Progressive Collapse of Concrete Buildings

PROGRESSIVE COLLAPSE RESISTANCE OF CONCRETE BUILDINGS

Ying Tian

Department of Civil and Environmental EngineeringUniversity of Nevada Las Vegas

OUTLINE

• Historical events of progressive collapse• Design standards and available approaches• Gaps in existing knowledge and research needs• Experimental study of progressive collapse resistance of

RC beams• Numerical simulation of axially restrained RC frame

beams• Numerical simulation of RC flat-plate buildings at the risk

of progressive collapse• Structural laboratory at UNLV

“Progressive collapse is defined as the spread of an initial local failure from element to element resulting, eventually, in the collapse of an entire structure or a disproportionately large part of it.”

--- ASCE 07-10

HISTORICAL EVENTS OF PROGRESSIVE COLLAPSE

Ronan Point apartment, 1968, UK

• Precast concrete wall and floor system.

• Explosion caused by a gas leak blew out one of the precast wall panels on the 18th floor, triggering the partial collapse of the building.

• Attention to progressive collapse was initiated.

(Nair, 2004)

Commonwealth Avenue apartment, 1971, Boston

• RC flat-plate structure• Likely construction over-load, poor material

properties in cold weather, and inadequate positioning slab top bars caused punching shear failure at roof level.

• Punching shear failure propagated to the ground level.

• Attention to progressive collapse was initiated.

(King and Delatte, 2004)

Alfred P. Murrah Building, 1995,Oklahoma City, Oklahoma

• RC frame structure with transfer girders designed in accordance with ACI 318-71.

• Discontinuity of reinforcement in both the positive and negative moment reinforcement.

• The blast from the bomb destroyed column G20 below the transfer girder and may have destroyed or severely damaged columns G24.

• 168 people died.

Sampoong Department Store, Seoul, South Korea

• RC flat-plate structure• Punching shear failure initiated from an interior slab-column connection at the

top story.• Contributing factors for the failure included reduced slab effective depth and a

35% increase in dead loads due to the change of use at the 5th floor (Gardner et al. 2002).

• Killed 501 people.

DESIGN STANDARDS

Both consider progressive collapse as dynamic and nonlinear event.

ASCE/SEI Committee, Disproportionate Collapse Standards and Guidance, is currently develop new standard modified from DOD ‐2009. 

Design Approaches

• Indirect Design - emphasizes providing minimum levels of strength, continuity, and ductility to ensure structural integrity.

• Direct Design - includes the Specific Load Resistance and the Alternate Path approaches.

Relies on an integrated system of tie forces for developing tensile membrane or catenaryaction. Horizontal ties and vertical ties.

Indirect design – DOD procedure

• Indirect Design - emphasizes providing minimum levels of strength, continuity, and ductility to ensure structural integrity.

• Building must bridge across a removed element.

Location of column removal considered in DOD 2009

Moment before column removal

Moment after column removal

mg

P

u

t

P

P

0

0.5

1

1.5

2

2.5

3

3.5

0 0.5 1 1.5 2 2.5

Time (s)

Dis

plac

emen

t / S

tatic

Dis

plac

emen

t P = 0.9Pu

P = 0.7Pu

P = mg

m

Dynamic Loading Effects Due To Sudden Removal of Supporting Column

(undamped SDOF system)

5% damping ratio

Three analysis procedures permitted: • Linear Static (consider M-factor)• Nonlinear Static (consider Nonlinear Dynamic Increase factor) • Nonlinear Dynamic

Force‐driven nonlinear static analysisLoad applied considers DIF for tributary area surrounding the lost element

Dynamic Increase Factor (DIF) for concrete structures

(Marchand et al. 2009) –Protection Engineering Consultants

GAP IN EXISTING KNOWLEDGE AND RESEARCH NEEDS

• Actual strength of critical element such as beams and beam-column joints

• Actual deformation capacity of critical element such as beams under large deformation

• Participation of slabs in resisting progressive collapse

• Risk of progressive collapse of flat-plate structures• Appropriate retrofit techniques for progressive

collapse prevention

EXPERIMENTAL RESEARCH

• In collaboration with Dr. Youpo Su at Hebei Polytechnic University (China)

• Investigated flexural capacity of RC frame beams where axial restrains exist

• Both static and dynamic loading tests were conducted.

