You are on page 1of 11

JCSS and IABSE Workshop on Robustness of Structures

Design-Oriented Progressive Collapse Assessment of Steel Framed Buildings


Anastasios G. Vlassis
Imperial College London anastasios.vlassis@ imperial.ac.uk

Bassam A. Izzuddin
Imperial College London b.izzuddin@ imperial.ac.uk

Ahmed Y. Elghazouli
Imperial College London a.elghazouli@ imperial.ac.uk

David A. Nethercot
Imperial College London d.nethercot@ imperial.ac.uk

Abstract
This paper introduces a relatively simple yet accurate methodology to evaluate the efficacy of steel framed buildings to resist progressive collapse initiated by sudden local column failure. Depending on the required level of sophistication, the proposed procedure can be implemented at various levels of structural idealisation, considering either the full structure without the damaged column or critical substructures. In either case, the procedure accounts for the dynamic effects of column failure and considers ductility demand and supply in assessing the potential for progressive collapse, where three main assessment stages are proposed. The first stage involves the determination of the nonlinear static response of the damaged structure/substructure under gravity loading using either simplified or detailed modelling techniques, including important factors such as the connection response as well as compressive and tensile membrane actions. The second stage employs a newly developed simplified dynamic approach, where the maximum dynamic response is estimated from the nonlinear static response using an energy equivalence criterion. The final stage assesses the vulnerability of the structure to progressive collapse by comparing the ductility demand in the connections, corresponding to the maximum dynamic response, to the ductility supply, considering the deformation capacity of individual connection components. The applicability of the proposed approach to progressive collapse assessment is demonstrated through a case study, which considers sudden removal of a peripheral column in a typical steel-framed composite building. This study shows that such structures can be prone to progressive collapse, especially due to the inability of secondary beam connections to transfer the gravity loads to the adjacent undamaged parts of the structure. Several remedial measures are investigated, including the benefits of additional reinforcement in the slab over the connections, the quantification of which is made possible by the proposed approach. Keywords: Progressive collapse, column removal, ductility, composite connections

1 Introduction
Progressive collapse is a relatively rare event as it requires both an extreme loading to cause the initial damage and a structure that lacks adequate continuity, ductility, and redundancy to resist the subsequent spread of damage. However, the recent escalation of terrorist attacks on buildings has highlighted the necessity to consider progressive collapse mitigation as a basic design criterion because of the significant

JCSS and IABSE Workshop on Robustness of Structures casualties that can result when collapse occurs and the economic loss associated with service disruption for large-scale building facilities. The main feature of progressive collapse is that the total cumulative damage is disproportionate to the original cause [1]. As it is not feasible to foresee all possible sources of collapse initiation, a rational progressive collapse design should aim at localizing rather than preventing damage on the whole. This goal can be generally achieved through structural robustness, since the inherent redundancy and ductility of a robust structure preclude premature failure modes that can lead to progressive collapse. Although building authorities in most western countries have introduced guidance to mitigate the risk of progressive collapse and enhance structural integrity [1-3], there is a broad consensus that the issue is addressed only in a very limited way [4]. Such guidance vary from the prescriptive, for example tying force requirements [3], to performance based [1,2], though in the latter case there is still a clear need for simplified design-oriented methods that consider ductility issues in the context of the nonlinear dynamic structural response under an extreme event. This paper presents such a method for progressive collapse assessment of steel framed buildings under an extreme dynamic event leading to instantaneous column removal. The new design-oriented method can be implemented at various levels of structural idealisation, depending on the required level of sophistication, in the sense that it can be applied either to the full structure, excluding the damaged column, or more practically to critical substructures. Regardless of the level of structural idealisation, three main assessment stages are considered: i) determination of the nonlinear static response, ii) estimation of the dynamic response due to instantaneous column removal, and iii) comparison of ductility supply and demand to establish the potential for progressive collapse. The paper proceeds by describing the components of the proposed assessment method, and finally presents a case study considering the assessment of a typical steel-framed composite building subject to sudden removal of a peripheral column. This study highlights i) the inadequacy of tying force requirements alone towards preventing progressive collapse, ii) the necessity of considering the nonlinear dynamic response and ductility demand/supply in progressive collapse assessment, iii) the susceptibility of typical composite building structures to progressive collapse, and iv) the benefits of additional connection reinforcement in the slab towards improving structural robustness.

