Directions in structural‐fire safety design for steel buildingsa

Abstract

Fire safety considerations have an impact on the design of almost all structures, ranging from small residential structures to high‐rise buildings. One important aspect of building fire safety is structural‐fire safety, with the goal of preventing or delaying collapse of structures during severe fires. Structural‐fire safety is of particular importance for steel structures, because of the high thermal conductivity of steel. Structural‐fire safety design is most often accomplished using non‐engineered prescriptive approaches. However, there is increasing interest in engineered structural‐fire safety design for potential advantages in safety, economy and design flexibility. This paper provides an overview of engineered structural‐fire safety design for steel buildings, and discusses some of the challenges in this field.

Introduction

Over the last several decades, and particularly after the events of September 11, 2001, there has been on-going discussion and debate on the most appropriate approaches for providing structural-fire safety for buildings. There is strong interest in moving from the traditional non-engineered prescriptive-based approach for structural-fire safety to an engineered approach. This interest is driven by the need to provide improved safety for building occupants and first responders in building fires; to enable greater flexibility in fire safety design for complex and unusual buildings; and to provide more cost-effective fire protection.^{14,17,20,28,45,46}

While providing fire safety in buildings is an age-old concern, many issues, problems and concerns with current practices in the U.S. came into sharp focus with the fire-induced collapses of the 110-story twin towers of the World Trade Center and of the 47-story WTC Building seven following the attacks of September 11, 2001. The detailed investigation of these collapses by the National Institute of Standards and Technology^45,46 cited a number of critical concerns regarding inadequate knowledge of the structural engineering profession to predict response of structures to fire. NIST concluded their WTC study with a number of major recommendations related to structural fire safety that go beyond the events at the World Trade Center, and more broadly address important changes needed in U.S. building design and construction practices. More specifically, NIST cited the need to improve the technical basis for structural-fire engineering and the need to move toward performance-based fire safety design. NIST also noted the important need to educate structural engineers on basic principles of structural-fire safety design, as well as the need to train fire-protection engineers and fire services personnel on the behavior of structures subjected to fire.

The need for a better understanding of structural fire response and the need for better education in structural fire-safety design were key conclusions of a National Research Council study on Making the Nation Safe from Fire – A Path Forward in Research.⁴² These same conclusions were made in the NIST-SFPE Workshop for Development of a National R&D Roadmap for Structural Fire Safety Design and Retrofit of Structures.⁴⁴ This workshop examined critical research needs for the U.S. to move forward in our capability to better design structures for fire events and to reduce the high cost of fire protection. The 2007 National Workshop on Structures in Fire,²⁸ sponsored by NIST and the National Science Foundation, further emphasized the need to improve structural fire safety design practices in the U.S. and identified major gaps in research and education. The 2015 Forum on Performance-Based Structural-Fire Engineering¹⁷ organized by the Applied Technology Council examined the incentives for performance-based structural-fire safety design, emphasizing the highly uncertain level of safety provided by traditional non-engineered prescriptive approaches.

This paper provides a review of design approaches for structural-fire safety design of buildings. The emphasis of this paper is on steel building structures, although many of the issues discussed in this paper are applicable to other materials. Additionally, the emphasis of this paper is on U.S. practices for structural-fire safety design, although similar practices are employed in most countries.

The Role of Structural-Fire Engineering in Overall Building Fire Safety

A fundamental goal of building fire safety design is life safety. In the event of a building fire, the goal is to prevent deaths and serious injuries to building occupants and emergency responders. Additional goals can include limiting property damage, limiting the spread of fire to adjoining buildings, and limiting the environmental impact of a fire.

There are many aspects to building fire safety design, most of which are heavily regulated by building codes and standards in most countries. Some of the key aspects of building fire safety design include the following^43,50:

• Building layout
  • -Fire department access
  • -Escape routes (number, size and layout of hallways and stairwells, emergency fire escapes, etc.)
  • -Occupancy limits
  • -Compartmentation
  • -Places of safe refuge
• Limit ignition sources
  • -Safety requirements for building power systems, electrical wiring, mechanical equipment, etc.
  • -Limit activities permitted in building
• Limit combustible contents
• Limit fire hazard of building materials, finishes, and cladding
  • -Limits on flammability, speed of flame spread, heat release rates, etc.
  • -Limits on smoke and toxic gas generation
• Detection and alarm systems
• Active protection systems
  • -Sprinklers (water, foam, etc.)
  • -Smoke control systems (HVAC controls, pressurized stairwells, smoke vents and fans, etc.)
  • -Automatic door closing systems
  • -Automatic elevator controls
  • -Fire extinguishers
  • -Fire department response
• Passive protection systems
  • -Fire resistive compartment walls and ceilings
  • -Fire resistive doors
  • -Fire and smoke stops
• Human behavior factors
  • -Tenability limits
  • -Evacuation timing
  • -Response to fire cues
  • -Response to exit signs
  • -Special needs of mobility impaired
• Structural-Fire Safety

Many of the fire safety features listed above are intended to prevent building fires from occurring in the first place, or in the event that a fire occurs, to prevent small fires from becoming large fires. In cases where fires grow large, structural-fire safety becomes a primary concern. The high gas temperatures developed in a building fire, often in excess of 1000°C, can cause severe structural damage or structural collapse. By preventing collapse, structural-fire safety is essentially the last line of defense against large-scale fatalities and injuries, in the event that other fire safety features were absent, did not function, or were overwhelmed.

