17TH STREET BRIDGE CANOPY FAILURE INVESTIGATION
Atlanta, Georgia

Project Background
17th Street Bridge
3-span trapezoidal, variable depth steel box girders with cast-in-place concrete deck
830 ft. long, 137 ft. wide Connects Midtown Atlanta to Atlantic Station Spans over 20 traffic lanes (Interstate 75/85
Downtown Connector) Carries 6 lanes of traffic and features
pedestrian sidewalks on north and south side
Construction completed in 2004
Engineer of Record URS Corporation General Contractor C.W. Matthews Contracting
Co., Inc.

Overall view of 17th Street Bridge

Canopy-Fence Structure
Attached to south parapet wall Structural support frames
Galvanized steel Built-up column and cantilever beam
assemblies 64 total along south side (with 5 additional
column assemblies) Spaced at 10'-6" on center
Architectural shade screen
Stainless steel tube shapes bolted to frames Perforated 14 gauge stainless steel sheets

Canopy fence along south parapet of bridge

Canopy-Fence Structure

Section through bridge looking east

Elevation view of canopy fence

Canopy-Fence Structure
Undulating appearance due to varying support frame geometries
Pitch of cantilever arms varied
Type A Type B Type C Type D
Overall size and geometry of individual elements unchanged

Canopy frame geometries

Canopy-Fence Structure
Epoxy adhesive anchors
7/8" diameter 4 anchors per frame Anchored to outside (south) face of concrete parapet
through flange in column assembly No specification for anchor material or adhesive system No specification for hole diameter or embedment depth Sufficient embedment to develop tension service load
of 4,000 lb. per anchor
Adhesive anchor

Canopy Failure
August 13, 2011 at 11:20 PM
Approximately 7 years after construction
190 ft. long section of canopy-fence detached from south parapet and fell onto roadway below
19 canopy support frames along east end fell
No injuries and only minor vehicle damage reported
GDOT immediately removed remaining canopy structure on south side

Canopy frames on roadway below

Canopy Failure

Field Investigation
Visual assessment
Concrete deck Parapet Anchor holes Canopy frames
Materials sampling for laboratory study
In-situ load testing of anchor rods

Field investigation

Visual Assessment Concrete Deck

Distress marks and scrapes on south sidewalk
Indicate inward rotation and impact of frames during failure
Land surveyor contracted to survey locations and extents of distress on bridge deck

Distress marks at Location 54

Distress marks at steel pedestrian perch

Visual Assessment Concrete Deck

Distress marks at Location 55

Distress marks at Location 56

Visual Assessment Concrete Deck

Visual Assessment Parapet

Spalls at anchor holes at Location 58

Spalls at top of parapet at Location 66

Visual Assessment Anchor Holes

Within failure area, anchors pulled out
Exception: lower 2 rods at Location 67 shear failure of rods
Inspected using optical borescope and videoscope
Average depth of embedment = 8" Anchor holes cored Diameter of holes = 1-1/8" Holes spacing as shown in design drawings

Borescope inspection

Videoscope inspection

Visual Assessment Anchor Holes

Epoxy observations
Generally bonded to substrate concrete
Frequent voids/air pockets, especially in back 1.5" of hole
Varied in color
Opaque dark gray Opaque light gray Translucent brown
Wet epoxy components extracted at 3 locations

Wet epoxy component extracted from inside hole 56B

Visual Assessment Anchor Holes
Videoscope Observations

Visual Assessment Anchor Holes

Anchors outside failure area were torch cut during removal
168 torch-cut rods inspected
112 observed with measurable withdrawal (pull-out)
Minimum 1/16"
Maximum 1-3/4"
Scorch marks indicate torch cut between column and parapet at some locations

Anchor withdrawal away from failure

Torch-cut anchor away from failure

Visual Assessment Canopy Frames

Observed in medians below bridge and later at off-site storage yard
19 frames in failure area
Types of distress observed:
Bent and missing connection plates Failed/bent/twisted plates in
column/cantilever assemblies Bent anchor rods Miscellaneous paint scrapes
Field-modified holes for anchor rods

Missing connection plates on cantilever

Field-modified hole for anchor rod

Visual Assessment Canopy Frames

Other observations

Top Left: Black paint marks on Frame 52
Top Right: Bent anchor rods

Bottom Left: Welded plates for eccentric anchorage on Frame 51
Bottom Right: Yellow paint on bottom of cantilever tube assembly

Materials Sampling
Wet epoxy materials from holes 6 concrete cores from south face of
parapet (at anchor hole locations) 4 concrete cores from north face of
parapet
3 for compressive strength testing 1 for petrographic study

Coring on south face of parapet

In-Situ Load Testing
Direct tension testing Performed on 5 existing anchors
away from failure area Tested rods showed no evidence of
prior withdrawal Only 1 of 5 rods failed below
capacity of test equipment (~15,000 to 20,000 lb.)

