Effect of earthquake damage on the behaviour of composite steel frames in fire

Abstract

Fire loading following earthquake loading is possible in any building in a seismic-prone area. However, most design approaches do not consider fire following earthquake as a specific loading case. Moreover, seismic design philosophies allow a certain degree of damage in structural elements which make structures more vulnerable when subjected to post-earthquake fire. This study uses three-dimensional numerical models to investigate the effect of earthquake damage on the fire resistance of composite steel-frame office buildings. A total of two types of earthquake damage, fire insulation delamination and residual lateral frame deformation, are investigated. It is concluded that earthquake damage can significantly reduce the fire resistance of composite buildings, with delamination of fire protection having the greatest effect. The results of this study can be used by designers to improve the post-earthquake fire resistance of composite buildings.

Keywords

composite construction earthquake engineering fire engineering fire following earthquake fire insulation

Introduction

Fire occurring after earthquake has been demonstrated to be a major threat for buildings in seismic-prone regions. Damage caused by fire following earthquake (FFE) can be worse than that caused by the earthquake itself (Scawthorn et al., 2005). Recorded experiences indicate that FFE has caused numerous deaths and extensive losses. For example, it was estimated that 80% of total damage and fatalities in the 1906 San Francisco seismic event were caused by fires following the earthquake (Scawthorn, 2008). In the 1923 Tokyo earthquake, the losses due to fire following the earthquake accounted for over 70% of total building losses and resulted in 140,000 deaths (Scawthorn, 2008). More recently, in the 1995 Kobe earthquake, there were approximately 7000 buildings destroyed by FFE (Faggiano, 2007).

Although the damage caused by post-earthquake fire can be very significant, it is not considered as a specific loading case in seismic design. Moreover, the current philosophy of seismic design permits a certain degree of damage to the structural elements, connections and fire proofing during seismic action, which potentially makes structures more vulnerable when subjected to post-earthquake fire. With the substantial damages and losses recorded in past events, it seems necessary to understand and quantify the behaviour of structures under multi-hazard events such as FFE.

Several previous studies have been conducted on the behaviour of earthquake-damaged steel frames in fire. Della Corte et al. (2003) developed numerical models to investigate the response of steel moment resisting frames (MRFs) under post-earthquake fire scenarios. They showed that the drift ratio is an important parameter that affects the fire resistance. In addition, it was observed that the seismic design philosophy adopted can significantly affect the performance of steel structures under post-earthquake fire. Faggiano et al. (2008) applied pushover and coupled thermal–mechanical analysis to evaluate two-dimensional (2D) steel MRFs subject to FFE. The results showed that there was a small reduction in fire resistance when the state of a frame after earthquake reached near collapse level. However, the fire resistance of the frames was still the same when the frame did not exceed the operational performance level limit.

Zaharia and Pintea (2009) presented post-earthquake fire resistance of two different seismically designed frames (moderate and severe seismic regions). The results showed that the frame designed for stronger seismic effects had a higher fire resistance rating in the case of post-earthquake fire. Memari et al. (2014) presented post-earthquake fire performance of MRFs with reduced beam section (RBS) connections. Post-earthquake fire exposure was only at the location of the RBS connections based on the assumption that fire insulation was damaged at the RBS connection after earthquake. The results showed that a post-earthquake fire produces smaller inter-storey drift ratios (IDRs) than the earthquake itself. No structural collapse occurred as a result of applied post-earthquake fires. Behnam and Ronagh (2015) performed pushover and thermal analyses to investigate the behaviour of 2D steel frames. Fire analysis only was also performed on an undamaged frame, as a benchmark. The results revealed that the damaged frames have lower fire resistance than the corresponding undamaged frame.

All of the aforementioned studies have focused on non-composite steel frames. Most of these structures have been analysed assuming 2D plane-frame behaviour, without considering the effects of out-of-plane behaviour. By contrast, this study focuses on composite steel-frame, concrete floor construction as is commonly used in mid-rise commercial frame structures. Although some essential aspects of the behaviour of these structures in fire can be captured in two dimensions, such an approach misses load redistribution in the out-of-plane direction which is a key load-carrying mechanism available at large deflections, as identified by the Cardington fire tests (Gillie et al., 2001; Kirby, 1998). Thus, when a composite slab is present, three-dimensional (3D) behaviour must be considered in an analysis to obtain realistic behaviour on a steel frame in fire. Additionally, previous research on post-earthquake fire behaviour has focused on unprotected steel frames. There is as yet a lack of detailed research into the influence of fire insulation on steel frames during FFE.