The primary objective of structural-fire safety design is to prevent or delay collapse of a building in the event of a large fire. This supports the broader goal of protecting life safety, as structural collapse clearly endangers occupants and first responders throughout the entire building, and not just in those areas impacted by flame and smoke. An additional objective of structural-fire safety for some structural elements, such as floor systems or walls, is to contain a fire in the compartment of origin in order to slow the spread of fire within the building or to adjacent buildings. This objective also supports the goal of protecting life safety.

Many of the building fire safety design features listed above do not involve structural engineering, and are typically the responsibility of fire protection engineers, contractors, or the architect. However, structural-fire safety, with the basic goal of preventing structural collapse, is clearly a structural engineering undertaking. As discussed later, however, on most building design projects, structural engineers play little or no role in structural-fire safety design.

Examples of Structural Damage and Collapse due to Fire

As a reminder of the severity of fire effects on structures, Figures 1 to 4 show examples of structural damage due to fire. Figures 1 and 2 show fire-induced collapse and damage to single story steel buildings located near Austin, Texas. Figure 3 shows the remains of the Windsor Building in Madrid, Spain after a severe fire in 2005 where portions of the upper stories completely collapsed.³⁷ The building was demolished shortly after the fire. Figure 4 shows the Faculty of Architecture Building at the Technical University in Delft, the Netherlands, after a fire in 2008 where an entire wing of the building completely collapsed.¹⁸ This building was also demolished shortly after the fire.

FIGURE 1. Fire induced collapse of a steel building near Austin, Texas in 2008

FIGURE 2. Fire induced damage to steel structure in Austin, Texas in 2009

FIGURE 3. Fire induced damage to Windsor Building in Madrid, Spain in 2005 (photo courtesy of J. Roesset)

FIGURE 4. Fire induced damage to Faculty of Architecture Building in Delft, the Netherlands in 200818

It is also important to recognize that fires in structures are a very common occurrence. For example, in 2019 in the U.S., a structure fire occurred, on average, once every 65 seconds.¹ Structural damage and collapse in high rise structures, such as that shown in Figures 3 and 4, are infrequent. However, the type of fire induced damage shown in Figures 1 and 2 are everyday occurrences.

The important point here is that, like wind and earthquake, fire can impose large demands on structures, with the potential for extensive structural damage and collapse. However, the difference is that in the current state of practice, structures are engineered for wind and earthquake loads by structural engineers. In contrast, the majority of structures are not engineered for fire, and indeed most structural engineers have little education or knowledge on how structures behave under fire loading.

Fire Effects on Structures

As described above, fires can cause severe structural damage or collapse. In this regard, it is important to recognize that the impact of a fire on a structure is the result of two major factors:

• effects of thermally induced forces and deformations
• effects of elevated temperatures on material properties.

In a simple sense, it can be stated that the primary effect of a fire on a structure is that the fire causes heating of structural elements. The heating of structural elements, in turn, results in the two major effects listed above. The first effect is that heating of structural members causes thermally induced deformations: i.e., thermal expansion, thermal contraction (during cooling phase of fire), thermal bowing, etc. When these thermally induced deformations are restrained, very large thermally induced forces are developed in structural members and connections. Thus, very large axial forces, bending moments, shear forces, etc. are generated in structural members and connections during both the heating and cooling phases of a fire. The investigation by NIST into the collapse of the 47-story World Trade Center Building 7 on 9/11⁴⁶ cited the thermal expansion of a long span floor system as a major contributing cause to the collapse.

The second major effect that comes from heating of structural members is degradation in material properties. Both steel and concrete materials lose strength and stiffness as temperature increases, resulting in significant reduction in member strength and stiffness. Thus, at elevated temperatures seen in a fire, the bending capacity of beams is reduced, the buckling capacity of columns is reduced, the shear strength of bolts is reduced, etc. Figure 5 shows elevated-temperature stress-strain curves for samples of ASTM A992 steel (specified F_y = 345 MPa, F_u = 448 MPa) plotted up to 0.5-percent strain. Note the significant reduction in strength and stiffness as temperature increases. Temperatures of unprotected steel in a severe fire can be on the order of 1000°C.⁹ As is evident from Figure 5, very little strength remains at this temperature. In the end, the development of structural damage or structural collapse in a fire is the result of some combination of these two major effects, i.e., the effects of thermally induced forces and deformations, and the effects of the degradation in material properties at elevated temperatures.

FIGURE 5. Stress-strain curves for samples of ASTM A992 steel for temperatures ranging from 20 to 1000°C32

Approaches to Structural-Fire Safety Design

For purposes of this paper, approaches to structural-fire safety design will be divided into two broad categories, as illustrated in Figure 6: (1) Non-Engineered Prescriptive Design, and (2) Engineered Design. The design approaches are then further subdivided into Element Level Design and System Level Design.

FIGURE 6. Approaches to structural-fire safety design

The current basis for providing structural-fire safety for the vast majority of buildings in the U.S. and in most other countries is the traditional non-engineered prescriptive approach. This approach entails following a series of rules prescribed in building codes, and is based on the concept of specified hourly fire resistance ratings for different elements of a structure, established using building standards such as ASTM E119⁶ and ISO 834.²³ To satisfy the prescriptive approach, structural elements (columns, beams, segments of composite floor systems, etc.) are tested in a furnace following a prescribed time-temperature curve that normally exposes the element to temperatures in excess of 1000°C. The element must survive this exposure for the specified hourly rating without failure. “Failure” is defined in different ways by these standards, depending on the element and on the standard. However, the failure criteria generally relate to loss of load carrying ability, or the loss of containment ability. Because of the limited size of furnaces, the elements tested are often a significantly reduced size compared to their actual use in a building. Detailed descriptions of prescriptive structural-fire safety requirements are available from many sources. See, for example, SFPE,⁵⁰ Buchanan and Abu,⁵ ASCE,⁹ Malhotra,³⁴ Wang et al.⁵⁶ and AISC.² Within the U.S. this basic prescriptive approach to structural-fire safety design has been in building codes and in practice for more than 100-years.⁴⁵ Passing fire resistance tests for structural steel elements most commonly requires insulating the steel members to limit temperatures in the member. This is commonly accomplished by application of Spray Applied Fire Resistive Materials (SFRM), commonly referred to as “fire-proofing” material. Other options of insulating steel include the use of gypsum board, intumescent coatings, concrete, and others.²

