Introduction

Steel-concrete composite elements consist of steel profiles combined with concrete, working together to withstand loads. These composite elements offer several advantages over both steel and concrete structures. Compared to concrete structures, notable benefits include the potential elimination of formwork and shoring, reduced construction time, increased dimensional accuracy and reductions in self-weight and structural volume, which lead to decreased foundation volume and costs. In comparison to steel structures, significant advantages include reduced consumption of structural steel, less need for fire and corrosion protection for steel profiles and increased structural rigidity (Rossi et al., 2020).

Structures must be designed to withstand all applied loads throughout their service life, ensuring adequate levels of safety, performance and durability while maintaining compatible construction and maintenance costs. In this context, connections, also known as “nodes”, play a crucial role in achieving these objectives. They are responsible for joining parts of the structure or connecting a part of it to an external element. In other words, connections are devices or means of connection responsible for transferring loads between the elements that constitute the structure and/or between its supports (Nicoletti et al., 2023).

In this context, steel-concrete composite connections are those where the transfer of forces between structural elements, as well as rigidity and strength, are provided simultaneously by both steel and concrete components. These connections can be utilized in beam-column and beam-beam joints. The primary advantage of composite connections is their increased strength and rigidity compared to solely steel connections (He et al., 2022; Nicoletti et al., 2023).

In addition to the significance of the connections themselves, it is important to highlight that the safety of structures during fire situations has become a prominent global concern. Analyzing structural safety in these scenarios has two main objectives: protecting the lives of building occupants and minimizing property damage. In this context, fire safety heavily relies on understanding fire dynamics. Accurate interpretation of fire scenarios enables designers to develop systems that improve risk management (Kodur; Kumar; Rafi, 2020).

Regarding the behavior of steel-concrete composite connections in fire situations, various knowledge gaps have been identified and/or proposed for future research in the literature. Chronologically, the following points can be noted:

1. Wald et al. (2006) emphasized that technical standards and their standard fire testing methods are based on isolated elements, which do not adequately reflect the global behavior of the structure. Specifically, Wald et al. (2006) stated that connections analyzed globally perform better in fire situations than the same connections analyzed in isolation;

2. Ranzi and Bradford (2007) stated that there was a lack of studies characterizing the behavior of shear connectors at elevated temperatures. Additionally, the authors advocated for the need for new experimental studies to better identify their constitutive models in both the longitudinal and transverse directions;

3. Santiago et al. (2008) advocated for the need for additional studies on steel-concrete composite connections subjected to elevated temperatures to optimize the component method;

4. Dong, Zhu and Prasad (2009) experimentally evaluated the thermostructural behavior of steel-concrete composite frames in fire situations. The authors studied the number and location of compartments heated by the furnace. The structural behavior was highly dependent on the number and location of compartments subjected to thermal loading. This underscores the importance of considering the global behavior of the structure in fire designs, rather than focusing solely on isolated elements. Building on these experiments, Dong, Zhu and Prasad (2009) noted the significance of considering the global behavior of structures, with appropriate interactions between elements, in fire design, as considering only the behavior of isolated elements leads to inaccurate results;

5. Dong and Prasad (2009a, 2009b) mentioned that experimental results clearly point out the need for comprehensive numerical modeling and analysis of these experiments to gain a better understanding of the fire-induced structural response of frames. It is also recommended that more resources should be allocated to perform experiments on larger and more complex structural systems under fire loading to capture three-dimensional effects;

6. Pucinotti et al. (2011) advocated for additional studies to understand the behavior of steel-concrete composite connections in fire situations regarding the influence of seismic actions and the slab type;

7. Haremza, Santiago and Silva (2013) highlighted and experimentally demonstrated the influence of axial restraints on the performance of steel-concrete composite connections in fire situations. The authors emphasized the importance of understanding the effect of flexural moment combined with axial loads on connections, especially during the cooling phase;

8. Agarwal, Selden and Varma (2014) recommended studying the influence of reinforcement in steel-concrete composite connections under fire conditions;

9. Pucinotti et al. (2015) commented on the necessity to investigate the influence of steel strength on the behavior of steel-concrete composite connections subjected to high temperatures;

