Limit analysis fast methods for assessment of progressive collapse

potential in RC structures
*

A. Ioani, I. Petran, M. Botez and L. Bredean

Technical University of Cluj-Napoca, Romania

Abstract
This study presents a comparative analysis of results obtained by using the LSP adopted in the
GSA (2003) Guidelines with respect to similar data furnished by two limit analysis based
methods. The first method takes into consideration only the contribution of beams, and the
second one considers the behavior under gravity forces of the entire two-way floor system. If
the floors, spanning above to a column lost in the first floor are of identical strength, then they
will all stand or they will all fall. For this reason, only an isolated bay should be considered in
the analysis and not the whole structure. These new, fast and easy methods require plastic
analysis of beams and slabs, and involve only hand calculations. Two models representing a
three-story framed structure are seismically designed, and their progressive collapse potential
is evaluated at lost corner column case. To assess the reliability of results, the classical DCR
values (GSA) are compared to plastic analysis DCR values, named by authors DCRp and
obtained following this approach. The new DCRp values are in good agreement with DCR
values obtained from the Linear Static Procedure-GSA (2003) (DCRp = 1.96 vs. DCR = 2.37),
and lead to similar conclusions regarding the potential for progressive collapse. If the second
method is applied and the contribution of slabs is considered, the magnitudes of inelastic
demands and DCRp values are obtained via the yield-line analysis by using the virtual work
method. Preliminary results shown reductions of DCRp values in the range of 55%, followed
by a change of the verdict from HIGH to LOW potential for progressive collapse; the verdict is
confirmed by advanced NS and ND analyses.
Keywords: progressive collapse, DCR, yield-line theory, limit analysis, RC structures.
_______________________________
*
Corresponding author:
Technical University of Cluj-Napoca, Faculty of Civil Engineering.
15 C. Daicoviciu Street, 400020, Cluj-Napoca, Romania.
E-mail: mircea.botez@yahoo.com.

Phone: +40 741 291 291

1.0 Introduction
The progressive collapse is a consequence of a local failure of a structural component that leads to
the entire failure of the structure or to a significant part of it. Known also as a disproportionate
collapse, the assessment of the safety of an existing building is based, in many design codes or
international guidelines [1, 2] on the structures capacity to bridge over a removed load bearing
element (column, wall. Practically a structure should be analyzed considering that a column is
suddenly removed from the structure, and two major changes may lead to important increases of
the internal forces: double span effect in horizontal elements bridging over the removed column,
and dynamic effect applied to existing gravity loads [3]. Four basic procedures are recommended
for the progressive collapse analysis: linear static (LS), linear dynamic (LD), nonlinear static (NS)
and nonlinear dynamic procedure (ND). Many others theoretical analyses and experimental
programs have underlined the following aspects:
- The LS procedure has advantage in that it is theoretically simple, can be applied for
buildings of 10 stories or less in height with relatively simple layouts, but a lot of manual
work is required to evaluate the DCR in each analysis step and to remodel the structure in
successive iterations [4]; in general, the LS procedure generates more conservative
decisions for progressive collapse [4, 5,6,7];
- Discrepancy between the conservative results obtained via LS analysis and those given by
ND procedure, are discussed in recent papers [8,9]; they identified the principal reason of
this discrepancy: in the LS analysis the contribution of the slabs is neglected, while in the
ND procedure the strength and deformation properties of the slab, considered as a primary
component, are taken into consideration;
- The nonlinear static method (NS) assumes both material and geometric nonlinear behavior,
but still does not consider the real dynamic effect; the dynamic increase factor DIF=2.0
recommended by NS procedure [1], leads to conservative results [6,7, 8, 9, 10];
- The ND procedure involves the nonlinear modeling of members and connections and it is
considered as a more precise tool for evaluation of progressive collapse potential but
requires a more complex analysis and considerable computational time [4, 6, 8].
To the above mentioned aspects, the following observations have been taken into consideration
when the authors have developed two fast and simple methods to evaluate the potential for
progressive collapse of RC framed structures:
- Vertical load-bearing components (columns), in general, are capable to accommodate the
supplementary internal forces generated by the removal of a column; consequently only the
behavior of primary horizontal components is the main concern in this investigation;
- Modern design and especially the seismic design, provides beams and slabs with sufficient
ductility and shear capacity; consequently only the flexural capacity is considered [11];
- All important increases in bending moments and in deformations are concentrated only in
the structural bays directly associated with the removed column; consequently, instead of
studying the entire structure (Figure 1a) only the behavior of one simple bay (Figure 1b)
might be investigated;
- The floors (beams and slabs) spanning to a column lost in the ground story are connected
through the upper floors columns and they are forced to undergo identical vertical