To address these knowledge gaps, this study extends Suwondo et al.’s (2017) previous work on post-earthquake fire behaviour of composite steel-framed structures. The 3D numerical models are employed to simulate post-earthquake fires on earthquake-damaged structures. A total of two types of damage, fire insulation delamination and residual lateral frame deformation, are investigated to represent earthquake damage. The consideration of the two types of damage will be explained in the section titled ‘Earthquake damage’. The main objectives of this study are to understand the influence of the following:

Delamination of the fire insulation on primary beams due to earthquake damage;

Delamination of the fire insulation on columns due to earthquake damage;

Residual deformation on the post-earthquake fire behaviour of composite steel frames.

The remainder of this article is structured as follows: a description of the generic building studied, the types of damage considered, the modelling approach adopted, the results of the study, and conclusions.

Generic building

A generic five-storey composite steel frame is analysed in this study. Figure 1 shows the structure’s plan and elevation. The building is designed for high seismicity using a MRF with medium ductility according to Eurocode EN 1993-1-1 (CEN, 2005a), EN 1994-1-1 (CEN, 2004a) and EN 1998-1-1 (CEN, 2004b). In this study, steel behaviour was assumed to be elastic-perfectly plastic for both columns and beams with a yield stress of 355 MPa and Young’s modulus of 210 GPa at ambient temperature. The compressive strength of concrete in the slabs was taken as 30 MPa, and the yield strength of the rebar is 450 MPa. Thermal expansions of steel and concrete are taken as 1.35 × 10⁵°C⁻¹ and 9 × 10⁵°C⁻¹, respectively. The concrete and steel properties at elevated temperatures follow the recommendations in Eurocode EN 1992-1-2 (CEN, 2004c) and EN 1993-1-2 (CEN, 2005b). The concrete slabs have a thickness of 130 mm and the rebar mesh consists of 6-mm diameter bars at 200-mm centres each way as shown in Figure 1(c).

Figure 1.

Generic frame: (a) plan view of building, (b) elevation view of structural frame and (c) composite floor slabs.

Primary beams and columns are assumed to be protected with a 10-mm thickness of lightweight insulating material which has thermal conductivity of 0.2 W/mK, specific heat of 1100 J/kgK and density of 300 kg/m³. In order to utilise tensile membrane action in the slab panels, which are surrounded by the primary beams, the secondary beams are left unprotected, as is common in a performance-based fire design. The total design load at the fire limit state (1 × dead + 0.5 × live) is taken as 5.5 kN/m². This load level results in a load ratio for the internal columns at bottom level and primary beams of 0.3 and 0.5, respectively. The load ratio is defined as the ratio of the applied load to the bearing capacity of a member. Although permitted by the Eurocode EN 1991-1-1 (CEN, 2002a), no further load reduction is considered in the structure in order to be conservative. For simplicity, the columns are assumed to have the same section size at all heights. Only the ground floor is exposed to fire. The ground floor is assumed to have the worst fire scenario since the columns have the largest load ratio compared to the upper floor. Moreover, the plastic hinges will only form during earthquake in the columns at ground floor so they are most likely to be damaged at ground level.

Earthquake damage

A key aspect of post-earthquake fire analysis is determining a structure’s physical state after an earthquake as this represents the initial condition for the subsequent fire event. It is well known that providing detailed information regarding earthquake damage is difficult due to randomness and uncertainties of both structural properties and earthquake vibration. Accordingly, in this study, two generic types of earthquake damage are considered: fire insulation delamination and residual lateral frame deflections as a result of plastic deformations developed during the earthquake.

Fire insulation delamination

The philosophy of seismic design permits a certain degree of damage to structural elements. Past events have demonstrated that earthquakes can cause damage to active fire protection systems such as sprinklers and reduce the effectiveness of firefighting capability. Thus, passive fire protection systems, such as sprayed fire-resistive material (SFRM), play a critical role in mitigating the effect of the post-earthquake fire on the structural system in a building. However, SFRM may also be damaged by becoming detached. Both experimental and field observations have indicated that SFRM can delaminate under static and dynamic load situations (Arablouei and Kodur, 2015; Braxtan and Pessiki, 2011; Kodur and Arablouei, 2015; Wang et al., 2013). SFRM delamination is most likely to occur in locations of high strain, such as where plastic hinges form.