Deflection

Ver

tical

Loa

d

Compressive arch action

δtu δcu

Pcu

Ptu

Pyu Capacity based on yield-line theory

Tensile arch (catenary) action

Typical Behavior of RC Frame Beams

Compressive arch action and catenary action

(Bao, 2008)

Prototype Structure and Test Specimen

Prototype structure and typical geometry of test specimen

Monotonic Loading Test Setup

12 specimens were tested: 9 under static loading (1/2‐scale), 3 under different loading speed (1/3‐scale)Test variables: (1) reinforcement ratio, (2) span‐to‐depth ratio, and (3) loading speed

Following concrete crushing

Prior to final failure

A3: 2.7 m x 0.3 m x 0.15 m, Pcu = 249 kN, PACI = 147 kNB1: 4.2 m x 0.3 m x 0.15 m, Pcu = 125 kN, PACI = 77 kNB2: 5.7 m x 0.3 m x 0.15 m, Pcu = 83 kN, PACI = 55 kNAll 3φ 14 at top and bottom, ρ = 1.13%

1

1.5

2

2.5

3

0 0.3 0.6 0.9 1.2 1.5

Flexural Reinforcement Ratio (%)

Stre

ngth

Enh

ance

men

t Fac

tor α

with symmetrical reinforcement with asymmetrical reinforcement

A4A1

A5

A3

A2A6

1

1.5

2

2.5

3

0 2 4 6 8 10

Span / Depth (l /h )

Stre

ngth

Enh

ance

men

t Fac

tor α

with symmetrical reinforcement with asymmetrical reinforcement

A3

B1B2

A6

B3

n

(a) (b)

Effect of Reinforcement Ratio Effect of Span‐to‐depth ratio

-200

-150

-100

-50

0

50

100

150

200

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6

Center Deflection / Beam Depth (δ/h )

Ver

tical

Loa

d P

(kN

)

-45

-30

-15

0

15

30

45

Ver

tical

Loa

d P

(kip

)

Hor

izon

tal R

eact

ion

N (k

N)

Hor

izon

tal R

eact

ion

N (k

ip)

Specimen C1 Specimen C2 Specimen C3 Peak Load Pcu

C1: 2.7 m x 0.2 m x 0.1 m, loading rate 0.2 mm/s, Pcu = 91.6 kNC2: 2.7 m x 0.2 m x 0.1 m, loading rate 2 mm/s, Pcu = 96.4 kNC3: 2.7 m x 0.2 m x 0.1 m, loading rate 20 mm/s, Pcu = 108 kNAll 2φ 12 at top and bottom, ρ = 1.3%

Observations from monotonic loading tests

• Compressive arch action resulting from axial restraint contributed at least 50% extra loading capacity beyond the capacity estimated without considering axial restraining forces and strain harderning.

• Load resistance under catenary action may not provide higher capacity than under compressive arch action.

• High loading speed slightly increases beam flexural stiffness and load resistance.

Dynamic Loading Tests

Test variables: Load level, reinforcement ratioFour specimens were tested: D1 to D4, 5700 mm x 300 mm x 150 mm (1/2‐scale)D1: no axial restraint was appliedD1 and D2: ρ = 1.2 %, D3: 1.8 %, D4: 2.4%Each specimen was tested multiple times with different weight of mass blocksLoad release time less than 10% of natural period

Lower weight of mass blocks: study the dynamic response of a specimen within its elastic range  

Higher weight of mass blocks: detect the dynamic load‐carrying capacity

0

5

10

15

Def

lect

ion

(mm

)

0

15

30

45

Hor

izon

tal F

orce

(kN

)

0

15

30

45

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8Time (s)

Res

train

ing

Mom

ent (

kN-m

)

Midspan deflection

Quarterspan deflection

(a)

(b)

(c)

Dynamic response under lower level of load

0

30

60

90

Cen

ter D

efle

ctio

n (m

m)

P = 18.8 kN

P = 23.9 kN

P = 44.0 kN

P = 38.9 kN

0

30

60

90

0 0.5 1 1.5

Time (s)

Cen

ter D

efle

ctio

n (m

m)

P = 54.6 kN

0 0.5 1 1.5

Time (s)

P = 44.0 kN

P = 28.9 kN

P = 53.5 kN

Concrete crushing

Flexural yielding D1 D2

D3 D4

Dynamic response under higher level of load

Dynamic response of axial restraining force and restraining moment

-200

-150

-100

-50

0

50

100

150

0 0.05 0.1 0.15 0.2 0.25 0.3-200

-150

-100

-50

0

50

100

150

0 0.05 0.1 0.15 0.2 0.25 0.3-200

-150

-100

-50

0

50

100

150

0 0.05 0.1 0.15 0.2 0.25 0.3

Res

train

ing

Mom

ent

(kN

-m)

Axi

al F

orce

(k

N)

(a) (b) (c)

Time (s) Time (s) Time (s)

At peak deflection At peak deflection At Concrete Crushing

Specimen D2 Specimen D3 Specimen D4

Diagonal Crack

Concrete Spalling

Edge Column Center Column

(a) Damage pattern of Specimen D3 (P = 54.6 kN) Damage pattern of Specimen D3(P = 54.6 kN, approximately the load capacity)

Damage pattern of Specimen D3(P = 53.5 kN, collapsed)

Damage Pattern

Observations from dynamic loading tests

• Typically assumed 5% damping ratio for cracked concrete structures was verified.

• Compressive arch action still exists under dynamic loading scenario considered by DOD and can significantly increase the dynamic loading capacity.

