks or development of new cracks can be monitored
and DIC results are relatively simple to interpret.

CONCLUSION

The effectiveness of three-dimensional digital image correlation for bridge

structural health monitoring has been demonstrated. In the laboratory, DIC located
non-visible cracks in a reinforced concrete beam using the axial strain contour plots
and quantified spalling of a concrete block. In field testing, spalling quantification
of a full scale bridge abutment was performed using a projected pattern to measure
surface geometry. Using photogrammetric targets, the opening of joints and cracks
were tracked over 222 days. Full-field strain and displacement was monitored on a
retaining wall and pier. Expansion of existing cracks and development of new
cracks were quantified with the long term full-field results. Future work will focus
on understanding how the natural, benign changes in a bridge over time (e.g.
seasonal temperature fluctuations) can be discriminated from mechanical damage.
The results of this paper show that DIC has great potential for quantitative bridge
inspection. Although applications discussed in this paper are bridge related, these
same techniques can be applied to numerous other large scale concrete or metal
structures.

ACKNOWLEDGEMENTS

The authors gratefully appreciate the financial support for this work provided by
the U.S. Department of Transportation (Grant No. RITARS-11-H-UML). Any
opinions, findings, and conclusions or recommendations expressed in this material
are those of the authors and do not necessarily reflect the views of U.S. DOT.

REFERENCES
1. Bridging the Gap, American Association of State Highway and Transportation Officials,
(2008).
2. Infrastructure Fact Sheet Transportation Bridges, American Society of Civil Engineers,
(2009).
3. Reliability of Visual Inspection for Highway Bridges, United States Department of
Transportation Federal Highway Administration, 476-481 (2001).
4. Damage Identification and Health Monitoring of Structural and Mechanical Systems from
Changes in Their Vibration Characteristics: A Literature Review, Los Alamos National
Laboratory, (1996).
5. A Review of Structural Health Monitoring Literature: 1996-2001, Los Alamos National
Laboratory, (2003).
6. Chiang, C., "Displacement Measurements of Highway Bridges Using Digital Image Correlation
Methods," SPIE, (2011).
7. Albert, J., "Pilot Studies on photogrammetric bridge deformation measurements,"
8. Santini-Bell, E., "Digital Imaging for Bridge Deflection Measurement of a Steel Girder
Composite Bridge," Transportation Research Board, (2011).
9. Brogan, P., "Digital Image Correlation Application to Structural Health Monitoring," University
of New Hampshire, (2010).

686
10. Gamache, R., "Non-instrusive Digital Optical Means to Develop Bridge Performance
Information," Non-Destructive Testing in Civil Engineering, (2009).
11. Malesa, M., Monitoring of Civil Engineering Structures Using Digital Image Correlation
Technique, International Conference on Experimental Mechanics, (2010).
12. McCormick, N., Structural Health Monitoring Extreme Condition Monitoring, National
Physical Laboratory
13. Yoneyama, S., Bridge Deflection Measurement Using Digital Image Correlation,
Experimental Techniques, (2007).
14. Sas, G., Photographic Strain Monitoring During Full-Scale Failure Testing of Ornskoldsvik
Bridge, Structural Health Monitoring, (2012)
15. Kntz, M., Digital Image Correlation Analysis of Crack Behavior in a Reinforced Concrete
Beam During a Load Test, Canadian Journal of Civil Engineering, (2006)

687