2 Nonlinear Static Response


The first component of the proposed assessment method is the static response of the steel framed system excluding the damaged vertical support member. Since the actual dynamic response due to sudden column removal is likely to involve large deflections and induce inelastic material behaviour, both geometric and material nonlinearity should be considered. The most accurate representation of the nonlinear structural response is obtained from detailed modelling of the overall structure (Fig. 1a), including the interactions of all the structural components. However, in view of the associated modelling complexity and computational demand, critical substructures, where ductility demands are concentrated, can offer accurate and efficient assessment of structural robustness, provided realistic boundary conditions are assumed. Such substructures may include the range of floors above the damaged column (Fig. 1b), an individual floor plate modelled in detail (Fig. 1c) or as a grillage (Fig. 1d), or individual floor beams (Fig. 1d). An important feature of the proposed methodology is that the nonlinear static response under gravity loading

JCSS and IABSE Workshop on Robustness of Structures may be obtained at various levels of idealisation either from detailed nonlinear finite element models [5,6], or from simplified models. In the latter case, the nonlinear response of a floor beam (Fig. 1d) including catenary action may be determined from simplified mechanical models [7], the response of a floor grillage (Fig. 1c) may be assembled from the individual responses of floor beams assuming a specific deformation mode, and the response of a sequence of floors (Fig. 1b) may be similarly assembled from the individual floor plates (Figs. 1c-d). Clearly, if the involved floor plates are identical in terms of structure and loading, it is sufficient to consider the nonlinear static response of a single floor, since the load sharing along the line of the damaged column would be negligible.

(a)

(c)

(d)

(b)

(e)

Fig. 1 Levels of idealization for progressive collapse assessment.


3

JCSS and IABSE Workshop on Robustness of Structures An important consideration in establishing the nonlinear static response of a selected substructure is the realistic modelling of connections. In this respect, it is essential to utilise connection models that capture the interaction between axial and bending actions, especially when the associated beams are subject to significant axial restraint from the surrounding structure. As shown in the case study, not only does this aspect of the connection response impinge on the tensile catenary/ membrane action of the substructure, but it can lead to considerable compressive arching action that can be very beneficial to the overall response. Finally, the nonlinear static response of the selected substructure is obtained under an assumed distribution of gravity loading at the instant of column removal, including the unfactored dead load plus a percentage of the unfactored imposed load that can vary from 25 to 50%. It is noted, however, that load scaling factors exceeding unity should be considered, since the dynamic response under such instantly applied loading will exceed the static response, as discussed in the following section.

3 Simplified Dynamic Assessment


The structural response under gravity loading due to instantaneous column removal is dynamic, closely resembling the dynamic response under instantaneously applied gravity loading, as illustrated in Fig. 2 for a simple structural system. For linear elastic structures, it is well-established that a dynamic amplification factor of 2 is associated with this type of step loading, and this appears to be the basis of the GSA and DoD guidelines [1,2]. When the structural response is nonlinear, however, such an approach based on dynamic load amplification is not applicable and can lead to considerable inaccuracy and unrealistic demands on the structural resistance.
Instantaneously applied gravity load

Gravity load

max

Instantaneously removed column (a) (b)

Fig. 2 Rationale of simplified dynamic approach.