The prescriptive approach can be followed and applied in the absence of any understanding of structural behavior and in the absence of any structural engineering calculations. In many projects, the structural engineer of record has little or no involvement in the process of providing for structural-fire safety. Consequently, in this paper, this approach is referred to as non-engineered.

In contrast, in an engineered structural-fire safety design, fire is treated as a load on a structure, much the same as wind or earthquake loading. The response of the structure to design fire scenarios is calculated, and the structure and the structural fire protection are designed to achieve desired performance objectives while minimizing cost. In an engineered structural-fire safety design, structures are engineered for fire by structural engineers.

In the literature, engineered structural-fire safety design is often referred to as performance-based structural-fire safety design, or as performance-based structural-fire engineering. However, the term engineered structural-fire safety design is used in this paper to emphasize the fact that analysis and design for structural-fire safety is conducted by structural engineers with this approach, in contrast to the traditional prescriptive approach.

Predicting Structural Performance in Fire

Undertaking an engineered structural-fire safety design requires predicting the response of a structure to fire. This generally involves three steps. The first step is to estimate the thermal environment of a fire, normally characterized in terms of a gas temperature vs time curve. Guidance for characterizing the fire environment for structural-fire safety design is provided in the literature (see for example Refs. [9,15,26]) and in building standards.^11,51 Note that typical building fires reach temperatures on the order of 1000°C and maintain these high temperatures within a localized area of a structure for periods of time on the order of 20 to 60 minutes.^9,19,27,45

The second step in structural-fire analysis is heat transfer analysis to determine temperatures in structural elements as a function of time and location within the member. This involves modeling radiative and convective heat transfer from the hot fire gases and burning objects to the surface of the structural member, and conductive heat transfer within the member.⁵⁶ These heat transfer calculations can be done using a variety of methods and tools, including finite element programs. For typical building fires, peak temperatures in steel members and connections can reach on the order of 600°C for protected members (i.e., members insulated with SFRM) and on the order of 1000°C for unprotected member.

Once member temperatures have been determined, the final step in a structural-fire analysis is to compute structural response under the actions of these high temperatures combined with other loads on the structure at the time of the fire. This generally requires modeling temperature-dependent and time-dependent nonlinear material properties and nonlinear geometry.

Thus, the three keys steps in predicting structural response to fire are:

1. characterize thermal environment of fire
2. determine temperatures of structural elements
3. compute structural response (i.e. compute internal forces and stresses; compute deflections and rotations, etc.).

Each of the three steps listed above can be accomplished using a variety of approaches, methods and tools. However, the actual application of these techniques can be divided broadly into the following two approaches:

• element-level design;
• system-level design.

The concepts of element-level versus system level design approaches are discussed in the following section.

Element-Level versus System-Level Design

The element-level design approach considers the fire response of individual structural members, such as individual columns or beams. This approach gives only limited consideration to how the member interacts with the rest of the structural system. As such, the element-level design approach generally neglects the effects of thermally induced forces and deformations or considers them in a highly simplified manner, but still considers the effects of elevated temperatures on reducing the stiffness and strength of the member. The load acting on the member under consideration is normally taken as some combination of dead and live load, but does not include member forces generated by restrained thermal deformation. For example, in U.S. codes,⁴ the loads assumed to be acting on a structure during a fire are taken as 1.2D + 0.5L, where D is dead load and L is live load. For element-level fire design, the required strength of a member is taken as the forces (bending moment, axial force, shear forces, etc.) in the member under this load combination. Element-level structural-fire design often does not consider the response of connections, only that of members.

In a system-level design, the effects of a fire on the structural system are considered, rather than just on individual members. A system-level analysis considers both of the major factors affecting structural response in fire: the effects of thermally induced forces and deformations, as well as the effects of temperature induced loss of member stiffness and strength. A system-level analysis normally considers a more comprehensive model of a structure or portion of a structure, including structural members, connections, floor systems, bracing, etc. The required strength of members or connections in fire is based on externally applied loads (e.g. 1.2D + 0.5L) plus any forces generated by restrained thermal deformations.

At present, engineered structural-fire safety design using an element-level approach is within reach of the structural-fire engineering profession. While the tools needed for system-level performance-based structural-fire engineering are available, practical application of comprehensive system-level approaches are not common.

Element-Level Structural-Fire Design

Traditional prescriptive-based structural-fire safety requirements in building codes are fundamentally an element-level approach. Individual elements are tested in furnaces, and there is either no consideration or highly simplified consideration of how these elements interact with the rest of the structural system. In the U.S., fire resistance ratings are determined by furnace testing conducted in accordance with ASTM E119.⁶ Furnace tests are conducted on individual columns, beams, small portions of floor systems, small portions of walls, etc. Connections, such as beam-to-column connections, are not normally considered in ASTM E119, as is typical of an element-level approach. In regard to the three key steps in predicting structural response to fire for the prescriptive approach, the thermal environment is characterized by the standard gas temperature versus time curve specified in ASTM E119. The other two steps, heat transfer analysis and structural response analysis are accomplished by testing, not by analysis.