• Dynamic increase factor of 2 could be too conservative for force controlled actions.

• Another series of tests is being conducted to further identify dynamic loading effects (mainly evaluate DIF proposed by DOD and dynamic deformation capacity).

NUMERICAL SIMULATION OF AXIALLY RESTRAINED RC FRAME BEAMS (ONGOING)

• Current DOD progressive collapse design guideline considers dynamic loading condition. The response of structure from an analysis (deformation and force demand) can be highly sensitive to the definition of beam flexural capacity.

• To reduce uncertainty in an analysis, appropriate nonlinear model is need for frame beams surrounding the lost column.

• Using traditional ACI code approach to define M-ϕ (or M-θp) in a nonlinear analysis cannot effectively capture the dynamic response under both compressive arch action and catenary action.

• Numerical analysis needs to consider the geometry nonlinearity when solving system equations.

Using fiber section to define flexural property

• Cross section is divided into several layers (regions) to have fibers along the beam or column.

• Material property is defined at stress-strain level.

• Confinement effects due to transverse reinforcement can be explicitly considered.

• Can be used for irregular cross sections.

• Current fiber section can only define flexural and axial loading behavior.

• Involves higher computational cost.• Available in SAP newer editions.

Simulation of axially restrained beams tested

• OPENSEES was adopted• Concrete 1 was used to define material

property for concrete• Confined concrete model for peak stress

and ultimate compress strain proposed by Scott et al. (1982) was use for cover concrete and core concrete.

• Steel 2 was used to define material property for reinforcing bars.

• Model (Bond_SP01) proposed by Zhao and Sritharan (2007) was considered.

• Zero-length section was used to define bond-slip property.

• Ultimate goal: nonlinear static and dynamic analysis of multi-story RC frame building designed w/ seismic loading (assisted by Ken Zhang) and w/o seismic loading (assisted by Sang-in Choi).

Concrete property (Concrete 1 model)

-200

-150

-100

-50

0

50

100

150

200

0 50 100 150 200 250 300

Load

and

Axi

al F

orce

(kN

)

Vertical Displacement at Center Column (mm)

Load (measured) Average Axial Force (measured)

Load (calculated) Axial Force (calculated)

Pu (ACI)

Simulation results

Symmetrically reinforced beam (ρ = 1.5%)

NUMERICAL SIMULATION OF RC FLAT-PLATES(ONGOING)

• Flat-plate buildings, especially those designed prior to 1980s, could be vulnerable to a progressive collapse.

• ABAQUS using shell elements is used to conduct nonlinear analysis.

• Concrete damaged plasticity model was used to simulate the property of concrete under tri-axial state of stress.

• Rebar layer was used to simulate tension and compression mats of slab flexural reinforcement.

• Preliminary analyses have been conducted.• Research assisted by Jinrong Liu.

40

4”

Inclined Crack

Behavior of two slab-column connections under simulated gravity loading

0

1

2

3

4

5

0 0.5 1 1.5

Center Deflection (in.)

G1.0

G0.5

First Yielding

Two‐way shear strength (ACI 318‐08)

(Tested at University of Texas at Austin)

(ρ=0.50%)

(ρ=0.50%)

(ρ=0.99%)

Test results of slab-column connections by (Elstner and Hognestad, 1956)

For flat‐plates with low‐to‐moderate reinforcement ratios, punching shear failure is actually controlled by flexure rather than shear.

Calibration of modeling parameter

0

0.2

0.4

0.6

0.8

1

0 0.003 0.006 0.009 0.012

Twist Angle (rad)

Torq

ue (t

onf-m

)

0

10

20

30

40

50

0 0.5 1 1.5 2 2.5

Deflection (in)V

ertic

al S

hear

(kip

s)

0

10

20

30

40

50

60

0 0.2 0.4 0.6 0.8

Deflection (in)

Ver

tical

She

ar (k

ips)

Specimen A-13, ρ = 0.55% Specimen 6AH, ρ = 0.6% Specimen T-2

P1 P2>P1

Applied LoadApplied Load

Column

Slab

Lateral Load

Column

Slab

Lateral Load

Test Result FE Simulation Result

Simulation results for a one story flat-plate building

Peak Dynamic Rotation Demand (rad.)

STRUCTURAL ENGINEERING LABORATORY AT UNLV

Renovated from a gymnasium

Strong floor

Strong floor: 32 ft long, 28 ft wide, and 4 ft thick reinforced concrete slab with a matrix of embedded anchors

Anchor unit

CONCLUSIONS

• Lateral restraining effect existing in an actual moment frame may significantly increase beam flexural capacity.

• Even though such effect is generally neglected in a normal design, it can be considered for progressive collapse resistance under extreme loading conditions.

• Fiber section can best describe the strength and stiffness properties of RC frame beams.

• Flat-plate buildings, especially older flat-plates, could be at high risk of progressive collapse.

• Input for industry is needed to better improve current design practice for progressive collapse.

QUESTIONS?

Thank You