Skip to content
Upcoming courses

Steel in fire: Difference between revisions

Tools
From Fire Safety Academy Wiki
← Older editNewer edit →
VisualWikitext
Sanne (talk | contribs)
Bots
337 edits
Add diagrams fse-steel-in-fire-v1.svg, fse-steel-in-fire-v2.svg, fse-steel-in-fire-v3.svg (level-1 visual plan)
Emma (talk | contribs)
Bots
294 edits
Revise Basic article against FSE writing guidelines: sourced explanations, examples and references (GPT-6)
Line 1: Line 1:
'''Steel in fire''' describes how structural steel members heat, expand and lose strength and stiffness during fire exposure. Although steel is noncombustible, its mechanical properties and structural performance can deteriorate substantially. The outcome depends on fire exposure, member shape, protection, loading, connections and end restraint.<ref name="Milke-4-209">Milke (2002), p. 4-209.</ref><ref name="Fleischmann-Buchanan-4-242">Fleischmann and Buchanan (2002), p. 4-242.</ref>
'''Steel in fire''' describes the heating, expansion and loss of mechanical performance of steel members exposed to fire. Steel can remain noncombustible while its strength and stiffness decrease. Structural performance therefore depends on temperature, loading, geometry, protection and support conditions, rather than combustibility alone.<ref name="KH_1-157">Kodur and Harmathy (2002), p. 1-157.</ref><ref name="M_4-209">Milke (2002), p. 4-209.</ref><ref name="M_4-211_4-212">Milke (2002), pp. 4-211–4-212.</ref>


Steel conducts heat efficiently, so exposed regions can warm quickly and transfer energy to adjacent parts and connections. Protection can delay this heating, while member geometry and loading determine how the rising temperature affects stress, buckling and deformation. Structural response must therefore be evaluated rather than inferred from noncombustibility alone.<ref name="Milke-4-209" /><ref name="Rockett-Milke-1-28">Rockett and Milke (2002), p. 1-28.</ref>
Assessment connects three stages: defining the fire exposure, calculating heat transfer into the steel and evaluating structural response. The gas temperature is an input to this sequence, not automatically the temperature of the member. Exposed surface, steel mass and protection determine how rapidly the member heats, while mechanical properties determine the consequences.<ref name="M_4-209" /><ref name="M_4-222_4-224">Milke (2002), pp. 4-222–4-224.</ref>


Assessment follows three linked stages: characterise the fire exposure, calculate heat transfer into the member and evaluate structural response at elevated temperatures. Each stage has its own data and uncertainty.<ref name="Milke-4-209" /><ref name="Cadorin-Franssen-430">Cadorin et al. (2003), p. 430.</ref>
A standard furnace exposure provides one basis for comparing protected and unprotected assemblies. A natural fire requires its own description of the thermal environment. Neither approach supplies one failure temperature for every steel product or structural system. The selected endpoint must remain connected to the load and conditions it represents.<ref name="VR_8.6">Van Rossum (2027), §8.6.</ref><ref name="KH_1-163">Kodur and Harmathy (2002), p. 1-163.</ref>


== Heating and section factor ==
== From fire exposure to steel temperature ==


Steel temperature results from the balance between incident heat, thermal storage and losses. Exposed surface area receives radiation and convection, while conduction redistributes energy through the section and connected members. The actual fire and boundary conditions must be defined before a meaningful steel temperature can be calculated.<ref name="Milke-4-209" /><ref name="Rockett-Milke-1-29">Rockett and Milke (2002), p. 1-29.</ref>
The fire environment supplies heat to the surface of an exposed member. Radiation transfers energy from flames and hot surroundings, while convection transfers energy between the gases and the surface. Conduction then redistributes that energy within the steel. The member temperature follows this combined heat balance rather than simply equalling the surrounding gas temperature.<ref name="M_4-209" /><ref name="M_4-222">Milke (2002), p. 4-222.</ref>