When considering an element-level engineered structural-fire safety design, the fire response of individual elements is considered, just as in the prescriptive-based approach. However, in engineered structural-fire safety design, member response predictions are based primarily on engineering calculations rather than on furnace testing. Basing fire response predictions on calculations rather than on furnace testing leads to significant advantages and opportunities. For example, typical furnaces in the U.S. used to test floor beams and composite floor systems have a maximum length of approximately 5 m. However, typical floor spans in actual buildings are more commonly on the order of 10 to 15 m. Thus, it is not possible to test a realistic size floor beam or floor system in a furnace, leading to significant questions on scale effects and boundary conditions. However, an engineered design can consider the actual dimensions of the floor system under consideration. Furnaces are also limited in the types and magnitude of loads that can be applied to a structural element. For example, most fire resistance ratings for steel columns in the U.S. are based on a furnace test where no load is applied to the column. The fire resistance rating is based only on the measured temperature in the column, which is used as a surrogate for structural failure. In an engineered approach, the loading on the column can be included as part of the evaluation process, along with various boundary conditions at the column ends. This leads to a much more realistic evaluation of column response in fire. However, the greatest advantage to using engineering calculations is that it allows the structural engineer the freedom to explore a variety of design alternatives to find the best structural-fire safety design for a given situation.

When approaching an engineered structural-fire safety design using an element-level approach, the three steps noted earlier must be considered in the design, i.e., (1) characterize thermal environment of fire; (2) compute temperatures of structural members; and (3) compute structural response. For an element-level design, each of these steps can be considerably simplified in comparison to what is needed in a system-level design. Further, for an element-level design, there is considerable information and guidance available in building standards. In the case of steel structures, building standards that provide guidance include:

• Eurocode 3 –Design of Steel Structures. Part 1-2: General Actions – Actions on Structures Exposed to Fire.¹²
• AISC Specification for Structural Steel Building. Appendix 4 – Structural Design for Fire Conditions.³

A first step in an element-level structural-fire design is characterizing the thermal environment of the fire. This is most commonly represented in terms of a gas temperature versus time curve. For an element-level design, all that is needed is the temperature exposure of the individual element under consideration rather than the temperature exposure for the entire structural system. In large building fire events, the fire often travels around floors of a building and spreads from floor-to-floor, often with several floors burning at the same time.^18,19 Describing the thermal exposure of such a fire event for a system-level design is complex. However, estimating the exposure for a single structural element can be often handled in a simpler manner.

Options for describing the thermal exposure for a single structural element design include:

• code specified standard fires (ASTM E119, ISO 834, etc.)
• compartment fire models
• parametric or empirical gas temperature-time curves (usually derived from compartment fire model analysis)
• local fire models.

The use of standard fires, such as ASTM E119, for an engineered design has some advantages in that it maintains a closer tie to existing building standards and may therefore be more readily accepted by building authorities. However, adopting some type of compartment fire analysis can lead to a more realistic representation of the fire environment by more explicit consideration of the building under consideration, including fuel loads and ventilation conditions. Compartment fire analysis can be done using available (and usually free) software such as CFAST,²⁵ Ozone,¹⁰ or Branzfire.⁵⁵ Considerable guidance for developing compartment fire models, including recommended fuel loads for various building occupancies, is available in Eurocode 1.¹¹ When an analysis is conducted to provide a more realistic representation of a fire for the conditions (fuel load, ventilation, compartment layout, etc.) in a particular building, the resulting representation of the fire is sometimes referred to as a “natural fire.”

As an example of developing a “design fire” for an element-level analysis, the software OZone¹⁰ was used to conduct a compartment fire analysis. This analysis was conducted for a building compartment that was approximately 6 m × 7 m × 3.5 m in dimensions, with a fuel load representative of an office occupancy per Eurocode 1. The analysis was conducted using a range of possible ventilation conditions, ranging from a high level of available ventilation to a low level of available ventilation. The resulting gas temperature vs time curves are shown in Figure 7. Note that if all other conditions remain the same (fuel load, compartment size, compartment boundary thermal properties), a high level of ventilation will generally result in higher peak gas temperatures but a shorter fire duration (the high level of ventilation results in rapid burning of the available fuel). In contrast, a low level of ventilation will result in lower peak gas temperatures but longer fire duration (the low level of ventilation results in slower burning of the available fuel). Also shown in Figure 7 is the standard ASTM E119 gas temperature-time curve. Note that the standard ISO 834 curve is very similar to ASTM E119.

FIGURE 7. Gas temperature vs time curves

The advantage of the compartment fire analysis is that it recognizes that fires have a limited duration of burning at any one location within a building, and therefore an individual structural element (column, beam, etc.) has a limited exposure time to high temperatures. The actual time of exposure of an individual structural element to high temperatures will depend on factors such as fuel load and ventilation. The standard fire, which is the same for all buildings, does not provide an estimate of actual duration of exposure.

Compartment fire analysis, while providing a more realistic estimate for fire exposure for a single structural element, requires judgment. It can be quite difficult to define compartment sizes, thermal properties of compartment walls, fuel load, and ventilation. However, these issues can often be addressed by running parametric studies where input variables are varied under a reasonable range of possible conditions, thereby bracketing possible gas temperature versus time curves. An example of such a parametric study is presented by Kirk²⁷ in describing the analysis of possible fire scenarios for investigation of the fire-induced collapse of the Faculty of Architecture Building at the Delft University of Technology (see Figure 4). After such a parametric study, a designer can choose several gas temperature-time curves that will be used for evaluating structural member response. The two gas temperature-time curves shown in Figure 7 provide an example of such a design approach.