Load testing of anchor rod

In-Situ Load Testing

Results of in-situ load tests

Anchor Maximum Applied Load

Number

(lb.)

13A

10,436

33A

17,634

33B

21,094

34A

20,374

41A

15,619

Anchor tested to failure Load test stopped prior to failure

Design 4,000 lb. service load

Anchor 13A tested to failure Data collection unit for load testing

Laboratory Investigation
Petrographic studies of concrete
Common normal weight concrete Moderately week paste-aggregate bond
Compressive strength testing of concrete
Average compressive strength = 6,787 psi
Assessment of adhesive material
Bisphenol A/Glycidyl Ether-based epoxy Poor mixing Poor proportioning Incomplete filling of holes

Core 55d split for visual assessment

Laboratory Observations
Core 50d split showing incomplete filling of anchor hole

Laboratory Observations
Various anchor rods collected from debris

Analysis
Estimate actual tensile loads on anchors during service
Estimate of typical frame self-weight = 2,075 lb.
Estimate of tributary weight of shade canopy for single frame = 550 lb.
Total = 2,625 lb.
Design service = 4,000 lb.
Center of gravity located approximately 2'-6" inboard of south face of parapet
Direct tension on lower anchors under gravity loading

2'-6"

Lower anchor bolts in tension

Elevation view of canopy fence

Analysis

Estimate of tensile reaction at single bottom anchor due to overturning moment

(self-weight only) for various assumed bottom anchor withdrawals (frame

rotations).

Tension @ Bottom Anchors

Service Load, P
(lb)

anchor (in)

Tensile reaction at bottom anchor, Tbot (lb)

1600 1500

2,625

0

1,053

1400

2,625

0.5

2,625

1.0

2,625

1.5

1,107 1,161 1,216

T (lb)

1300 1200

2,625

2.0

1,268

1100

2,625

3.0

2,625

4.0

1,376 1,483

1000

0

1

2

3

4

anchor (in)

Collapse

Collapse

Collapse

Collapse

Collapse

Collapse

Collapse

Collapse

Collapse Sequence
Canopy support frames experienced significant inward rotation as lower anchors partially withdrew due to sustained tensile loading and creep.
At time of failure, lower anchors at Frames 55 and 56 were first to fully release.

Visual Assessment Concrete Deck

Collapse Sequence
Canopy support frames experienced significant inward rotation as lower anchors partially withdrew due to sustained tensile loading and creep.
At time of failure, lower anchors at Frames 55 and 56 were first to fully release.
Frames west of 55 and east of 56 fell in opposite directions due to restraint from eccentrically supported Frame 51.
Upper anchors were subjected to bending and increased tensile demand as frames rotated and eventually pulled free.

Collapse Sequence
Frames pulled adjacent frames as they rotated inward. Once free of parapet, frames began to translate (slide) back
over the parapet to the south. Moving away from point of initiation, frames experienced
increased translation as they were pulled south by adjacent frames while rotating. Collapse arrested at Frame 48 to west
Rate of loading Quality of mixing of epoxy adhesive

Conclusion
Primary causes of canopy failure related to epoxy anchor adhesive:
Poor resistance to long-term creep Sustained tensile loading
Secondary causes
Disproportionate mixing of adhesive components Incomplete mixing of adhesive components Inclusion of air voids High temperature in service

Conclusion
Over time, the anchors connecting the supporting frames withdrew as the epoxy adhesive deformed and failed under a sustained tensile load that was substantially lower than the design service load.
Poor performance of epoxy anchors was similar to that observed by NTSB on collapse of ceiling panels in I-90 tunnel in Boston, July 2006.

Conclusion
Following collapse in Boston, design community has become more aware of poor performance of adhesive anchors under sustained load.
17th Street Bridge was designed and constructed approximately 2 years before the effect of sustained loading on adhesive anchors was identified by NTSB on the failure in Boston.

Recommendations
If GDOT elects to re-install the canopy structure on the south parapet, the use of a positive connection is recommended to secure the frames to the parapet wall.
If GDOT elects to leave the fence system on the north parapet in place, the existing adhesive anchors should be modified to use a positive connection to the parapet.