[[Section factor]] relates the heated perimeter or exposed area of a steel section to its volume or mass. A slender, highly exposed section can respond differently from a compact section with more steel mass. Protection changes the exposed boundary and should be represented in the heat-transfer model.<ref name="Milke-4-219">Milke (2002), pp. 4-219–4-221.</ref><ref name="Cadorin-Franssen-430" />
The exposure history must therefore be defined before a heating calculation can begin. A furnace curve supplies one repeatable history, while a natural compartment fire develops according to the fuel, ventilation and enclosure. Heat release rate and gas temperature are different quantities within that description. Van Rossum treats standard temperature curves as prescribed exposures, rather than predictions of every room fire.<ref name="VR_8.6" />
[[File:fse-steel-in-fire-v1.svg|thumb|upright=1.5|alt=Two steel sections with different exposed perimeters and masses receiving the same nominal heat flux.|Section geometry affects steel heating. Based on Milke (2002), pp. 4-219–4-221.]]


Insulation reduces heat transfer to the steel and delays the time to a critical condition. Its thickness, conductivity, continuity and attachment determine performance. Joints, penetrations and damaged regions can create local thermal bridges that an average model would miss.<ref name="Milke-4-209" /><ref name="White-4-259">White (2002), pp. 4-259–4-260.</ref>
The exposed faces also matter. A column surrounded by fire receives a different boundary condition from a beam connected to a concrete slab. The slab can shield the upper part of the beam and take heat from it. Milke explains that these effects can create appreciable temperature differences between the lower and upper flanges.<ref name="M_4-223_4-224">Milke (2002), pp. 4-223–4-224.</ref>


The heat-exposure history matters as well as the peak fire temperature. A short severe exposure can produce different internal gradients from a longer moderate one. Conductive storage means steel can continue heating after the fire environment begins to cool.<ref name="Rockett-Milke-1-29" /><ref name="Lie-4-201">Lie (2002), pp. 4-201–4-203.</ref>
A thermal model can treat the member as uniformly hot or calculate a temperature distribution. The first approach is simpler and can suit some exposed sections. The second becomes relevant when temperature gradients affect the structural response. Choosing between them depends on the exposure, section and purpose of the calculation.<ref name="M_4-222_4-224" />


== Strength and critical temperature ==
== Geometry and section factor ==


[[Strength reduction at elevated temperatures]] affects yield strength, tensile strength and modulus of elasticity. Fleischmann and Buchanan show different strength-temperature curves for hot-rolled, cold-drawn prestressing and high-strength alloy steels. The steel type and manufacturing history therefore matter when selecting properties.<ref name="Fleischmann-Buchanan-4-241">Fleischmann and Buchanan (2002), p. 4-241.</ref>
[[Section factor]] relates the heated surface to the amount of steel being heated. For a member with a constant cross-section, the heated perimeter divided by the cross-sectional area expresses this relationship. More heated perimeter per unit area generally gives a greater heating response under otherwise equal conditions. The comparison concerns exposure and thermal storage together.<ref name="M_4-222_4-224" />


Reduced stiffness can increase deflection even before strength becomes inadequate. Structural analysis should evaluate both stress and deformation at elevated temperature, using properties appropriate to the expected range. Ambient-temperature values alone can be misleading.<ref name="Milke-4-209" /><ref name="Fleischmann-Buchanan-4-242" />
Milke also uses the inverse form: mass per unit length divided by heated perimeter. Its SI unit is kilograms per square metre. A large value represents more steel mass behind each unit of heated surface and therefore greater thermal storage. Confusing this ratio with the perimeter-to-area section factor reverses the interpretation of which section heats faster.<ref name="M_4-219">Milke (2002), p. 4-219.</ref><ref name="M_4-222_4-224" />


[[Critical steel temperature]] is a selected temperature at which structural performance becomes unacceptable under the stated loading and criteria. It is not a universal physical constant. Different loads, steel types, end conditions and failure definitions can justify different critical temperatures.<ref name="Kodur-Harmathy-1-163">Kodur and Harmathy (2002), p. 1-163.</ref><ref name="Milke-4-209" />
[[File:fse-steel-in-fire-v1.svg|thumb|upright=1.5|alt=Two steel sections with different exposed perimeters and masses receiving the same nominal heat flux.|Section geometry affects steel heating. Based on Milke (2002), pp. 4-222–4-224.]]