10. Selden, Fischer and Varma (2016) recommended a detailed investigation into the behavior of steel-concrete composite connections during the cooling phase of a fire;

11. Song et al. (2017) suggested investigating the behavior of steel-concrete composite connections during the cooling phase as well as studying the catenary forces developed in the concrete slab;

12. Han, Xu and Tao (2019) mentioned the need for a study of steel-concrete composite connections involving stainless steel tubular columns filled with concrete in fire situations;

13. Ye et al. (2019) stated that there is a gap regarding the influence of cyclic loads on steel-concrete composite connections, particularly the effects of such loading on connections subjected to high temperatures;

14. Hajjar, Hantouche and Ghor (2019) emphasized the need to study the effect of creep in flexible steel-concrete composite connections;

15. Demonceau and Ciutina (2019) suggested studying the effect of cyclic loads on steel-concrete composite connections, in addition to negative moment and temperature effects; and

16. Liu, Hang and Burgess (2021a, 2021b) recommended investigating the influence of out-of-plane structure of the connection, particularly slabs, on the performance of steel-concrete composite connections, both at ambient temperature and in fire situations.

Overall, as highlighted by Wald et al. (2006), Dong, Zhu and Prasad (2009), Dong and Prasad (2009a, 2009b), Song et al. (2017) and Liu, Hang and Burgess (2021a, 2021b), there is a growing recognition of the need to study the global behavior of buildings in fire scenarios, particularly evaluating the role of connections in this context. Therefore, this study aims to assess, through numerical analyses, the influence of steel-concrete composite connections, fire location and the number of stories on the structural behavior during fire exposure. It is important to note that this paper is part of a broader study. Initially, the isolated behavior of two types of composite connections – composite flush end-plate and composite double web-angle connections – was examined under ambient and fire conditions. The subsequent focus of this study is to evaluate the influence of these connections on the overall structural behavior in fire situations. Consequently, beyond examining the global behavior of buildings, this research allows for a comparison of the performance of structural elements both in isolation and in an integrated system. It is important to note that no studies were found in the literature that conducted a global analysis of buildings under fire conditions, specifically evaluating the influence of beam-column connection type and the number of stories.

Methodology

The methodology of this study is divided into three main stages. The first stage provides general considerations for constructing numerical models. The second stage details the analysis of the isolated behavior of steel-concrete composite connections, focusing on determining the stiffness of beam-column composite flush end-plate connections and beam-column composite double web-angle connections under both ambient and fire conditions. Finally, the third stage outlines the considerations for analyzing building behavior under fire conditions, including the influence of beam-column connection type, number of stories and fire location.

Construction of numerical models and types of analysis

For the analysis of models under fire conditions, it is initially necessary to process thermal models, where the primary input data include not only the geometry of the model but also the time-dependent behavior of specific heat, thermal conductivity and the density of the materials constituting the model. Additionally, at this stage, it must also be specified which regions of the model are exposed to fire. To simulate these phenomena, the standard fire curve from ISO 834 (ISO, 1999) was used.

To model the fire action, the recommendations of EN 1991-1-2 (ECS, 2002) were considered, specifically:

1. aconvection heat transfer coefficient of 25 W/m²•K on exposed surfaces;

2. a convection heat transfer coefficient of 9 W/m²•K on unexposed surfaces;

3. an emissivity factor of 0.70 to account for heat transfer by radiation on exposed surfaces; and

4. a zero-emissivity factor on unexposed surfaces.

Additionally, it is important to note that the thermal properties of concrete, structural steel and reinforcing steel were assumed based on the recommendations of EN 1992-1-2 (ECS, 2004). In this context, the following should be highlighted:

1. EN 1994-1-2 (ECS, 2004) provides upper and lower limit values for thermal conductivity. In the numerical modeling, the average values between these limits were adopted;

2. the thermal elongation values proposed for limestone aggregate concrete were used; and

3. a moisture content of 3.0% by concrete weight was considered for specific heat calculations.

The Stefan-Boltzmann constant was assumed to be 5.67•10⁻⁸ W/m²•°C⁴ and the temperature corresponding to absolute zero was taken as -273 °C.