A new simplified dynamic assessment approach [8], based on energy equivalence, is proposed to address the above issue. According to this approach, the nonlinear static response can be directly used to assess the maximum dynamic response under an instantaneously applied gravity load, where the area under the static curve should be identical to the work performed by the step loading at the maximum dynamic displacement. This is illustrated in Figs. 3a-b, where the maximum dynamic displacements corresponding to two different levels of gravity loading are obtained for an idealised structural system, exhibiting a plastic resistance Rp and a post-yield stiffness which may be due to strain-hardening and/or catenary action. It is worth noting that Fig. 3b corresponds to gravity loading exceeding the plastic resistance, which is quite a realistic prospect for a structural system that has lost an internal supporting column.
4

JCSS and IABSE Workshop on Robustness of Structures

R P2 Rp

P P2 P1

Rp P1

max,1

max,2

max,1

max,2

(a)

(b)

(c)

Fig. 3 Determination of maximum dynamic response using the simplified dynamic approach. (a) P < Rp; (b) P > Rp; (c) Pseudo-static response.

The maximum dynamic displacements associated with different levels of instantaneously applied loading can be used to construct a pseudo-static loaddeflection response curve, as illustrated in Fig. 3c. This has the advantage that the pseudo-static response of individual members/substructures can be used to assemble the pseudo-static response of higher-level substructures (Fig. 1) using static analysis principles, provided a dominant SDOF deformation mode may be assumed.

4 Ductility Assessment
The maximum dynamic response, as obtained from the two previously discussed assessment stages, can be translated into deformation demands in various parts of the structure. The most critical parts where deformation demands are often concentrated are the beam-to-beam and beam-to-column connections, and hence this is the focus of the proposed assessment methodology. It is important to note that both the rotational and axial deformation demands are required for connection assessment, the latter becoming potentially very important when the associated beam is subjected to sufficient axial restraint leading to significant catenary action. The two types of deformation can be combined to provide a measure of ductility demand which can be compared to the connection ductility supply to provide a connection failure criterion, that can be directly used to assess the potential for progressive collapse. In the context of detailed models of connections, active connection components should have already been identified for the development of the connection mechanical model. The overall connection ductility capacity is directly related to the ductility supply of these components, including the end plates, angle cleats and bolts, which can be determined from experimental, numerical or analytical studies. When a simplified connection model is used, the connection ductility is typically expressed in terms of a single rotation value corresponding to the connection type under consideration. Recommended rotation capacities can be found in several guidelines, even though most of these refer to connections used for earthquake resistant design [9,10]. A clear shortcoming of this simplified assessment is that it ignores the influence of axial connection deformation, which as mentioned before can become significant in the presence of tensile catenary action.

max

JCSS and IABSE Workshop on Robustness of Structures With respect to composite connections, in addition to ductility limits in the steel components, additional ductility limits may also be imposed by the slab. Considering the interaction between the steel connection components and the slab, possible failure modes are those associated with tensile reinforcement rupture in hogging moment regions, concrete crushing in sagging moment regions, as well as slip of the shear connection [11]. It is worth noting that when the response of the substructure under consideration is obtained by assembling the pseudo-static responses of individual members, this is typically achieved by assuming a deformation mode, which should be subject to a critical deformation profile that does not violate the deformation capacities of any of the members involved. The determination of the critical deformation profile is demonstrated in the following case study, where the assessment of progressive collapse due to sudden removal of a peripheral column is investigated for a typical steel-framed composite building.

5 Case Study
The proposed progressive collapse assessment procedure is applied to a typical 7-storey steel-framed composite building, where consideration is given to the instantaneous removal of a peripheral ground floor column. The building is designed as a simple construction, according to current UK design practice, and thus the required lateral restraint is provided by a pair of similarly sized braced cores. A conventional composite steel and concrete construction is selected for the superstructure floors, which are designed to carry equal gravity loads. Connection design and detailing is carried out in accordance with UK design guidelines [12]. Flexible end-plate connections are mainly employed for the main beams, while fin plate is the predominant connection type for the secondary beams. With respect to structural integrity, the tying force requirements specified in 2.4.5 of BS 5950 [13] have been satisfied.
6000 Removed column 6000