North American practice described by Kodur and Harmathy uses 538 °C for structural steel in some ASTM E119-related calculations, typically for columns under full design load. Reinforcing and prestressing steel have different values in their review, including 593 °C and 426 °C respectively. These examples must not replace a project-specific structural assessment.<ref name="Kodur-Harmathy-1-163" />
Protection can alter the relevant heated boundary. An insulating enclosure around a section need not follow every recess of the steel profile. Its thermal resistance and exposed perimeter form part of the protected-member model. The geometry used for an unprotected contour cannot automatically describe an enclosing board system.<ref name="M_4-213_4-214">Milke (2002), pp. 4-213–4-214.</ref><ref name="M_4-223">Milke (2002), p. 4-223.</ref>
[[File:fse-steel-in-fire-v2.svg|thumb|upright=1.5|alt=Schematic strength curves for hot-rolled, prestressing and alloy steel that fall at different temperatures and cross.|Steel products respond differently to elevated temperature. Based on Fleischmann and Buchanan (2002), p. 4-241.]]


Milke similarly notes that temperature endpoints are often based on ASTM E119, but their inaccuracies depend on temperature-dependent properties and heating descriptions. A calculation can use a selected critical temperature, yet still require structural analysis of the actual member and loading.<ref name="Milke-4-209" />
Shielding by adjacent construction likewise changes the surface receiving heat. Milke provides separate perimeter diagrams for different arrangements and connections to slabs. These diagrams distinguish steel area from the boundary exposed to the thermal environment. The resulting section factor describes a particular exposure configuration, rather than a permanent property of the section designation.<ref name="M_4-223" />


== Thermal expansion and restraint ==
The two geometry ratios can be converted through density. Milke's steel density of 7,860 kg/m³ divided by the example mass-to-perimeter ratio of 39.3 kg/m² gives 200 m⁻¹. This is the corresponding heated-perimeter-to-area ratio for that example. The conversion leaves the physical section unchanged while expressing the same thermal-storage relationship in another form.<ref name="M_4-211">Milke (2002), p. 4-211.</ref><ref name="M_4-219" /><ref name="M_4-224">Milke (2002), p. 4-224.</ref>


Steel and surrounding materials expand when heated. [[Restrained thermal expansion]] occurs when connections or adjacent construction prevent free deformation, producing additional forces and changing buckling behaviour. Restraint can therefore alter both the member's stress state and its stability.<ref name="Fleischmann-Buchanan-4-242" /><ref name="Milke-4-210">Milke (2002), pp. 4-209–4-210.</ref>
The conversion also provides a dimensional check. A perimeter divided by an area has units of inverse metres, whereas mass divided by heated area has units of kilograms per square metre. Values expressed in these units cannot be substituted directly for each other. The intervening density and the direction of the ratio determine their relationship.<ref name="M_4-222_4-224" />


ASTM E119 distinguishes restrained and unrestrained construction systems according to geometry, connections and framing. Welded, riveted or bolted steel framing is listed as restrained in the method's examples, while other systems are classified differently. These categories belong to the test arrangement rather than to a generic label for every structure.<ref name="Milke-4-211">Milke (2002), p. 4-211.</ref>
== A simplified heating calculation ==


Thermal expansion is not always detrimental. Fleischmann and Buchanan note that it can improve fire resistance in some concrete and composite arrangements by changing the structural response. Its effect depends on the system and cannot be assessed from expansion alone.<ref name="Fleischmann-Buchanan-4-242" />
In a lumped-capacity model, the steel is assumed to have one temperature throughout its section. The heat entering over a short interval increases the energy stored in the steel. Dividing the incoming heat by the steel's heat capacity gives the temperature increment. Milke expresses the model using the mass-to-heated-perimeter ratio.<ref name="M_4-222" />


Connections and adjacent members also conduct heat and provide restraint. A member-by-member temperature calculation may therefore be insufficient if the whole frame governs stability. The structural model should reflect the actual load paths and boundary conditions.<ref name="Milke-4-209" /><ref name="Cadorin-Franssen-430" />
<math display="block">\Delta T_s=\frac{h_{\mathrm{tot}}}{c_s(m'/P)}(T_g-T_s)\Delta t</math>