The effect of fire insulation delamination on the steel can jeopardise the structural stability of the building. There are several studies focused on studying the effect of fire insulation delamination on the fire resistance. Tomecek and Milke (1993) investigated the behaviour of steel columns with loss of protection material based on predicted thermal response and thermal criteria. The results showed that there is significant reduction in the fire resistance of steel columns. Milke et al. (2002) also investigated the effect of the loss of fire protection on the thermal response of the steel column. They found that the temperature rise in the column is mainly caused by the area of missing fire protection regardless of the protection thickness. They concluded that the column size has a relatively small effect on the reduction in fire resistance. Wang and Li (2009) conducted an experimental study to investigate the behaviour of steel columns with partial loss of fire insulation. They found that the failure mode of specimens can be yielding when the fire protection damage is short or buckling when the fire protection damage is long.

The above studies indicate that passive fire protection system damage has a significant role in determining the vulnerability of steel structures to FFE. Therefore, it is important to quantify the delamination in the assessment of a building subjected to FFE. Accordingly, in this study, it is assumed that fire insulation delamination may occur at both ends of the primary beams and at the bottom of the columns where the plastic hinges may occur during an earthquake, as shown in Figure 2. In these delamination regions, the steel is assumed unprotected. The length of delamination is assumed to be 5% or 10% of the member length.

Figure 2.

Fire insulation delamination on the beam and column.

Residual deformation

Residual frame deformation is the irreversible lateral deformation that remains after an earthquake. This deformation can be dangerous since it enhances P–Δ effects that can increase stress on the structure under loads acting after the earthquake as illustrated in Figure 3. The repetitions of plastic deformations during the earthquake also cause some reduction of mechanical properties. However, for MRFs designed according to Eurocode 8, the plastic deformation is relatively small. In this case, the reduction of mechanical properties is negligible (Della Corte et al., 2003). Therefore, an ideal elastic-perfectly plastic structure with a non-degrading component is considered in this study.

Figure 3.

Residual deformation of structure after earthquake.

As discussed in Bruneau et al. (1997) and Della Corte et al. (2002), the structural damage is mainly caused by IDR. Thus, there is good agreement that the IDR can be used for measuring damage of a frame under an earthquake. The IDR is the ratio between storey displacement relative to the adjacent storey and the storey height. IDR may also be associated with local plastic deformation (Gupta and Krawinkler, 2000).

The level of residual deformation after an earthquake was here estimated by performing a ‘pushover analysis’. In a pushover analysis, the building is pushed incrementally using a specific lateral load to arrive at a target displacement, and the load is then reduced to zero again. Lateral storey forces on the structure are applied in proportion to the product of storey mass and fundamental mode shape (ATC, 1996). The target displacement (represented by the roof displacement) is the expected displacement of the building when subjected to a design earthquake. The residual deformation pushover analysis results can be used as an initial step for the subsequent fire event. Pushover analysis is an approximate method, and dynamic effects are not directly considered.

According to ASCE standard 41-06 (ASCE, 2007), the level of earthquake damage can be categorised into three different performance levels: immediate occupancy (IO), life safety (LS) and collapse prevention (CP) that represent minor to major damage. These performance levels are tied to the IDR as an indication of global stability of the structure. The IDR value is less than 0.7%, 0.7%–2.5% and 2.5%–5% for performance level of IO, LS and CP, respectively.

This study considers two different levels of damages, damage 1 and damage 2. Damage 1 is LS level with IDR 2%, and damage 2 is CP level with IDR 4.5%. Local buckling is not a possible failure mechanism in the composite structural form analysed and so was not considered here. This assumption is in line with the philosophy of seismic design that performance levels of structures do not exceed the intended level when subjected to design earthquake. Material deterioration other than plasticity was also not considered. Steel in the case of damage states considered here is unlikely to suffer permanent changes to its key parameters. Concrete may suffer from cracking during earthquake, which is crudely captured in the model and possibly spalling during a fire. However, experimental and real-fire experience suggests that these effects are at worst secondary matters when considering structural behaviour during a fire. Spalling in the case of steel-deck composite construction is largely contained, for example. According to previous studies, spalling becomes important in elements with more than 4- to 5-cm cover (Majorana et al., 2010) or made of high-strength concrete (Kodur, 2005).

The two damage types discussed above are considered to represent earthquake damage in this study. It is apparent that when fire insulation delamination occurs, residual deformation always exists as a result of plastic deformation. Thus, the combined effect of fire insulation delamination and residual deformation could be considered. However, the aim of this study is to examine and quantify the effect of the two damage types separately in order to investigate which damage most significantly affects fire behaviour. For this reason, the two types of damage have been analysed separately in this study.