displacements. If all floors are of identical strength and gravity loads, then they will all
stand or they will all fall; for this reason, instead of analyzing a substructure with a number
of floors (Figure 1b), the proposed methods may assess the potential for progressive
collapse by investigating the behavior of an isolated typical floor (Figure 1c).

Figure 1. Simplified structural models in the limit analysis

Starting from these observations, the authors developed two simple limit analysis based methods,
to assess the potential for progressive collapse of RC structures. The first method considers the
beams as a primary component in the flexural resisting mechanism, while the second one
introduces the contribution of slabs. Principles of the methods and comparative results with
respect to the current procedures (LS, NS and ND) are discussed. The proposed methods need only
simple hand calculations, give rapidly verdicts, and these verdicts are in good agreement with
those indicated by the classical procedures.

2.0 Theoretical Basis of the Methods

A RC frame structure is subjected to the missing column scenario - the corner column case (Figure
2). The entire structure is loaded by amplified gravity forces: 2(DL+0.25LL) [1]. After the
removal of a ground corner column (Figure 1a), demands in structural components increase, large
inelastic deformations may occur in the affected bay, leading at limit to the formation of plastic
hinges at the ends of the beams and of yield-lines in slabs. The limit stage appears when plastic
hinges and yield-lines (from the first to the top floor), through their limit resisting moments (Mp,
respectively m), cannot sustain anymore the vertical loads. The fundamental upper-bound theorem
of plasticity is applied to express the condition to avoid progressive collapse. For a chosen
distribution of plastic hinges in beams and for an adopted pattern of yield-lines in slab, the
principle of virtual work is applied: external work (Lext) done by loads going through unit
displacement will be equal to the internal work (Lint) performed by the hinge/yield-line moments
going through hinge rotations from unit displacement (=1) [11, 12].
2.1. First Plastic Method
The internal work can be evaluated considering only the contribution of beams (Lbint), through the
internal work done by the limit/plastic moments acting in the j-plastic hinge (Mp,j, j=1n), where
n-represents the total number of plastic hinges developed in beams, from the first (I) to the top

floor (F = I,II,N), and N represents the number of floors (Figure 2a).

The particularity of the damaged model is that equal displacements are imposed at each level,
that is 1= 2== N= and, consequently, when a unit displacement is imposed in the virtual
work procedure ( = 1), and the beams are of equal length (Lx=Ly=L = the clear span of the
beam), then 1= 2= j== 1=n==1/L. Consequently:
n

j =1

j =1

Lbint = M p , j j = M p , j =

1 n
M p, j
L j =1

(1)

Figure 2. Plastic analysis of floor beams through the principle of virtual work
If the loading is equal, the external work Lbext also has a particularity, being equal at any floor:

Lbext = N Lbext, floor .

(2)

The gravity forces acting on the slabs ps are transferred to adjacent beams, according to the
tributary area, as linear loads (Figure 2b). If the loads are amplified by DIF = 2.0 [1], the external
work for a typical level where two beams are involved in the mechanism (Figure 2b), is:

1
2

1
2

Lbext, floor = 2 DIF ( Db + Ds ) L 1 = 2 2( Db + Ds ) L = 2 Db L +

1
ps L2
2

(3)

where: Db - is the dead load of the beam and ps - is the gravity load acting on the slabs.
If the internal energy (Lbint) exceeds the external energy (Lbext), then the flexural capacity of the
beam mechanism is sufficient to sustain the amplified loads:

Lbint Lbext =>

1 n
1
M p , j N Lbext, floor = N ( Db L + p s L2 )

L j =1
4

(4)

M
j =1

p, j

N (2 Db L2 +

1
p s L3 ) .
2

(5)

The last condition can be expressed as a ratio between the plastic demand and the plastic capacity.
Similar to DCR from GSA (2003), the authors introduced the plastic Demand-Capacity-Ratio
(DCRp) with the acceptance criterion:

DCR p =

1
p s L3 )
2
1.0 .