Once “design fire” gas temperature-time curves are established (standard fire, compartment fire analysis, etc.), the fire response of individual structural members can evaluated. As an example of possible element-level design approaches, consider the problem of determining the required thickness of SFRM (spray applied fire resistive material) for a W12 × 120 column that has an effective length of 3.8 m, and is made of ASTM A992 steel (minimum specified F_y = 345 MPa).

As an example, using prescriptive code requirements, say the building code required a 3-hour fire resistance rating per ASTM E119 for this W12 × 120 column. ASTM E119 permits the fire resistance rating to be determined using a furnace test with no load applied to the column. The acceptance criteria is that in a 3-hour exposure to the standard fire curve (Figure 5), the temperature of the steel column cannot exceed an average of 538°C (1000°F), and cannot exceed 649°C (1200°F) at any one thermocouple location (ASTM E119 specifies locations along the column where temperatures must be measured). Thus, for columns, ASTM E119 is simply a heat transfer test. The implications of the acceptance criteria is that if the column temperature remains below the specified levels, the column will retain sufficient capacity to safely carry loads during a fire.

As noted above, the ASTM E119 furnace test for steel columns is basically a heat transfer test. However, heat transfer can be calculated, and therefore does not necessarily need to be done with a physical test specimen in a furnace. So, the simplest application of an engineered element-level structural-fire design for this example column would be to use calculations to determine the required thickness of SFRM rather than physical testing. Heat transfer calculations for structural-fire applications can be done with a variety of readily available tools. For this example, the program OZone, which was used for the compartment fire analysis shown in Figure 7, was also used to conduct the heat transfer analysis. In conducting this heat transfer analysis, information is needed on the thermal properties of the SFRM, including thermal conductivity, specific heat, and density. OZone has a built in library of SFRM properties that can be used for heat transfer analysis. In this regard, it would be helpful if SFRM manufacturers provided more comprehensive temperature-dependent thermal property data on their products. Nonetheless, using more generic thermal properties of SFRM, such as those provided in OZone, can still lead to useful design solutions.

Figure 8 shows the results of the heat transfer analysis. The figure shows the ASTM E119 gas temperature-time curve, and also plots the steel temperature for the W12 × 120 column as a function of time, for three levels of protection: (1) no protection; (2) 10 mm SFRM; and (3) 30 mm SFRM. As a simple acceptance criterion based on ASTM E119, the column will be considered acceptable if the temperature remains below 500°C. Based on these simple calculations, it can be determined that 30 mm SFRM is needed for a 3-hour fire resistance rating.

FIGURE 8. Heat transfer analysis of W12 × 120 column for standard ASTM E119 fire

Rather than using a standard fire, the heat transfer analysis described above can be repeated for a gas temperature-time curve derived from a compartment fire analysis. Say for example, the 3.8 m long W12 × 120 column is exposed to the gas-temperature-time curve shown in Figure 7 as “Compartment Fire Analysis – High Ventilation.” Results are plotted in Figure 9. This figure shows the compartment fire gas temperature-time curve, and also plots the steel temperature for the W12 × 120 column as a function of time, for two levels of protection: (1) no protection; and (2) 10 mm SFRM. This analysis shows that with 10 mm SFRM, the steel temperature does not exceed 400°C for this fire event.

FIGURE 9. Heat transfer analysis of W12 × 120 column for compartment fire

As a final example of simple engineered element-level structural fire design, the axial compressive capacity of the 3.8 m long W12 × 120 column will be calculated throughout the compartment fire event. The case of the column with 10mm SFRM will be considered. Using the column temperatures plotted in Figure 9, the corresponding column compressive strength for each temperature will be computed. For a given column temperature, the column strength will be calculated using Appendix 4 of the 2016 AISC Specification.³ Results are shown in Figure 10. This figure shows gas temperature, column temperature, and column design compressive strength as a function of time. The elevated temperature column buckling strength equations in AISC Appendix 4 are applicable only for temperatures in excess of 200°C. Consequently, column capacity in Figure 10 is plotted only from temperatures greater than 200°C.

FIGURE 10. Strength analysis of W12 × 120 column for compartment fire

The analysis shows that the lowest compressive design strength of the column occurs at about 60 minutes and is equal to 3650 kN. Say, for example, that the required compressive strength of the column based on a load combination of 1.2D + 0.5L is 3200 kN. Since the design strength of the column exceeds the required strength, the column with 10 mm of SFRM is acceptable.

The examples discussed above involved determining the required fire insulation for a steel column. Similar types of analyses can be conducted for steel and composite beams, and other structural members.

These examples were intended to illustrate that element-level engineered structural-fire safety design is possible with currently available tools combined with reasonable engineering judgment. By replacing furnace testing with engineering calculations, a designer has the ability to explore a wider range of fire safety strategies, and can more realistically evaluate the expected fire performance of real structural elements. For example, conducting a furnace test on a 30m long steel truss would be nearly impossible because of the limited number of furnaces worldwide available of this size (there may be none). However, calculating the fire response of a 30 m long steel truss can be done. When ASTM E119 was first developed nearly 100 years ago, the tools for computing heat transfer and structural response of structural elements subjected to fire were not available, and so furnace testing was a reasonable approach. However, with the availability of good computational tools for fire modeling, heat transfer analysis, and structural analysis, the continued sole reliance on furnace testing is no longer needed or reasonable.