As a result, the processing of thermal models provides temperature isotherms over time for all elements constituting the model’s structure.

Next, the thermostructural models were developed. At this stage, in addition to the geometry, the main input data included boundary conditions, loading, material properties (density, modulus of elasticity, constitutive model and thermal elongation) and finally, the result file from the thermal analysis. Thus, when processing the thermostructural model, the temperature effects are simultaneously considered along with the applied loads on the model.

In summary, the thermal models consist of a single processing step in which the structure is heated based on the standard fire curve from ISO 834 (ISO, 1999). The thermostructural models, on the other hand, undergo two processing steps: the first involves the application of external loads and the second applies the temperature effects.

In the thermal models, a “Heat Transfer” analysis was used, with a total time of 7,200 seconds (two hours), an initial time increment of 1 second and minimum and maximum increments of 0.0001 seconds and 60 seconds, respectively, with a maximum temperature variation per increment of 300 °C.

In the thermostructural models, the processing step for external load application was modeled using “Static, General” analyses, with an initial time increment of 0.01 and minimum and maximum increments of 0.00001 and 0.1, respectively. The heating phase of the structure was then modeled using “Visco” analyses. The processing step size and increments were adopted to match those of the thermal models to maintain consistency with the temperature input.

In both steps of the thermostructural models, the “Dissipated Energy” stabilization method was employed with a coefficient of 0.002 and geometric non-linearities were included. The Newton-Raphson solution method was adopted and the stopping criterion was based on error in processing due to the loss of load-bearing capacity of the structural element.

Analysis of the isolated behavior of steel-concrete composite connections

As previously mentioned, the global analysis of buildings under fire situations, analyzing the influence of beam-column connection type and number of stories, is part of a broader study. The first step of this study consisted of analyzing the isolated behavior of steel-concrete composite connections.

Initially, numerical models were calibrated based on experimental tests from the literature. The model calibration, based on experimental results from existing literature, allowed for a more accurate determination of the numerical modeling strategies to be used in the proposed parametric study. The variables examined included the profiles used in beams and columns (in profile sets), the negative longitudinal reinforcement ratio, the degree of steel-concrete interaction, the slab type, the moment orientation, the concrete strength (considering both normal and high-strength concrete) and the steel strength (considering both conventional and high-strength steel). By parameterizing these variables, a total of 180 steel and concrete composite connection models were generated, as organized in the chart shown in Figure 1.

For subsequent analysis of the building’s behavior and to investigate the overall behavior of steel and concrete composite beam-column connections under fire conditions, six profile combinations were defined. For the composite beams, the VS 350 x 38 profile was selected as a pre-dimensioned option suitable for 8 m spans typically found in steel and concrete composite structures. Representative connections for corner, side and central columns were identified. Additionally, the study considered loads consistent with 10- and 3-story buildings to examine the impact of load levels. Given a column spacing of 8 m, steel with a yield strength of 25 kN/cm² and a distributed load of 10 kN/m² on the slabs, six column sections were dimensioned as presented in Table 1. The profile combinations were varied, resulting in six distinct sets outlined in Table 2. The dimensioning calculations were performed using Equation 1, where “A” is the required cross-sectional area of the column, “N” is the normal stress applied to the column and “fy” is the steel’s yield strength.

In the numerical models of the central and lateral columns (sets C2, C3, C5 and C6), cruciform configurations were adopted, where beams are connected to two opposite faces of the columns. In contrast, the numerical models representing corner columns (sets C1 and C4) feature a single beam segment connected to the column. Figure 2 illustrates an example of a cruciform numerical model and a corner column in a composite beam-column connection with double web angles.

The behavior of connections was examined under both ambient temperature and fire conditions for all model sets, considering the influence of six variables. Seventeen models were created for each profile combination to facilitate this analysis. Table 3 provides a detailed description of these models, including the parameterized variables and their corresponding values.