Post-earthquake fire loading and failure

A total of two different fire scenarios, a ‘standard fire’ (ISO, 1975) and parametric fire EN 1991-1-2 (CEN, 2002b) as shown in Figure 4, were used to simulate post-earthquake fires in this research. The fire load of 511 MJ/m² taken in the parametric fire is calculated according to Annex E of EN 1991-1-2 (CEN, 2002b), which is for office buildings. It should be noted that the parametric fire curve depends on several factors, such as ventilation, fire load, fire protection, and so on. When an earthquake occurs, these factors may change, which makes the appropriate parametric fire curve for use before an earthquake different from that for use after an earthquake. Opening factors of 0.06 and 0.08 are taken for the parametric fire before and after earthquake, respectively. This assumption is considered as there is a possibility of window breakage or damage to the building envelope after an earthquake that can increase the opening factor of the compartment. Moreover, firefighting measures may be different between fire in normal conditions and fire after earthquake. Using appropriate parameters from Table E.2 in Eurocode EN 1991-1-2 (2002b), the assumed parametric fire after an earthquake (PEF) results in higher temperatures and a longer period than those before an earthquake (F) as shown in Figure 4.

Figure 4.

Gas temperature–time curves.

Since in modern buildings, open-plan offices are the most common, full-floor compartments are considered and the hot gases are assumed uniform in the fire compartment. With these gas temperature–time curves, the procedure in Eurocode EN 1993-1-2 (CEN, 2005b) was used here to calculate the steel temperatures and a numerical heat transfer analysis to obtain concrete slab temperatures. Temperature–time curves for each fire scenario and for the steel and concrete are plotted in Figure 5.

Figure 5.

Structural temperature: (a) standard fire, (b) parametric fire before earthquake (F) and (c) parametric fire after earthquake (PEF).

In this study, two approaches are applied to determine the capacity of the frame structure exposed to post-earthquake fire. First, failure is defined when the primary beams are not able to transfer loads effectively or to reliably provide compartmentation. This stage can be crudely characterised when the mid-span deflection of a beam exceeds L/20, where L is beam span (Dwaikat and Kodur, 2011). Second, failure is defined by buckling of columns which is identified when vertical displacement of the top of a column rapidly reduces as temperature increases (Rezvani and Ronagh, 2015). It should be acknowledged that in both cases, the notional threshold of failure is somewhat arbitrary. However, for the purpose of this study, these serve as a reasonable and practical measure of structural performance.

Numerical model and validation

The general finite element software ABAQUS was used to analyse the generic building. Steel beams and columns were discretized using two-node linear beam elements (B31). The concrete slabs were modelled using four-node shell elements with reduced integration (S4R). To represent rebar in the slab, layers with appropriate steel material properties were specified in the shell elements. Tie constraints between steel beams and the concrete slabs were applied to represent full composite action. For simplicity, it was assumed that the beam-to-column and secondary beam-to-primary beam connections behaved as rigid and pinned connections, respectively, which approximates common practice for steel frames designed for seismic regions. Connection failures are not considered here.

Structural performance of elements under fire can be affected by surrounding structure, due to the effect of thermal restraint (Ali and O’Connor, 2001; Laím and Rodrigues, 2016; Rodrigues and Laím, 2017; Valente and Neves, 1999). The axial restraint can generate extensive unforeseen forces in the element during fire that may cause unpredictable structural failure, in contrast to the rotational restraint, which can avoid a sudden failure. Thus, 3D frames as shown in Figure 6 are developed to consider thermal restraint affected by surrounding structure. A mesh size of 0.5 m × 0.5 m is used for the first floor slab, and a mesh size of 1.0 m × 1.0 m is used for the upper slab to save the computing cost. A total of 8 elements are meshed for all columns and 18 elements for the beams.

Figure 6.

Finite element mesh of the generic frame.

A non-linear static analysis procedure was used because it was less time-consuming and easier to interpret than a dynamic procedure. Numerical convergence problems were prevented by use of artificial viscous damping in the analyses. An appropriate dissipated energy fraction should be specified by trial and error (Abaqus, 2014). However, the results’ accuracy can be affected significantly if the fraction is too high. In this research, the default energy dissipation factor of 0.2 × 10³ is used, and the ratio of the energy dissipated by viscous damping to the total strain energy is limited by an accuracy tolerance of 5%.