N ( 2 Db L2 +
n

(6)

p, j

j =1

If the building has beams of identical strength at any floor, it is sufficient to check the DCR p value
for a typical floor (Figure 2b):

DCR p =

2 Db L2 +
4

M
j =1

1
p s L3
2
=

1
ps L3
2
1.0 .
M p ,1+ M p , 2+ M p ,3+ M p , 4
2 Db L2 +

(7)

p, j

Expression (7) shows how fast and simple can the first method be: for a model of equal spans
(Lx=Ly=L) and equal strengths of the beams, only the mechanism for a typical floor should be
checked; that means the evaluation of only six terms: Db, ps, Mp1, Mp2, Mp3 and Mp4.
2.2 Second Plastic Method
The second method takes into consideration not only the resisting mechanism of the beams
(through plastic hinges) but also the flexural capacity of the floor slabs (through yield-lines). All
the assumptions of the yield-lines theory [12] should be applied. When the progressive collapse
resistance of the building is investigated, only the slabs from the affected bays are considered.
Usually, the slabs at any floor are of identical dimensions and reinforcement, and consequently of
identical strength. The assumed yield-line pattern used in the analysis is illustrated in Figure 3.
In the internal and external work, the slabs supplementary contribution ( Lsint , Lsext ) is introduced:

Lint = Lbint + Lsint ; Lext = Lbext + Lsext .

(8)

The beams contribution loaded only by the linear load Db (Figure 3) is:

Lbext = 2( DIF Db

1
1
L 1) = 2 2 Db L = 2 Db L ;
2
2

Lbext =

1 4
M P, j .
L j =1

(9)

Figure 3. Yield-line pattern in slabs and plastic hinges in beams.

If the slab top reinforcement over the beam 2 and 4 are identical ( i2 = i4 = i ), the slabs
contribution (Figure 3) will be:

1
2

1
3

1
3

Lsext = ( DIF p s )(2 L L) = 2 p s L2 =

2
p s L2
3

Lsint = m2 L By + m4 L Ax + mL Ax + mL By ; Ax = By =
Lsint = m(2 + i2 + i4 ) = 2m(1 + i )

(10)

1
; m2 = i2 m ; m4 = i4 m
L

(11)
(12)

where m is the ultimate (limit) positive moment / unit length given by the bottom reinforcement
(in the field), and (- i m ) is the negative ultimate moment/unit length over the supports.
The condition to avoid the collapse mechanism for the floor system (beams and slabs) will be:

2
1 4
M P , j + 2m(1 + i ) 2 Db + ps L2

L j =1
3

Lbint + Lsint Lbext + Lsext =>

(13)

The new expression for the plastic Demand-Capacity-Ratio DCR p is:

DCR p =

2 Db L2 +
4

M
j =1

p, j

2
p s L3
3

+ 2m (1 + i )L

(14)

3.0 Testing the Methods

A three-story RC framed structure having two spans and five bays of 6.0 m each, was completely
designed and detailed for a zone of low seismicity (ag = 0.08g); beams (250x450mm) are of
identical dimensions and strength, as well as the slabs (h = 150mm) and columns (400x400mm).
The compressive strength of concrete (C25/30) is fck = 25 N/mm2 and the steel is of S500 type (fyk
= 500 N/mm2). The beams have 2D18mm bars as top reinforcement, respectively 2D14mm bars in
the bottom part. The slabs have identical reinforcement in both directions: D8mm/15cm in the
bottom part (field) and D10mm/15cm over the beams, as negative bending moment reinforcement.
In order to test the accuracy of the proposed simplified methods with respect to the classical GSA
(2003) procedures (LS, NS and ND), two 3-D models (without slabs and with slabs) are analyzed
using SAP2000.
3.1 First Plastic Method Results
The simplified model (Figure 4a), where only columns and beams are considered as primary
components, was loaded with the amplified gravity forces 2(DL+0.25LL). Following the step-bystep linear static procedure [1], demands (in terms of bending moments) in beams are evaluated.
Separately, the expected flexural capacity of the rectangular cross-sectional beams is computed,
and DCR values are established by hand calculations (Figure 4a).