System-Level Structural-Fire Design

A system-level structural-fire safety design represents a more advanced and comprehensive engineered approach than an element-level design. In the system-level approach, the fire modeling, heat transfer analysis, and structural analysis are all done with a more comprehensive representation of the complete building system. The actual movement of a fire through a building is evaluated, and the response of the entire structural system to the fire is computed. In the structural analysis, the connection and interaction of structural elements is considered, so as to predict the response of not just a single element, but the entire structural system. In a system-level analysis, both of the key factors affecting structural response to fire are considered: the effects of thermally induced forces and deformations, and the effects of elevated temperatures on material properties.

An example of what is likely the most advanced system-level structural-fire evaluation conducted to-date is the analysis by NIST of the collapse of the World Trade Center Towers⁴⁵ and of the collapse of WTC Building 7.⁴⁶ For these analyses, detailed three-dimensional models were developed of the building architectural features and furnishings (for the purposes of fire modeling), as well as of the structural system. Detailed simulation of fire movement through the buildings was done using the computational fluid dynamics based software FDS – Fire Dynamics Simulator³⁵) developed at NIST. Heat transfer analysis and structural response analysis were then done using advanced finite element software to predict structural response throughout the fire event and to ultimately predict the onset of collapse. The overall study serves as a model for a system-level structural-fire evaluation.

At present, the type of advanced and detailed system-level analysis conducted by NIST in the WTC studies is likely not practical for most building design projects, although it may be in the future. Nonetheless, considerable work has been done to move in the direction of system-level structural-fire engineering design for real building projects, although considerably more work is still needed. To realize the full advantages of engineered structural-fire safety design, system-level analysis is needed. The remainder of this section will discuss some of the issues involved in system-level structural-fire design.

One of the difficult aspects of system-level structural-fire engineering is representing the thermal environment of the fire on the structural system. In major building fires, the fire moves horizontally and vertically through the building. Architectural trends for new buildings have moved away from well compartmented spaces to much more open systems. Many office buildings have large open floor plans with little or no compartmentation and often with large floor-to-floor openings. Partitions, when present, are often not fire-rated and not full story height. Overall, modern buildings have more varied and complex architectural and structural forms. When considering system-level representations of fires for the purposes of structural-fire response analysis, compartment fire models become highly questionable, and more advanced fire modeling is needed.

For more advanced fire modeling, a tool of choice is the NIST developed software FDS.³⁵ This computational fluid dynamics based software allows a full three-dimensional representation of the development, growth and movement of fire in complex spaces. This software is available for free from NIST, but requires highly knowledgeable individuals for proper use.

An example of the type of information that can be developed with FDS is shown in Figure 11. This figure, reproduced from the NIST WTC study,⁴⁵ shows a snapshot of upper layer gas temperatures on the 94th floor of Building WTC 1; 15 minutes after aircraft impact. Note that the horizontal extent of very high temperatures does not include the entire floor. The analysis also showed that the fire traveled around the floor, and that multiple floors were burning at different locations within each floor as any given instant.

FIGURE 11. FDS prediction of upper layer gas temperatures predicted on the 94th floor of world trade center building 1; 15 min after aircraft impact45

The FDS analysis of the WTC towers showed that the thermal environment of a large building fire is complex, and interpreting this thermal environment in the context of structural response analysis is difficult. A much simpler analysis of fire movement was conducted for the 2008 fire of the Faculty of Architecture Building in Delft, the Netherlands. This fire moved horizontally and vertically throughout the building for about 7-hours before a portion of this reinforced concrete building completely collapsed (Figure 4). Although the fire event lasted about 7-hours, the duration of high temperatures, say above about 400°C, seen by any single structural element was estimated to be about 30 to 40 minutes, based on photographic evidence and simple fire modeling.^18,27,36 For an element-level structural evaluation in this building, the thermal environment was approximated in a fairly simple manner by a single gas temperature-time curve.²⁷ However, for a more comprehensive system-level structural evaluation, characterizing the thermal environment of such a large fire becomes challenging.

One of the major barriers in moving forward with system-level structural-fire engineering is the lack of guidance on design-fires. We need to better understand and characterize fires in buildings with large non-compartmented spaces with horizontal and vertical (floor-to-floor) movement of fire. Some of the questions that need to be answered are:

• What is the horizontal variation of gas temperatures for a fire in a large open space or a fire in a partitioned space (but not fired rated partitions), and how does this affect structural response?
• How do fires move horizontally and how does this affect structural response?
• Are simultaneous fires on adjacent floors important for structural response?
• How can we simplify the representation of large complex moving fires for the purposes of structural system response evaluation?

Answering these questions requires close collaboration between fire modeling specialists and structural-fire engineering specialists. In this regard, it would be valuable to have more detailed investigations and studies of actual large fire events. Progress has been made in studying traveling fires in buildings,^13,29,49,52 although much still needs to be done. In the end, the goal is to develop guidance for design-fires for system-level structural-fire design that align with modern building architecture.

In addition to representing the thermal environment of a fire, system-level structural-fire design also requires system level models for heat transfer and structural analysis. Computing heat transfer from the fire environment to the surface of the structural members and then computing heat transfer within the members can be done with commercially available finite element software such as Abaqus and ANSYS, as well as many others. A major challenge, however, is interfacing the results of fire modeling software like FDS with a finite element program for structural heat transfer analysis. During the WTC investigation, NIST developed such an interface in a preliminary manner,⁴⁵ but much work is still needed in further development and simplification of such an interface.

The last step in a system-level structural-fire design is a system-level structural analysis. This can be attempted with commercial finite element programs like Abaqus, ANSYS and others, or by special purpose structural-fire analysis software such as SAFIR²² and Vulcan. Note that typically the same software can be used for both the heat transfer analysis and structural response analysis. The structural response analysis must include the capabilities for temperature-dependent nonlinear material properties as well as nonlinear geometry. While the finite element packages noted above include these features, developing such advanced nonlinear temperature-dependent finite element models requires considerable expertize and experience.