Using existing experimental and analytical work available in the literature, a series of validations were carried out to confirm that the results from the analysis give an acceptable level of accuracy. To validate the pushover analysis, the ‘European calibration frame’ previously analysed by Vogel (1985) was chosen. The frame was subjected to proportionally distributed gravity loads and concentrated lateral loads as shown in Figure 7. The yield stress of all members was 235 MPa, and Young’s modulus was 205,000 MPa. The result in Figure 8 shows good agreement between this study and the previous analysis and confirms that the analysis considers P–Δ effects correctly.

Figure 7.

Dimension and loading condition of the European calibration frame.

Figure 8.

Load-deflection behaviour of the frame.

To validate the fire analysis, a steel–concrete composite frame was chosen which is a simplified version of the Cardington British Steel fire test (Kirby, 1998) and which was previously analysed by Gillie (2009). Figure 9 shows the geometry of the composite frame. A load of 5.48 kN/m² is applied as a gravity load over the slab. Temperature loadings are applied on the shaded area. The secondary beam is heated to 800°C. The lower surface of the slab is heated to 600°C with a linear gradient of 4.6°C/mm. The structure was then cooled to ambient temperature. The comparison in Figure 10 shows that the general pattern and magnitude of the deflections are in good agreement in all cases and also confirm that the model can be used for thermal–mechanical analysis of steel–concrete composite frames.

Figure 9.

Geometry of simplified version of the first Cardington test (Gillie, 2009).

Figure 10.

Comparison of mid-span beam deflections.

Results and discussion

Impact of delamination of fire insulation on the primary beams

First, frames with and without delamination of the fire protection on the beams were analysed. The unprotected secondary beams, the primary beams with and without delamination and the slab were heated. In order to clearly investigate the effect of delamination on the primary beams, the column remained at ambient temperature. This is because heating columns would conceal the influence of delamination on the primary beams.

Figure 11 shows mid-span deflections of primary beam B12 (see Figure 1). It can be seen that the beam deflection exceeds the limit for all fire scenarios except under a parametric fire before earthquake. In general, the delamination reduces failure time of the beams, but the effect of the length of the delamination on the beam deflections is small.

Figure 11.

Mid-span beam deflection of primary beam B12 against beam temperature: (a) ISO 834 standard fire, (b) EC parametric fire before earthquake and (c) EC parametric fire after earthquake.

It can be observed that the influence of fire insulation delamination is negligible, up to 130°C. But beyond this temperature, there is a sudden increase in deflection. This is expected because the temperature of the delamination region reaches 400°C, the point at which the yield stress of steel starts decreasing. The differences become more obvious when temperatures increase further. However, as can be seen in Figure 11, the differences become insignificant again when the temperature is above 800°C using the standard fire. At this point, the temperature difference between protected and unprotected beams becomes small (due to prolonged high temperatures overwhelming the fire protection) so that the delamination has a small effect. Besides, at high temperature, the role of the concrete slab becomes dominant in supporting the loads since the beams have lost almost all strength and stiffness.

Figure 12 shows a comparison of the vertical deflection contours of concrete slabs without delamination and with 10% delamination. It shows that the delamination influences tensile membrane action in the concrete slab. When no delamination occurs, tensile membrane action can be fully mobilised and the slab bends into a bowl-like shape. However, the delamination on the beam reduces the vertical support to the slab. The tensile membrane action is reduced, and the deflected shape is more like that of a one-way slab or like catenary action in a beam. This behaviour will have a negative effect on structural strength since catenary action provides a weaker load-carrying mechanism than tensile membrane action. In this case, the benefit of tensile membrane action cannot be assumed in design.

Figure 12.

Concrete slab vertical deflection contours after 90-min standard fire exposure (unit in m).

Impact of fire insulation delamination on the columns

In this case, the delamination occurs only in the columns. The whole ground floor including beams, columns and slab are exposed to fire. As discussed earlier, the column failure is defined when the vertical displacement of the column drops suddenly. Figure 13 shows vertical displacement of the top of column C2 (see Figure 1) for all cases. The fire resistance time of the column very significantly reduces (by around 70%) when the delamination occurs.

Figure 13.

Vertical displacement of C2: (a) ISO 834 standard fire, (b) EC parametric fire before earthquake and (c) EC parametric fire after earthquake.