Figure 4. Simplified model (no slabs): a) DCR values in beams; b) plastic analysis of floor beams
DCR values exceed the allowable value of 2.0 in all critical sections (Figure 4a) indicating that all
these sections are severely damaged and also, the possibility of a generalized (spatial) mechanism.
The progressive collapse potential in this case, according to GSA LS procedure, is HIGH. The
First Plastic Method requires only the analysis of two beams from one typical floor (Figure 4b).
The linear loads representing the self-weight of a beam (Db=2.81kN/m) and the gravity forces
brought by the slabs (ps = 6.3 kN/m2) are given. The limit moments (Mp,j) in plastic hinges are
evaluated for one beam, in two sections: Mp,1=Mp,4=116.4 kNm, and Mp,2=Mp,3=70.4 kNm.

Using the expression (7) where L=5.60m, the plastic DCR p value is easily computed:

1
p s L3
2
DCR p =
= 1.96 > 1.0
2(M p ,1+ M p , 2 )
2 Db L2 +

(15)

When DCRp is greater than 1.0, a plastic failure mechanism will occur and the verdict given by the
First Plastic Method is HIGH potential for progressive collapse, similar to the GSA-LS procedure
verdict (Figure 4a). In addition, the new introduced ratio (DCRp) is very close to the average value
of the classical DCR [1] (DCRp = 1.97 vs. DCRav =2.37). If we consider the 10% rule [11], the
DCRp values are even closer (DCRp = 2.16 vs. DCRav =2.37).
3.2 Second Plastic Method Results
A new and more complex 3-D model (Figure 5) where beams and slabs are considered as primary
components, is analyzed using one dimensional finite elements (for columns and beams) and shell
elements for slabs (SAP2000). The potential for progressive collapse is evaluated following three
classic procedures (LS, NS and ND), and the GSA (2003) [1] acceptance criteria are applied.

Figure 5. Complete model (with slabs): a) DCR values (LSA); b) plastic rotations (NSA) in mrad
The LS procedure (Figure 5a) shows that DCR values change significantly; the allowable value
DCR = 2.0 is exceeded only in one beam section and after a second iteration (with a plastic hinge
inserted in the critical section), all DCR will satisfy the GSA acceptance criteria, and the verdict
will be: LOW potential. To validate this verdict two advanced analyses (NS and ND) have been
performed and the results are detailed and discussed in another paper [9]. NS analysis with
2(DL+0.25LL) where DIF=2.0, shows that plastic hinges occur in beams, but the plastic rotations
are very small compared to the allowable rotation limit (max = 3.1mrad vs. allow = 35mrad).
Consequently, the potential for progressive collapse is LOW (Figure 5b). The most complex and
costly procedure, the nonlinear dynamic analysis under gravity load (DL+0.25LL) confirmed the

verdict (LOW potential): the plastic rotations in beams are even smaller (max = 0.4mrad vs. allow
= 35mrad), indicating that DIF = 1.5 would be an appropriate value for the linear analysis. The
Second Plastic Method formula (14) requires the evaluation of the limit moment in plastic hinges
(Mp,j) and the limit moments/unit length in slabs. Spandrel beams are considered as L-flange
beams, and two plastic moments shall be evaluated: Mp,1=Mp,4=171kNm, Mp,2=Mp,3=70.4kNm.
Using the slab reinforcement details, the positive limit moment in slabs is determined
(mx=my=m=19.8kNm/m); over the supports, the negative limit moment is (-mi) where i = 1.56.
The vertical loads on beams (Db = 2.81kNm) and slabs (ps = 6.35kN/m2) are as in the First Plastic
Method (Section 3.1). With the numerical data, the plastic DCRp (14) is computed:

2
p s L3
3
DCR p =
= 0.87 1.0
2(M p ,1+ M p , 2 ) + 2m(1 + i )L
2 Db L2 +

(16)

When DCRp value is smaller than 1.0, the external energy expended by the loads moving
downward is smaller than the internal energy dissipated in plastic hinges and along the yield lines.
Consequently, the potential for progressive collapse is LOW, identic with the verdict furnished by
LS, NS and ND procedures recommended by the codes [1]. The verdict is accurate, being
confirmed by advanced analyses, and it is reached rapidly through very simple hand calculations.