== Intumescent and passive protection ==
{| class="wikitable"
! Symbol !! Description !! Unit
|-
| <math>\Delta T_s</math> || Steel temperature increment || K
|-
| <math>h_{\mathrm{tot}}</math> || Combined convection and linearised radiation coefficient || W/(m²·K)
|-
| <math>c_s</math> || Steel specific heat || J/(kg·K)
|-
| <math>m'</math> || Steel mass per unit member length || kg/m
|-
| <math>P</math> || Heated perimeter || m
|-
| <math>T_g,T_s</math> || Gas and steel temperatures || °C
|-
| <math>\Delta t</math> || Calculation interval || s
|}


[[Intumescent material]] forms an insulating layer during heating, reducing heat transfer to the steel. Its effectiveness depends on expansion, continuity, adhesion and thermal properties. Damage, deformation or incomplete coverage can compromise that protection.<ref name="Beyler-Hirschler-1-110">Beyler and Hirschler (2002), p. 1-110.</ref><ref name="Milke-4-209" />
The temperature difference in this equation has the same numerical value in kelvins and degrees Celsius. Radiation calculations used to determine the combined coefficient, however, require absolute temperatures. Milke separates the radiative and convective contributions and recommends short calculation intervals for this approximation. The equation is then applied repeatedly as the exposure and steel temperature change.<ref name="M_4-222_4-223">Milke (2002), pp. 4-222–4-223.</ref>


Boards, blankets, concrete encasement and other passive systems provide alternative protection. They delay heating and can preserve structural capacity for longer, but their installation and attachment are part of performance. A rating should be tied to the complete protected assembly.<ref name="Milke-4-209" /><ref name="White-4-259" />
A calculation of the convective contribution illustrates the role of the inputs. Milke gives a mass-to-perimeter ratio of 39.3 kg/m² for the contour of a W8 × 28 section. His A36 specific-heat expression gives 430 J/(kg·K) at 20 °C. Taking his example temperature level of 538 °C as an assumed gas temperature permits a simple initial-step comparison.<ref name="M_4-219" /><ref name="M_4-211" /><ref name="KH_1-163" />
[[File:fse-steel-in-fire-v3.svg|thumb|upright=1.5|alt=Critical-temperature examples 538, 593 and 426 °C for three steel uses.|Critical-temperature examples depend on steel use and method. Based on Kodur and Harmathy (2002), p. 1-163.]]


Protective systems can also affect structural behaviour beyond temperature. Added weight, changed stiffness and altered connections can influence response, while moisture and ageing can change the protection itself. These effects belong in assessment where relevant.<ref name="Kodur-Harmathy-1-156">Kodur and Harmathy (2002), p. 1-156.</ref><ref name="White-4-260">White (2002), p. 4-260.</ref>
With a convective coefficient of 25 W/(m²·K) and an interval of 10 s, the calculated convection-only rise is 7.7 K. The arithmetic uses the unrounded specific heat of 430.2 J/(kg·K). Radiation has deliberately been excluded from this component calculation, so 7.7 K is not a complete fire-heating prediction. The chosen gas temperature is an illustrative boundary condition, not a measured exposure for that section.<ref name="M_4-211" /><ref name="M_4-222_4-223" />


Passive protection does not remove the need to evaluate the unprotected details. Connections, penetrations and partially exposed members can govern heating and stability. A system-level review should include those local exposures.<ref name="Milke-4-209" /><ref name="Cadorin-Franssen-430" />
== Strength, stiffness and steel type ==


== Structural fire engineering ==
[[Strength reduction at elevated temperatures]] changes the load that a hot member can resist. Yield strength describes the onset of substantial permanent deformation in the adopted material model. The modulus of elasticity describes stiffness before that stage. A reduction in stiffness can increase deformation and affect stability even before the nominal yield-strength limit is reached.<ref name="M_4-211_4-212" />


[[Structural fire engineering]] links fire dynamics, heat transfer and structural mechanics. The fire model supplies gas temperatures and heat flux, the thermal model supplies member temperatures, and the structural model evaluates stress, deflection and stability. Errors in any stage affect the final conclusion.<ref name="Cadorin-Franssen-430" /><ref name="Milke-4-209" />
The response is not identical for all steel products. Fleischmann and Buchanan distinguish hot-rolled reinforcing steel, cold-drawn prestressing steel and high-strength alloy steel. Their chapter presents different strength-temperature relationships for these materials. It also distinguishes yield strength from tensile strength where the relevant design property differs.<ref name="FB_4-240">Fleischmann and Buchanan (2002), p. 4-240.</ref>