It can be observed that the column vertical deflections are initially positive due to thermal expansion. After a smooth and low rate of generation, the vertical deflection reaches a maximum value and then starts decreasing as the mechanical properties of the steel column reduce with time and temperature. At failure, the column starts to buckle, and the deflections reverse, when the axial force reaches a peak as shown in Figure 14.

Figure 14.

Axial force of C2: (a) ISO 834 standard fire, (b) EC parametric fire before earthquake and (c) EC parametric fire after earthquake.

Figure 15 shows horizontal displacements against the height of column at certain gas temperatures. It can be seen that the effect of delamination is still negligible at gas temperatures up to 500°C. After that, the delamination increases the horizontal deflection of the column. The deflection of the column with 10% delamination is much higher than that of the column with 5% delamination at high temperature. This deflection increases P–Δ effects on the columns, but it is essentially insignificant on the fire resistance of the columns. This is in line with the previous study by Wang and Li (2009) that showed that the length of fire insulation delamination on the column has small effect on the failure time.

Figure 15.

Horizontal displacement of C2 exposed to ISO 834 standard fire: (a) gas temperature at 500°C, (b) gas temperature at 700°C and (c) gas temperature at 825°C.

Impact of residual deformation

As explained in the section titled ‘Earthquake damage’, the residual deflection damage is represented by pushing the building horizontally prior to fire. This can be achieved by performing a three-step procedure. First, the building is subjected to gravity load. Second, a non-linear pushover analysis is performed to simulate the earthquake. Figure 16 shows lateral load versus top storey displacement. It can be seen that the pushover analysis results in residual deformation that can be used as the initial step for the subsequent fire event. Damage 1 and damage 2 have residual deformations at the first floor level of 0.08 m (IDR 2%) and 0.18 m (IDR 4.5%), respectively. Finally, a thermal–mechanical analysis of the damaged frame is conducted to simulate a fire.

Figure 16.

Pushover curves for damage 1 and damage 2.

Figure 17 shows mid-span deflection of primary beam B12 with and without damage. In general, the fire resistance on the beam slightly reduces due to the initial deflection after damage. It can be seen that there is an initial deflection in the primary beam for the damaged frame. However, the patterns of deflection in fire in the undamaged and damaged frames are very similar. For the case of standard fire and parametric fire after earthquake, the analysis stops before the beam deflections exceed the limit since failures in the heated column due to buckling occur earlier than failure in the heated beams. Hence, fire resistance of the composite frame is determined by failure in the column.

Figure 17.

Mid-span deflection of primary beam B12: (a) ISO 834 standard fire, (b) EC parametric fire before earthquake and (c) EC parametric after earthquake.

Figure 18 shows vertical deflection of the column C2. As can be seen, there is no collapse in the building when subjected to parametric fire before earthquake, but in other cases, runaway failure occurs. Overall, the damaged structure is more vulnerable than the undamaged structure. This is because the residual deformation increases moments in the column after earthquake due to P–Δ effects. The residual deformations reduce failure time from 98 to 92 min and from 78 to 76 min when exposed to standard fire and parametric fire after earthquake, respectively.

Figure 18.

Vertical displacement of C2: (a) ISO 834 standard fire, (b) EC parametric fire before earthquake and (c) EC parametric fire after earthquake.

Summary

This study has presented a numerical investigation of a steel-framed composite floor system under FFE. A total of two types of earthquake damage, fire insulation delamination and residual deformation, were considered. Failure of the structure was defined using two measures, a beam deflection exceeding span/20 and column buckling.

For clarity, Table 1 shows failure times for all cases examined. It can be seen that the earthquake damage can significantly reduce the fire resistance of the composite building. The reduction in fire resistance times result mainly from fire insulation delamination, particularly in the columns, rather than residual deformation. Moreover, the results show that there is no failure for the structure exposed to FFE using a parametric fire suitable for the pre-earthquake condition except when the delamination occurs in the columns. Therefore, the building can resist the applied load during earthquake or fire separately. However, the earthquake event can result in damage to the structures that make the structure more vulnerable to post-earthquake fire than the corresponding undamaged structure. Besides, a fire after earthquake can be more onerous than that in normal conditions since an earthquake may increase the compartment opening factor by breaking the windows and also by affecting the efficiency of firefighting.

Table 1.

Failure time (min).