Numerical analyses were conducted using Abaqus^® software (Simulia, 2018), which supports linear, physical and geometrically nonlinear analyses. Additional details on the construction of numerical model portions and their associated parameters are documented in Nicoletti et al. (2023), which evaluates the behavior of steel-concrete composite flush end-plate connections under both ambient temperature conditions and fire scenarios.

Analysis of building behavior in fire situations based on beam-column connections, number of stories and fire location

To assess the influence of connections on the behavior of structures under fire conditions, eight building models were constructed in the Abaqus^® software (Simulia, 2018), subjected to vertical loadings. The parametric variables were the type of connection (beam-column connection stiffness), the position of the fire on the floor and the number of stories. Thus, eight exposed building models were analyzed as shown in Table 4. In turn, Figure 3 presents a flowchart of the parametrization of the building models.

The composite connections were represented by the strength and stiffness values obtained from the numerical models of isolated connections analyzed previously. These models were calibrated based on experimental results. It is worth recalling that isolated connection models were simulated, representing sections of lateral, central and corner columns of buildings with three and 10 floors.

Figure 4 depicts a schematic of the typical floor plan of the analyzed buildings. It was considered that the floor has a height of 3 meters.

The present research is part of a broader study. Initially, the behavior of composite flush end-plate connections and composite double web-angle connections was studied at ambient temperature and under fire conditions. These studies conducted parametric analyses evaluating the influence of the following variables: the load level, indirectly assessed through the profiles used in beams and columns (in profile sets), the position of the column in the floor plan, the negative longitudinal reinforcement ratio, the degree of steel-concrete interaction, the type of slab, the orientation of moments, concrete strength and steel strength.

For variations in load level and profiles, scenarios consistent with corner, side and central columns of 3-story and 10-story buildings were considered. Additionally, in the pre-dimensioning process, spans of 8,000 mm were considered, which are consistent with the use of steel-concrete composite beams. Further details on the parametric analysis and the preliminary studies conducted can be found in the Section “Construction of numerical models and types of analysis” of the present paper.

For the beams and columns, pre-dimensioned sections were adopted as determined previously. Specifically, the welded profile VS 350 x 38 was used for the beams and for the columns, the sections shown in Table 5 were utilized. In turn, Table 6 presents the stiffness values that were considered in each model, both at ambient temperature and in a fire situation. The values expressed in Table 6 consist of the stiffness values of the isolated connection models with default variables, determined through the analyses of isolated connections.

It is important to note that the stiffness values of the composite connections varied during the fire, considering the changes in deformations and displacements at the beam edges over time, as well as the variation in the applied force. To ensure structural safety, the minimum stiffness values were considered in the fire scenario analyses. Table 7 details the dimensions of the profiles used in each set.

Through this methodology, the aim was to evaluate the influence of connections on the structure’s behavior under fire conditions. Specifically, the main objectives of this study were to investigate the fire resistance time of the buildings and assess the influence of connection stiffness, fire floor location and force redistribution within the structure.

Given the extensive nature and number of elements in the model, beams and columns were represented using beam-type bar elements, specifically beam B31. Slabs were modeled with the shell element S4R. Abaqus^® (Simulia, 2018) allows the use of stress/displacement continuum shell elements in three-dimensional analyses. The S4R shell element is particularly suitable for structures where thickness is relatively small compared to lateral dimensions, making bending and shear effects negligible compared to stress in the direction normal to the surface. These two-dimensional elements have a simplified formulation, making them computationally efficient for structural analysis, especially in complex models.

To simplify the model and ensure computational feasibility, a “Tie Constraint” interaction was used between the edges of the slabs and the supporting beams. Table 8 details the mesh used for modeling the elements. A mesh sensitivity test was conducted to define the mesh for the parametric analysis. The computer used for processing was equipped with an Intel(R) Xeon(R) CPU E3-1225 3.30GHz and 16 GB of RAM. The element dimensions listed in Table 8 provided the best performance of the numerical model compared to the expected analytical results (Table 9). Finer meshes with smaller node distances resulted in increased processing times but stable results. For example, the processing time for the B03 numerical model (with 10 stories) was approximately 350 minutes (about 6 hours). Reducing the element spacing by 50% (to 25 mm for beams and columns and 125 mm for slabs) increased the processing time for the B03 model to approximately 720 minutes (12 hours), with no significant difference in results observed. The maximum deflections varied by less than 1%.