2375

Main beams

3000

3000

Transverse beam

Secondary beams

1500

Fig. 5 Plan view of the floor area affected by the peripheral column removal

(dimensions in mm). The geometry of the floor area directly affected by the peripheral column removal is shown in Fig. 5. To evaluate the susceptibility of the structure to progressive collapse as a consequence of the peripheral column loss, the lowest level of
6

JCSS and IABSE Workshop on Robustness of Structures structural idealisation, shown in Fig. 1e, is initially considered. At this level, the pseudo-static responses of the individual members, including the longitudinal main and secondary beams as well as the transverse primary beam, are established using the proposed methodology. Subsequently, a grillage-type approximation, at the next level of idealisation (Fig. 1d), is used to determine the overall dynamic resistance of the floor plate accounting for coupling effects, where the pseudo-static responses of individual members are assembled on the basis of an assumed deformation mode. In the present context, detailed slab modelling (Fig. 1c) and higher-level idealisations are not considered, which for the latter is realistic given that the various affected floor plates are identical in terms of structure and loading.

5.1 Pseudo-static Response of Individual Members


The pseudo-static response curves for the main, secondary and transverse beams are shown in Figures 6a, 6b and 6c, respectively. The curves correspond to various modelling assumptions regarding composite action, amount of slab reinforcement over the connections, and axial restraint. To establish the static response required by the simplified dynamic approach, the nonlinear structural analysis program ADAPTIC [14] is used. It is noted that for the beams in the main longitudinal direction of the building the percentage of the anticipated service loads is also given. The nominal value of the service loads has been calculated from the tributary area of the corresponding longitudinal strip. In line with the GSA guidelines [1], the level of service loads at the time of column removal is taken as DL+0.25IL, where DL and IL are the dead and the imposed loads carried by each strip. The principal conclusion that can be drawn based on the response of individual members is that the main beams, which are connected to the supporting columns through flexible end-plate connections, are generally more adequate than the secondary beams, which employ fin plate connections. When the main and the secondary beams are axially restrained, the response is enhanced due to compressive arching action. However, for the secondary beams, compressive arching action is much less pronounced due to connection yielding at a relatively early stage, thus compromising the dynamic load carrying capacity. The beneficial effects of additional tensile slab reinforcement are also illustrated in Fig. 6. Both the dynamic load carrying and deformation capacities of the beams increase with increasing amount of slab reinforcement, such an increase being more evident for the main and transverse beams. Importantly, the pseudo-static response of the main beams is very sensitive to the axial restraint provided by the adjacent structural members, principally due to compressive arching action. It is notable that the absence of axial restraint precludes the development of compressive arching action and results in a considerably reduced dynamic load carrying capacity. On the other hand, the relatively small compressive arching action in secondary beams, due to fin plate connections, results in a pseudostatic response that is rather unaffected by axial restraint. Finally, when composite action between the steel beams and the concrete slab is ignored, the pseudo-static response deteriorates considerably. Due to the reduced connection resistance and initial stiffness, the dynamic load carrying capacity diminishes dramatically, thus rendering the beams vulnerable to progressive collapse, despite the fact that the connection design satisfies structural integrity requirements of BS 5950 [13]. This demonstrates that tying force requirements alone cannot guarantee structural robustness, without consideration of ductility demand/ supply in the connections as determined by the nonlinear dynamic response.

JCSS and IABSE Workshop on Robustness of Structures

200% 25 Dynamic Load (kN/m)


150% 125% 100% 75% 50%

20

15

10

5 25% 0 0 100 200 300 400 (mm)

500

600

0% 700

Displacement,

(a)
16 14 Dynamic Load (kN/m) 12 10 8 6 4 2 0 0 100 200 300 400 500

100% Percentage of Service Loads (%) = 0.87%, w/ axial restraint = 2%, w/ axial restraint Bare-steel frame, w/ axial restraint = 0.87%, w/o axial restraint = 2%, w/o axial restraint Bare-steel frame, w/o axial restraint

75%

50%

25%

600 (mm)

700

800

0% 900

Displacement,

(b)
500 Bare-steel frame = 0.87% = 2.00%

Dynamic Moment (kN m)

300

200

100

0 0.000

0.030

0.060

0.090 (rad)

Rotation,

(c) Fig. 6 Pseudo-static response: (a) main beam, (b) secondary beams, (c) transverse beam.