Delamination	Standard fire	Parametric fire (F)	Parametric fire (PEF)
Delamination on beam
0%	130	–	100
5%	121	–	92
10%	120	–	90
Delamination on column
0%	100	–	77
5%	20	23	21
10%	20	20	20
Residual deformation
Undamaged	98	–	77
Damage 1	98	–	77
Damage 2	92	–	76

Conclusion

Based on this study, the following conclusions can be made for the structural configuration and damage types investigated:

Delamination of fire protection at the ends of primary beams, as is likely with seismic event induced damage to ‘strong-column, weak-beam’ structures, means that tensile membrane action may not be fully available as a reserve load-carrying mechanism in FFE (Figure 11). This has significant consequences for design because the benefits of tensile membrane action are often used for performance-based fire design of composite structures. Either designers should specify fire protection systems in these regions that are resistant to the deformations and strains that are likely to develop or the benefits of tensile membrane action should not be relied upon. The results are only weakly dependent on the length of the assumed delamination, so consideration of beam length comparable with a plastic hinge is sufficient.

Delamination of fire protection at the bottom of columns will cause a large reduction in fire resistance time (up to 70%) because columns will buckle with no alternative load paths available. This finding is in line with earlier experimental work (Kirby, 1998) and consequent recommendations that columns are always fully fire protected. For design purposes, either columns with sufficient bending strength to avoid delamination of fire protection should be used (with due consideration of the implications of this for seismic design) or, again, deformation resistant fire protection systems should be specified.

Residual lateral deflections of earthquake-damaged structures have a significant but lesser effect on their fire resistance. The increase in P–Δ moments in the columns may lead to premature failure earlier than in an undamaged structure, and this should be accounted for in design, but no particular changes to fire protection or structural systems appear to be required.

Fire protection systems that are tolerant of substantial strains should be investigated for application to building structures. Such protection systems do exist for application where blast resistance is needed, but they are expensive. It may be that local application of such systems in regions where plastic hinges are likely to form due to earthquake damage would be a practical and efficient solution.

The choice of design fire for use in an FFE analysis is not obvious. This study shows significant differences in predicted response between standard and parametric fires and between different parametric fires. The best choice of parametric fire is probably structure dependent and may not be apparent at the design stage. A range of likely fires should be considered if performance-based fire design is used for FFE loading.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship and/or publication of this article: This PhD research was funded by the Indonesia Endowment Fund for Education (LPDP).

ORCID iD

Riza Suwondo

References

Abaqus (2014) ABAQUS/Standard Analysis User’s Manual v14.2. Johnston, RI: SIMULIA.

Ali

O’Connor

(2001) Structural performance of rotationally restrained steel columns in fire. Fire Safety Journal 36: 679–691.

Arablouei

Kodur

(2015) Dynamic delamination of fire insulation applied on steel structures under impact loading. International Journal of Impact Engineering 83: 11–27.

ASCE (2007) Seismic Rehabilitation of Existing Buildings (ASCE/SEI 41-06). Reston, VA: ASCE.

ATC (1996) Seismic Evaluation and Retrofit of Concrete Building (ATC-40). Redwood City, CA: Applied Technology Council.

Behnam

Ronagh

(2015) Post-earthquake fire performance-based behavior of unprotected moment resisting 2D steel frames. KSCE Journal of Civil Engineering 19: 274–284.

Braxtan

Pessiki

(2011) Postearthquake fire performance of sprayed fire-resistive material on steel moment frames. Journal of Structural Engineering 137: 946–953.

Bruneau

Uang

C-M

Whittaker

(1997) Ductile Design of Steel Structures. USA: McGraw-Hill Professional.

CEN (2002a) Eurocode 1: actions on structures – part 1–1: general actions – densities, self-weight, imposed loads for buildings (EN 1991-1-1).

10.

CEN (2002b) Eurocode 1: actions on structures – part 1–2: general actions. Actions on structures exposed to fire (EN 1991-1-2).

11.

CEN (2004a) Eurocode 4: design of composite steel and concrete structures – part 1–1: general rules and rules for building (EN 1994-1-1).

12.

CEN (2004b) Eurocode 8: design of structures for earthquake resistance – part 1: general rules, seismic actions and rules for building (EN 1998-1-1).

13.

CEN (2004c) Eurocode 2: design of concrete structures – part 1–2: general rules – structural fire design (EN 1992-1-2).

14.

CEN (2005a) Eurocode 3: design of steel structures – part 1–1: general rules and rules for buildings (EN 1993-1-1).

15.

CEN (2005b) Eurocode 3: design of steel structures – part 1–2: general rules – structural fire design (EN 1993-1-2).

16.

Della Corte

De Matteis

Landolfo

et al . (2002) Seismic analysis of MR steel frames based on refined hysteretic models of connections. Journal of Constructional Steel Research 58: 1331–1345.