While the basic tools for advanced system-level structural analysis are available, there are still considerable limitations in our modeling capabilities. More data are needed on temperature-dependent material properties for steel and concrete, both during heating and subsequent cooling. For steel structures, more data and better models are needed for time-dependent material response of structural steel, bolts and welds. While considerable work has been done on time-dependent behavior of steel elements at elevated-temperatures,^{16,38,39,40,41} understanding of this phenomena is still limited. More research is also needed for improved and simplified modeling of connections in both steel and reinforced concrete structures. One very important area where good modeling capability is still lacking is spalling of concrete. However, despite these limitations and need for additional research, currently available finite element programs provide a very powerful tool for system-level structural response analysis.

One area of system-level analysis of particular importance to steel building structures is the development of methods to design floor systems for fire exposure, taking advantage of tensile membrane action and catenary action. An example of this behavior can be seen in Figure 12. This figure shows a photo of the Echelon Building in Austin, Texas after a severe fire in 2010. The floor system in this building consisted of unprotected lightweight steel joists supporting a concrete slab. An element-level analysis, using either prescriptive-based or engineered approaches, would have considered the steel joists as the primary load carrying structural element, and would likely have predicted failure of the unprotected joists for a severe fire event, such as that which occurred in this building. In reality, however, the floor system survived this severe fire without collapse due to the development of tensile membrane action in the concrete floor slab. This example demonstrates how the behavior of a structural system in a fire can be much better than the behavior of an individual element, and how an element-level approach would lead to an overly-conservative design. However, taking advantage of system-level behavior requires system-level analysis.

FIGURE 12. Floor system in echelon building after severe fire in 2010

Considerable work has been done in understanding membrane action in floor systems and translating this understanding to practical design methodologies. Much of this work started with the Cardington fire tests in the U.K. and work by Bailey⁷ followed by a number of other researchers (see for example Ref. [54]). An explanation of floor membrane action in fires and the use of membrane action in design can be found in Wang et al.⁵⁶ These methods can be used to significantly reduce or eliminate fire protection on steel floor beams, while safely maintaining load carrying capacity during severe fire exposures.

Tensile membrane analysis of floor systems is a good example of system-level analysis applied to a portion of a structure. More comprehensive system-level approaches to structural-fire design, although difficult and requiring a great deal of judgment, are being applied to actual building design projects. Some examples are described by Johann et al.,²⁴ Tonicello et al.,⁵³ Lamont et al.,³¹ Lane (2000),⁵⁷ Luo,³³ Plank,⁴⁸ Kohno and Okazaki,³⁰ Engelhardt and Heintz,¹⁷ and Block and Kho.⁸ The application of system-level structural-fire engineering to actual building design projects, however, is still rare, and typically done for very special and unique buildings with very special design consultants. However, these case studies demonstrate the feasibility and advantages for system-level structural-fire design, and provide a model for the future. As the structural-engineering profession advances beyond non-engineered prescriptive designs, it is expected that element-level engineered designs will become more common, with system-level designs remaining within the purview of specialized applications. Nonetheless, increasing use of system-level analysis and design will certainly inform and enhance element-level methodologies.

The descriptions of element-level and system-level structural-fire safety design provided above are intended to illustrate the range of possible approaches to engineered designs. There are, of course, a wider range of possibilities that combine aspects of both element-level and system-level design.

Moving Forward in Structural-Fire Engineering

Engineered structural-fire safety design holds the promise of improved safety, improved economy, and greater design flexibility compared to traditional prescriptive-based approaches. Engineered approaches are particularly attractive for steel buildings, where prescriptive-based requirements can sometimes impose a significant penalty, often unnecessarily so, on economy and architectural versatility. Interestingly, despite extensive and sustained worldwide research efforts in structural-fire engineering for a number of years, the impact of this research on the design of most buildings today is still small. Engineered structural-fire safety design is still restricted to a very small number of special building design projects and practiced by a very small number of specialized design consultants. To achieve the benefits of engineered structural-fire safety design on a more widespread basis will require continued work. In the view of the authors, some of the major remaining challenges include the following:

• Development of a methodology to quantitatively assess collapse risk under fire.

  There has been much discussion and debate on the level of safety provided by different approaches to structural-fire safety design, including traditional non-engineered prescriptive approaches as well as the various options for engineered approaches. At present, however, there is no objective way to compare different design approaches in regard to collapse risk. Until such a methodology is developed and accepted by the design and regulatory communities, progress will likely continue to be slow. Within the field of earthquake engineering, the “FEMA P695” methodology²¹ has provided a framework for assessing the collapse risk of different seismic force resisting systems in buildings, as well as defining an acceptable collapse risk. A similar effort is needed in the structural-fire engineering field for a quantitative assessment of collapse risk, with appropriate consideration of uncertainties involved in characterizing the fire hazard in a building, heat transfer analysis, and structural response analysis.

• Identify limits of applicability of different structural-fire safety design approaches.