400

0.120

0.150

Percentage of Service Loads (%)

= 0.87%, w/ axial restraint = 2%, w/ axial restraint Bare-steel frame, w/ axial restraint = 0.87%, w/o axial restraint = 2%, w/o axial restraint Bare-steel frame, w/o axial restraint

175%

JCSS and IABSE Workshop on Robustness of Structures

5.2 Assembled Floor Plate Response


As illustrated in Fig. 7, a linear deformation profile is assumed for the transverse primary beam, governed by its support rotation, as well as for the longitudinal beams, governed by their individual midspan deflection capacities. Superimposing the various profiles, an overall critical deformation profile can be readily established. Table 1 provides the overall critical deformation profile for four different cases: Case 1 both composite action and axial restraint are considered. A 0.87% reinforcement ratio is employed based on the minimum requirements of EC4 [15] for the limitation of crack width. Case 2 same as case 1 with 2% reinforcement ratio in the slab. Case 3 same as case 2 without consideration of axial restraint. Case 4 bare-steel frame design is assumed in for all beams, and axial restrained is considered.
PSB1

MRj

PSB2
j SB1

PSB3
SB2

PMB

RSB2 RSB3 RMB Fig. 7 determination of critical deformation profile.

It is worth noting that the deflection capacity of the main beam in the longitudinal direction governs the critical deformation profile in all cases. Based on the established deformation profile and the pseudo-static response curves for each of the corresponding members, the dynamic resistance Ri, including the moment resistance MRj at the support of the transverse beam, that each beam would develop in achieving this deformation can be easily determined. Moreover, the anticipated gravity loads can be translated into point loads Pi acting at the midspan of the longitudinal members (Fig. 7). The overall dynamic moment resistance at the critical deformation level as well as the moment demand at the support of the transverse primary beam due to these loads are summarized in Table 2. It can be seen that the substructure under consideration lacks sufficient dynamic resistance to prevent the initiation of progressive collapse in most of the cases. The only exception is the case involving a 2% reinforcement ratio and axial restraint of the main and secondary beams. Nonetheless, based on the calculated D/C ratios, it is clear that the estimated dynamic resistance is not much lower than the calculated demand, apart from the system ignoring composite action. This observation indicates the inherent robustness characteristics of composite steel framed structures. Of course, the specific conclusions of this study depend greatly on the accuracy of the adopted connection response models, including the associated ductility assessment, an area which requires extensive experimental validation. Furthermore, even if the adopted connection models were sufficiently accurate, it is still possible for cases 1 and 3 to
9

RSB1

SB3 MB

JCSS and IABSE Workshop on Robustness of Structures provide the required dynamic resistance without connection failure if more realistic slab models, accounting for two dimensional membrane action (Fig. 1c), are employed with the proposed progressive collapse assessment method.
Table 1 Critical deformation profile for the idealized floor system

Case No. 1 2 3 4

Deformation profile

(rad)

SB1

(mm)

SB2

(mm)

SB3

(mm)

MB

(mm)

0.0238 0.0274 0.0216 0.0163

35.8 41.2 32.3 24.5

107.3 123.5 97.0 73.4

178.8 205.9 161.7 122.4

235.4 271.1 212.9 161.1

Table 2 Overall moment demand and resistance Case No. 1 2 3 4 Demand (kN m) 2190.19 2190.19 2190.19 2190.19 Resistance (kN m) 2041.82 2440.61 1915.34 221.47 D/C ratio 1.07 0.90 1.14 9.89