For the mechanical properties, Earls’ (1999a, 1999b) perfect elastoplastic stress-strain behavior was adopted for the steel elements (beams and columns). At ambient temperature, the longitudinal modulus of elasticity for steel was set to Ea = 200 GPa, with a yield strength of fy = 250 MPa an ultimate strength of fu = 400 MPa. Concrete was modeled solely with its elastic behavior, with Ec = 30,588.56 MPa. This approach was chosen due to the large scale of the numerical model, which includes a high number of elements and nodes. Modeling concrete with plastic behavior would require significantly higher computational resources.

To account for variations in the stress-strain curves of steel and concrete due to fire, reduction coefficients from European standards EN 1992-1-2 (ECS, 2002) and EN 1994-1-2 (ECS, 2005) were applied to the constitutive models at room temperature. Stress-strain curves for various temperatures (0 °C, 20 °C, 100 °C, 200 °C, 300 °C, 400 °C, 500 °C, 600 °C, 700 °C, 800 °C, 900 °C, 1000 °C and 1100 °C) were input as data in the materials tab of Abaqus^® software (Simulia, 2018).

It should be noted that, except for the region exposed to fire, the concrete elements were subjected to service loads resulting in stresses lower than the compressive strength of concrete class C30 (with a compressive strength of 30 MPa), which aligns with the Ec value used. However, in regions with increased temperatures, material degradation and stress redistribution could mobilize the plastic behavior of the concrete. Thus, while representing concrete solely with elastic behavior is a limitation of this study, the extensive model (particularly the slabs) and the high computational cost required to include plastic properties made it unfeasible. The available computers were unable to process even the initial increment of the model.

To account for the variation in the stress-strain curve of steel and concrete due to fire, the reduction coefficients of European standards EN 1992-1-2 (ECS, 2004) and EN 1994-1-2 (ECS, 2005) were applied to the constitutive models at room temperature. As input data in the materials tab of Abaqus^® software (Simulia, 2018), stress-strain curves for different temperature values were inserted (20 °C, 100 °C, 200 °C, 300 °C, 400 °C, 500 °C, 600 °C, 700 °C, 750 °C, 800 °C, 860 °C, 900 °C, 1000 °C and 1100 °C). Additionally, to consider the thermal expansion of concrete and steel, the curves from the European standards EN 1992-1-2 (ECS, 2004) and EN 1994-1-2 (ECS, 2005) were applied, respectively.

Furthermore, due to the fact that the beam-type element B31 lacks degrees of freedom that allow for heat transfer analyses, thermal analysis was not conducted separately. Specifically, a coupled analysis was performed, consisting of the following steps:

1. application of a uniformly distributed service load on the slab, with an intensity of 3 kN/m². The service load was assumed to be equivalent to 30% of the ultimate normal load, which was considered to be 10 kN/m². This approach aligns with the methodology used in the parametric analyses of isolated steel-concrete composite connections, as described in Section 2.1. Horizontal loads were not considered; and

2. heating of the beams using the “Predefined field” tool, with temperature evolution following the standard fire curve (ISO, 1999). It constitutes a pattern of temperature evolution internationally recommended in standards and testing procedures for practical reasons (Costa; Silva, 2006).

Figure 5 and Figure 6 illustrate the two positions in which the fire was considered. In both cases, the fire was assumed to occur on the first floor. Since the slabs were modeled as shell elements and the concrete was modeled with only its elastic properties, heating was applied solely to the beams and columns to ensure greater reliability in the numerical model. Moreover, the exposure of elements to the fire only at the corner and the center is consistent with the methodology employed in the isolated parametric analysis of the connections, which resulted in the determination of the stiffness values of the springs used in the beam-column connections of the building (Table 6). Additionally, this methodology incorporates fire compartmentalization and aims to investigate the effects of fires occurring in two different locations on the first floor of the building: in a compartment located at the corner of the first floor and in a compartment located at the center of the first floor.