  Traditional non-engineered prescriptive approaches for structural-fire safety have a good safety track record, as few major buildings, with some notable exceptions, have collapsed due to fire. The prescriptive approach likely provides adequate safety for many ordinary structures that fall within the long experience base of the prescriptive rules. However, building architecture, materials, and structural design are continually innovating, resulting in buildings that are increasingly outside of our range of experience with prescriptive structural-fire safety design. For such buildings, relying solely on experience becomes increasingly problematic. For example, prescriptive structural-fire safety design is likely acceptable from a safety perspective for an ordinary 3-story office building. However, it seems reasonable that some level of engineered structural-fire safety design should be undertaken for a 110-story building with an occupancy of 20 000 people, although non-engineered prescriptive approaches are still normally used for such major structures. At present, there is little understanding of the circumstances where the application of prescriptive requirements may be problematic from a safety point of view. Work is needed to develop guidelines that connect the required level of engineering analysis and design for structural-fire safety with building characteristics and consequence of failure. Guidance is needed on when prescriptive approaches are adequate and when engineered approaches are preferred. Further, in the case of engineered approaches, guidance is needed on the level of analysis complexity appropriate under different circumstances. Developing such guidance would certainly benefit from a methodology to quantitatively assess collapse risk under fire, as described above. However, even in the absence of such a methodology, much progress can be made through consensus of knowledgeable stakeholders.

• Develop better guidance for element-level design.

  Element-level engineered structural-fire safety design has the potential for practical application in many building projects with currently available tools and currently available building standards. However, more guidance is needed on when thermally induced forces can be neglected in such designs, and in cases where they cannot be safely neglected, how these thermally induced forces can be handled in a simplified and practical manner. Additionally, guidance is needed on the appropriate approaches to connection design at the element-level.

• Develop better guidance for “design fires” for structural-fire safety design.

  There has been extensive work in characterizing compartment fires and local fires. However, as noted earlier, architectural trends for buildings have moved away from well compartmented spaces to much more open systems. Many office buildings have large open floor plans with little or no compartmentation and often with large floor-to-floor openings. For such building layouts, compartment fire analysis is often unrealistic. While progress has been made in developing guidance for characterizing fire in large building spaces, much additional work is needed to develop guidance for design-fires that align with modern building architecture and that appropriately consider the many uncertainties associated with characterizing fires in buildings.

• Continue and expand basic and applied research in structural-fire safety.

  These has been a significant amount of research in structural-fire safety over the last several decades, although the amount is very small compared to research in earthquake safety. Nonetheless, there is a critical need to continue and to expand research in structural-fire engineering for the field to move forward. A number of workshops have been held in the U.S. in recent years to discuss research needs in structural-fire engineering.^28,44,47 These workshops identified a number of basic research needs related to fire modeling, heat transfer analysis, and charactering elevated temperature structural response at the material, component, subassembly, and system levels. There are also important needs in applied research, to transfer the findings from basic research into practical design methodologies. There are still many research needs that remain unaddressed. In this regard, there has not a strong history of structural-fire engineering research in the U.S. Many of the basic developments in structural-fie engineering resulted from research in the U.K., Europe, Australia, New Zealand, and other countries. This changed after the events of 9/11when the U.S. National Science Foundation (NSF) started significant funding of structural-fire engineering research at U.S. universities. This enabled a number of structural engineering faculty to develop active research programs in structural-fire engineering and to add courses on structural-fire engineering to their graduate curricula. This NSF support began to address important research and education needs in the field. Perhaps even more importantly, the NSF university research support developed a community of young scholars in the U.S. with PhD’s in structural fire engineering. These individuals then went on to other universities to continue research and teaching in structural-fire engineering, or entered industry to practice structural-fire engineering, or went on to other research organizations such as NIST. Further, many of these PhD’s in structural fire engineering became involved in the development of codes and standards. Over a period of only a few years, the impact of NSF support, both in the research outcomes and in the development of highly trained individuals with a passion for the field, was remarkable. The NSF support of structural-fire engineering served as a major change agent to advance structural-fire engineering in the U.S. Unfortunately, within the last few years, NSF has stopped funding structural-fire engineering research, and in fact, prohibits submission of proposals in this area. This change in NSF policy serves as a detriment to the advancement of the structural-fire engineering in the U.S.

• Recognize that structural-fire safety must be the responsibility of structural engineers.

  Like wind and earthquake, fire can impose large demands on structures, with the potential for extensive structural damage and collapse. However, the difference is that in the current state of practice, structures are engineered for wind and earthquake loads by structural engineers. However, in current practice in the U.S. and in most other countries, structural engineers play little or no role in assuring structural safety in the event of a major fire. Structural-fire safety, with the basic goal of preventing structural collapse, is clearly a structural engineering undertaking. Professionally and ethically, structural-fire safety must be the responsibility of structural engineers.

Conclusions

This paper has provided a brief overview of approaches to structural-fire safety design, including traditional non-engineered prescriptive approaches and engineered approaches. In the U.S. and in many other countries, the basis for structural-fire safety design is the traditional non-engineered prescriptive approach; an approach that does not require the involvement of structural engineers on building design projects and an approach that has seen little meaningful change in over a hundred years. Nonetheless, there are important reasons to advance the state-of-practice in structural-fire safety by moving in the direction of engineered structural-fire safety design that is undertaken by structural engineers on building design projects. These reasons include improved safety, improved economy, and improved design flexibility. While the widespread application of engineered structural-fire safety design in practice still seems somewhat distant in the future, a concerted effort by the research, design, and building regulatory communities could have great impact in a reasonable time horizon. The structural engineering profession has made great strides in many areas, most notably in earthquake engineering. Similar great strides are need in structural-fire safety.

ACKNOWLEDGMENTS

The U.S. National Science Foundation is gratefully acknowledged for support of the authors’ research in structural-fire engineering through Awards Nos. 0700682, 0927819, 1031099, and CMS-0521086. Any opinions, findings, and conclusions or recommendations expressed in this paper are those of the authors and do not necessarily reflect the views of the National Science Foundation.

DISCLOSURE

The authors have no conflict of interest.