6 Conclusions
This paper presents a simplified design-oriented method for assessing the resistance of steel-framed buildings to progressive collapse under an extreme event that leads to sudden loss of a vertical support member. The proposed procedure is applicable at various levels of structural idealisation, considering either the overall structure or critical substructures where ductility demand is concentrated. Three distinct stages are followed in the proposed method: i) the determination of the nonlinear static response, ii) the estimation of the dynamic response using a new simplified method, and iii) the assessment of connection ductility supply and demand. The method benefits from the ability to accommodate detailed as well as simplified nonlinear analysis models, which can be applied as required at different levels of structural idealisation. The application of the proposed procedure in a case study has demonstrated that composite steel framed buildings with typical structural configurations can be prone to progressive collapse due to sudden loss of a vertical support member. This observation is principally attributed to the inability of secondary beam connections to safely transfer the gravity loads to the surrounding structure if a fin plate connection type is employed. The supply of additional reinforcement in the slab over the hogging moment regions can generally have a beneficial effect on both the dynamic load carrying and deformation capacities. The response can be further improved if axial restraint provided by the adjacent structure can be relied upon. Finally, the new method offers for the first time a rational design-oriented procedure that deals with dynamic effects and ductility demand/supply in progressive

10

JCSS and IABSE Workshop on Robustness of Structures collapse assessment. However, the application of the proposed method should follow extensive experimental validation and calibration, particularly in relation to the connection response and ductility limits under combined bending and axial actions.

Acknowledgement
The authors would like to acknowledge the support provided for this work by ARUP and EPSRC under a Case award scheme.

References
1. General Services Administration (2003). Progressive Collapse Analysis and Design Guidelines for New Federal Office Buildings and Major Modernization Projects. 2. Department of Defense (2005). Unified Facilities Criteria, Design of Buildings to Resist Progressive Collapse. 3. Office of the Deputy Prime Minister (2000). The Building Regulations 2000, Part A, Schedule 1: A3, Disproportionate Collapse, 1992 edition, fourth impression (with amendments) 1994, further amended 2000. 4. National Institute of Standards and Technology (2005). Final Report of the National Construction Safety Team on the Collapses of the World Trade Center Towers, NIST NCSTAR 1, Draft for Public Comment. 5. Song, L., Izzuddin, B. A., Elnashai, A. S., and Dowling, P. J. (2000). An integrated adaptive environment for fire and explosion analysis of steel frames Part I: analytical models, Journal of Constructional Steel Research; 53: 63-85. 6. Izzuddin B.A., Tao X.Y., and Elghazouli, A.Y. (2004). Realistic Modelling of Composite and Reinforced Concrete Floor Slabs Under Extreme Loading I: Analytical Method, Journal of Structural Engineering, ASCE, 130(12):1972-1984. 7. Izzuddin, B. A. (2005). A Simplified Model for Axially Restrained Beams Subject to Extreme Loading, International Journal of Steel Structures; under review. 8. Izzuddin, B. A. (2004). Ductility Assessment for an Idealized Elasto-Plastic Structural System Subject to an Instantaneous Applied Load, internal discussion document, Imperial College, London. 9. European Committee for Standardization (2005). EN 1993-1-8:2003, Eurocode 3: Design of Steel Structures - Part 1-8: Design of Joints. 10. American Institute of Steel Construction (2002). Seismic Provisions for Structural Steel Buildings, Appendix S. 11. Anderson, D., Aribert, J. M., Bode, H., Kronenburger, H. J. (2000). Design rotation capacity of composite joints, The Structural Engineer 78(6): 25-29. 12. British Construction Steelwork Association/Steel Construction Institute (2002). Joints in Steel Construction: Simple Connections. 13. British Standards Institution (2001). BS 5950: Structural use of steelwork in building, Part 1: Code of practice for design Rolled and welded sections. 14. Izzuddin, B. A. (1991). Nonlinear Dynamic Analysis of Framed Structures, PhD Thesis, Department of Civil Engineering, Imperial College, University of London. 15. European Committee for Standardization (2004). EN 1994-1-1:2004, Eurocode 4: Design of Composite Steel and Concrete Structures Part 1-1: General Rules and Rules for Buildings.

11

You might also like