It is acknowledged that the absence of consideration of direct exposure of the slabs to temperature variations may indeed influence the overall behavior of the structure. However, due to the type of finite element used for the slabs and the modeling restricted to the elastic properties of concrete, this methodology was chosen.

Additionally, regarding the definition of the regions exposed to heating, there are numerous possibilities. The approach adopted in this study is not necessarily the most critical but was considered consistent with previous studies (especially in the analysis of the isolated behavior of steel-concrete composite connections) and with the limitations of the available equipment for processing the numerical models. Furthermore, the fire exposure of steel elements, such as beams and columns, represents a more critical situation compared to that of a slab. In practice, slabs may have coverings on their upper and lower faces, which have the potential to delay the degradation of this structural element.

Figure 7 and Figure 8 respectively present the numerical models of buildings with 3 and 10 stories.

In the analysis of buildings under fire conditions, it is important to consider the possibility of significant deformation in the floor before structural collapse occurs. The resistance of the slab is directly related to the action of the tensile membrane within it, necessitating that such significant deformation occurs to activate the load-supporting mechanism provided by this membrane (O’Connor; Martin, 1998; Bailey, 2001; Vassart; Zhao, 2011).

It is important to emphasize that significant deformations in the floor can lead to a reduction in structural performance. This is due to the development of cracks in the concrete, excessive deformations in the reinforcement and subsequent changes in load distribution. Additionally, the floor’s inclination may be affected, along with a reduction in material strength when exposed to high temperatures. In this context, ISO 834 (ISO, 2014), along with several other studies (Tan; Nguyen, 2015; Piloto et al., 2017; Zheng; Zhang, 2016; Naser; Kodur, 2017), recommends considering a deflection of L/30 as the failure criterion for a structural element subjected to bending under fire conditions. For floors consisting of primary beams, secondary beams and slabs, it is advisable to establish a total deflection limit, accounting for the cumulative allowed deflections of each structural element. This approach is illustrated in Figure 9.

Thus, regardless of the distribution of beams, the deflection limit associated with failure is (L1 + L2)/30, where L1 represents the length of the secondary beams and L2 corresponds to the length of the primary beams. Therefore, irrespective of the beam distribution, the deflection limit associated with failure is (L1 + L2)/30, where L1 represents the length of the secondary beams and L2 corresponds to the length of the primary beams.

Other references, such as the European standard EN 1992-1-2 (ECS, 2004), establish limit deflection rates to assess structural elements subjected to bending under fire conditions. However, the criterion for the rate of deformation is not applied until the deflection limit of span/30 is exceeded (Vassart; Zhao, 2011).

To ensure the validity of the numerical model, a numerical model considering rigid connections was simulated. Table 9 presents the comparison between the vertical deflection at mid-span of some elements of the numerical model with the analytical deflection for the same elements. The analytical deflection was calculated using Equation 1.

Where δ is the analytical vertical deflection at the midpoint of the beam span, q is the uniformly distributed load on the beam, L is the longitudinal span length of the beam, Ea is the longitudinal modulus of elasticity of the steel and Itr is the transformed section moment of inertia.

The uniformly distributed load was obtained by decomposing the 3 kN/m² load on the beams, considering their load influence area. Additionally, the transformed section method involves converting the cross-section composed of the two materials into an equivalent section made of a single material. In this case, the geometry of the transformed composite section was obtained by converting the geometry of the concrete slab into a steel slab. For this purpose, the effective width of the slab was divided by α, where α = Es/Ec, which is the ratio of the modulus of elasticity of steel (Es) to the longitudinal modulus of elasticity of concrete (Ec).

Given that the maximum relative error was 14.3% and the maximum absolute error was 0.747 mm, the numerical model of the building was considered adequate to perform the parametric analysis.

Results and discussion

Figure 10 presents the behavior of maximum vertical deflection on the 1^st floor over the fire duration for building models with steel-concrete composite beam-column connections using non-extended end plates. In all cases, the location of the maximum deflection corresponded to the region where the structural elements were exposed to fire.