4.0 Conclusions
In the paper the progressive collapse of a three-story RC framed structure is assessed using two
new limit analysis based methods. Theoretical basis of the proposed methods are presented, as
well as the particularities in their use when the primary horizontal components (beams or slabs)
are of identical strength. The methods were tested on low-rise (3-story) and mid-rise (13-story) RC
buildings when two missing column scenarios are considered. The methods have the capacity to
easily introduce the slab contribution in the flexuralresisting mechanism formed by beams and
slabs. All the progressive collapse potential verdicts (LOW or HIGH) given by the methods were
confirmed by the classical procedures (LS, NS and ND) recommended in international design
codes (GSA and DoD); in the paper, only the corner column case for a three-story RC building is
discussed.
The First Plastic Method is recommended to be the first check against the risk for progressive
collapse, and it is considered by the authors as a successful replacement of the linear static
procedure because: it is faster than LSP, it is easy to be applied, does not require computer
programs, 3-D models or FE analyses; it requires only hand calculations, basic notions for the
plastic/limit analysis of beams using the well-known principle of virtual work, and always leads to
conservative decisions.
If the verdict reached by the First Plastic Method indicates o HIGH potential for progressive
collapse, the user should continue the analysis by considering the Second Plastic Method. This
method is more accurate because takes into account the supplementary contribution of slabs in the
resisting mechanism. In the limit stage, when plastic hinges are formed in beams, and yield-lines

in slabs, the same principle of virtual work shall be applied, and the plastic DCRp value shall be
computed for a typical floor.
If DCRp is smaller than 1.0, the structure has a LOW potential for progressive collapse, and this
verdict may be regarded as a final one; the accuracy of the results given by the Second Method
was confirmed by the performed NS and ND analyses. If DCRp is greater than 1.0, indicating a
HIGH potential for progressive collapse, the final decision will be released by a more sophisticate
ND analysis.
Both proposed methods need further validations for different models (different bay size, number of
stories, increased level of gravity loads) and for different damage cases; research is in progress.

5.0 Acknowledgements
The authors gratefully acknowledge the funding by the Executive Agency for Higher Education,
Research, Development and Innovation Funding (UEFISCDI), Romania, under grant PN II PCCA
55/2012 Structural conception and Collapse control performance based Design of multistory
structures under accidental actions CODEC (2012-2015), made in the frame of the Partnerships
Program Joint Applied Research Projects.

6.0 References
1. General Service Administration, Progressive Collapse Analysis and Design Guideline for
New Federal Office Buildings and Major Modernization Projects, Washington, USA, 2003.
2. Department of Defense, UFC- Design of Buildings to Resist Progressive Collapse,
Washington, USA, 2009.
3. Dat, P.X., Hai, T.K., Membrane Action of RC slabs in Mitigating Progressive Collapse of
Building Structures, Engineering Structures, 2011.
4. Kim, J., Kim, T., Assessment of Progressive Collapse-Resisting Capacity of Steel Moment
Frames, Journal of Construction Steel Research 65, 2009, pp. 169-179.
5. Marjanishvili, S.M., Progressive Analysis Procedure for Progressive Collapse, Journal of
Construction Facilities, 2004, pp. 79-85.
6. Tsai, M-H., Liu, B-H., Investigation of Progressive Collapse Resistance and Inelastic
Response for an Earthquake-Resistant RC Building Subjected to Column Failure,
Engineering Structures 30, 2008, pp. 3619-3628.
7. Kim, H-J., Progressive Collapse Behaviour of RC Structures with Deficient Details, PhD
Thesis, The University of Texas at Austin, 2006.
8. McKay, A., Gomez, M., Marchand, K., Non-Linear Dynamic Alternate Path Analysis for
Progressive Collapse: Detailed Procedures Using UFC 4-023-03, DDESB, 2010.
9. Botez,M., Bredean, L., Petran, I., Ioani, A., Accuracy and Efficiency in Progressive Collapse
Analysis: Real Structures vs. Successively Reduced Substructures, Protect 2013, 2013
under press.
10. Dinu, F., Dubina, D., Robustness Based Design of Steel Building Frames Under Extreme
Loads, STESSA 2012, Santiago de Chile, 2012.
11. Udilovich, K.V., Shleykov, I.B., Banchuzhnyy, M.V., Design of Flat-Plate Floors for
rogressive Collapse Using Yield Lines Analysis, Concrete International, Vol.32, No.7, 2010.
12. Jones, L.L., Ultimate Load Analysis of Reinforced and Pre-stressed Concrete Structures,
Chatto and Windus, London, 1968, 248pp.