17.

Della Corte

Landolfo

Mazzolani

(2003) Post-earthquake fire resistance of moment resisting steel frames. Fire Safety Journal 38: 593–612.

18.

Dwaikat

MMS

Kodur

VKR

(2011) A performance based methodology for fire design of restrained steel beams. Journal of Constructional Steel Research 67: 510–524.

19.

Faggiano

(2007) Fire after earthquake. In: Proceeding from WG1 meeting, Urban Habitat Constructions under Catastrophic Events. Available at: http://fire.fsv.cvut.cz/COST_C26_Prague/07-03-30_31/1-05%20Faggiano.pdf

20.

Faggiano

Esposto

Mazzolani

(2008) Risk assessment of steel structures under fire. In: The 14th world conference earthquake engineering, Beijing, China, 12–17 October.

21.

Gillie

(2009) Analysis of heated structures: nature and modelling benchmarks. Fire Safety Journal 44: 673–680.

22.

Gillie

Usmani

Rotter

(2001) A structural analysis of the first Cardington test. Journal of Constructional Steel Research 57: 581–601.

23.

Gupta

Krawinkler

(2000) Estimation of seismic drift demands for frame structures. Earthquake Engineering Structural Dynamics 29: 1287–1305.

24.

ISO (1975) Fire resistance test-elements of building construction (ISO 834:1975).

25.

Kirby

(1998) The behaviour of multi-storey steel frame building in fire. Technical report. United Kingdom: British Steel Swinden Technology Centre.

26.

Kodur

Arablouei

(2015) Effective properties of spray-applied fire-resistive material for resistance to cracking and delamination from steel structures. Construction and Building Materials 84: 367–376.

27.

Kodur

VKR

(2005) Guidelines for fire resistance design of high-strength concrete columns. Journal of Fire Protection Engineering 15: 93–106.

28.

Laím

Rodrigues

JPC

(2016) Numerical analysis on axially-and-rotationally restrained cold-formed steel beams subjected to fire. Thin-Walled Structures 104: 1–16.

29.

Majorana

Salomoni

Mazzucco

et al . (2010) An approach for modelling concrete spalling in finite strains. Mathematics and Computers in Simulation 80: 1694–1712.

30.

Memari

Mahmoud

Ellingwood

(2014) Post-earthquake fire performance of moment resisting frames with reduced beam section connections. Journal of Constructional Steel Research 103: 215–229.

31.

Milke

Ryder

Wolin

(2002) An investigation of the reduction in fire resistance of steel columns caused by loss of spray applied fire protection. Journal of Fire Protection Engineering 12: 31–44.

32.

Rezvani

Ronagh

(2015) Structural response of a MRF exposed to travelling fire. Proceedings of the Institution of Civil Engineers: Structures and Buildings 168: 619–635.

33.

Rodrigues

JPC

Laím

(2017) Fire response of restrained composite columns made with concrete filled hollow sections under different end-support conditions. Engineering Structures 141: 83–96.

34.

Scawthorn

(2008) Fire following earthquake. USGS file report 2008-2015, USA.

35.

Scawthorn

Eidinger

Schiff

(2005) Fire following earthquake. Technical Council on Lifeline Earthquake Engineering Monograph 26. Reston, VA: American Society of Civil Engineers.

36.

Suwondo

Gillie

Bailey

et al . (2017) Post-earthquake fire behaviour of composite steel-framed structures. In: Proceedings of the international conference of applications of structural fire engineering conference, Manchester, 7–8 September.

37.

Tomecek

Milke

(1993) A study of the effect of partial loss of protection on the fire resistance of steel columns. Fire Technology 29: 3–21.

38.

Valente

Neves

(1999) Fire resistance of steel columns with elastically restrained axial elongation and bending. Journal of Constructional Steel Research 52: 319–331.

39.

Vogel

(1985) Calibrating frame. Stahlbau 54: 295–301.

40.

Wang

W-Y

G-Q

(2009) Behavior of steel columns in a fire with partial damage to fire protection. Journal of Constructional Steel Research 65: 1392–1400.

41.

Wang

W-Y

G-Q

Kodur

et al . (2013) Approach for modeling fire insulation damage in steel columns. Journal of Structural Engineering 139: 491–503.

42.

Zaharia

Pintea

(2009) Fire after earthquake analysis of steel moment resisting frames. International Journal of Steel Structures 9: 275–284.