For a fire duration of 120 minutes, with a deflection limit of 26.6 cm (based on a span length of L = 800 cm and a deflection limit of L/30), Figure 10 shows that only model B02 exceeded this limit, doing so after approximately 45 minutes of fire exposure. This model is characterized by having three floors and the fire occurring in the central region of the floor.

It was also observed that the other three-floor model reached a maximum deflection of 24.9 cm, which is close to the limit, but only after 120 minutes of fire exposure. For the same 120-minute period, the 10-floor models (B03 and B04) exhibited maximum deflections of 11.6 cm and 18.7 cm, respectively. In general, the following observations were made:

1. when comparing the results of models where the fire occurred at the corner of the floor (B01 and B03) with those where the fire occurred at the center (B02 and B04), lower deflections were observed in the corner models. This is because structural elements at the edges are subjected to lower intensity loads; and

2. the 10-floor models, due to their more robust column sections, showed greater resilience in redistributing loads during the fire. None of these models exceeded the deflection limit of 26.6 cm. Additionally, in the 10-floor models, better redistribution of forces was noted. This behavior is attributed to the stiffer column sections and the larger number of beams, which facilitated a more efficient load path mechanism, as explained by the Vierendeel effect (Adam et al., 2020). Figure 10 shows that vertical deflection in these models stabilized at around 30 minutes of fire exposure.

Similarly to Figure 10, Figure 11 illustrates the behavior of maximum vertical deflection on the 1^st floor over the fire duration for building models with composite double web-angle connections.

Similarly to the behavior observed in building models with beam-column connections using non-extended end plates, only model B06 exceeded the deflection limit, which occurred at approximately 45 minutes. Overall, the connection type had minimal influence on the maximum vertical deflection during the fire. The largest difference was noted in models B01 and B05. The beam-column connection with a double angle section, being more flexible, allowed for faster redistribution of forces. Consequently, the maximum vertical displacement stabilized sooner, at around 60 minutes for B01 and 52 minutes for B05.

For a fire duration of 120 minutes, the maximum vertical deflection for B01 was 24.9 cm, whereas for B05, it was 13.1 cm, representing a 47.4% reduction. Figure 12 compares the vertical deflections on the first floor across all models after 120 minutes of fire exposure or upon reaching the deflection limit, as observed in models B02 and B06.

Figure 12 illustrates the influence of stiffer columns on the distribution of deflection across the floor. In models B03 and B07, the maximum deflection did not occur directly above the column.

It is also important to note that in the numerical analyses of isolated composite flush end-plate connections, a fire resistance time (TRF) of approximately 23 minutes was observed, with variations of less than three minutes between models. In contrast, models of isolated composite double web-angle connections exhibited a TRF of approximately 26 minutes, with variations of less than two minutes between models. Therefore, it is evident that the TRF was consistently higher in the global analyses of buildings. Even the models with the lowest TRF (B02 and B06) withstood fire exposure for approximately 45 minutes. The other six models endured 120 minutes of fire without reaching the deflection limit of L/30.

These findings underscore the importance of considering the global behavior of structures, including the proper interactions between elements, in fire safety design, particularly in connection analysis.

Conclusion

To evaluate the influence of connections, fire location and number of stories on the behavior of structures under fire conditions, eight numerical building models were developed. These models represent sections of lateral, central and corner columns in buildings with three and ten floors and were created using Abaqus® software. The connections were modeled based on previously determined stiffness values.

When comparing the results of steel-concrete composite structure models with fire located at the corner of the floor versus those with fire at the center, lower deflections were observed in the former. This difference is likely due to the fact that structural elements at the building’s extremities are subject to lower intensity loadings.

Additionally, in the 10-floor models, a more effective redistribution of forces was observed. This behavior is attributed to the stiffer column sections and the presence of a larger number of beams, which facilitated a more efficient load path, as explained by the Vierendeel effect. None of the 10-floor models exhibited vertical deflections exceeding the limit of 26.6 cm. Furthermore, in taller buildings, the maximum vertical deflection stabilized more quickly.

It is also important to highlight that the global analyses of the buildings consistently resulted in higher fire resistance times compared to isolated connection analyses, emphasizing the necessity and importance of conducting global assessments.

References

Publication Dates

History