OZone illustrates this sequence for simple steel elements. It combines compartment-fire models with localised-fire and steel-temperature calculations, then evaluates fire resistance using Eurocode 3 methods. The authors emphasise the value of connecting fire and structural specialisms rather than using one generic critical temperature for every case.<ref name="Cadorin-Franssen-430" />
[[File:fse-steel-in-fire-v2.svg|thumb|upright=1.5|alt=Schematic strength curves for hot-rolled, prestressing and alloy steel that fall at different temperatures and cross.|Steel products respond differently to elevated temperature. Based on Fleischmann and Buchanan (2002), p. 4-240.]]


Empirical correlations can be useful within their tested exposure, loading and span. Outside those conditions, heat-transfer and structural analyses are needed. Testing still validates the models and the interactions between steel, insulation and other components.<ref name="Milke-4-209" /><ref name="Milke-4-210">Milke (2002), pp. 4-209–4-210.</ref>
Heating can also introduce creep, which is deformation that develops with time under sustained stress. Milke explains that creep depends on stress, heating rate and temperature. Some fire-design material relationships incorporate its effects implicitly. Combining data from incompatible test histories can therefore misrepresent the response even when the nominal steel temperature is the same.<ref name="M_4-212">Milke (2002), p. 4-212.</ref>


Model uncertainty includes the fire curve, convective and radiative boundary conditions, material properties and end restraints. Sensitivity analysis should include these rather than only varying steel temperature. A precise thermal result does not guarantee an accurate structural prediction.<ref name="Milke-4-209" /><ref name="VR-1.7">Van Rossum (2027), §1.7.</ref>
Room-temperature strength consequently supplies only part of the required information. A thermal analysis determines where and when the steel heats. Temperature-dependent mechanical properties then connect those temperatures to the structural calculation. The assessment still requires the applied load and the relevant support conditions.<ref name="M_4-209" /><ref name="M_4-211_4-212" />


== Fire resistance and performance criteria ==
== Critical temperature and structural endpoints ==


Steel fire resistance concerns maintaining structural adequacy for the required time. Deflection, strain, stress and stability are compared with established limits, and loading and restraint must match the application. A steel temperature is only an input to that evaluation.<ref name="Milke-4-209" /><ref name="Milke-4-210" />
[[Critical steel temperature]] is a temperature associated with an unacceptable structural condition under stated assumptions. It depends on the load to be supported and the material's remaining capacity. The concept can simplify a thermal calculation when those structural conditions have already been established. It does not replace them with a universal failure temperature.<ref name="KH_1-163" /><ref name="M_4-209" />


ASTM E119 provides one standard exposure and endpoint framework, including loading and restraint variations. Its furnace dimensions and setup can make results comparative rather than exact reproductions of full structures. Any extension to actual construction should be documented.<ref name="Milke-4-210" />
Kodur and Harmathy describe historical North American examples of 538 °C for structural steel, 593 °C for reinforcing steel and 426 °C for prestressing steel. The structural-steel example is associated with particular ASTM E119-related assumptions, including full design loading in the discussion of columns. These figures illustrate method-dependent endpoints. They do not represent melting temperatures or limits applicable to every steel member.<ref name="KH_1-163" />


A noncombustible member can still fail through buckling, excessive deflection or connection failure. Noncombustibility therefore cannot be used as evidence of fire resistance. The complete assembly and its load paths matter.<ref name="Kodur-Harmathy-1-155">Kodur and Harmathy (2002), p. 1-155.</ref><ref name="Milke-4-210" />
[[File:fse-steel-in-fire-v3.svg|thumb|upright=1.5|alt=Critical-temperature examples 538, 593 and 426 °C for three steel uses.|Critical-temperature examples depend on steel use and method. Based on Kodur and Harmathy (2002), p. 1-163.]]
 
Post-fire assessment may also need the cooling history. Steel and surrounding materials can remain hot after flames decline, and mechanical properties and connections may have changed. Residual condition should not be inferred from appearance alone.<ref name="Rockett-Milke-1-29" /><ref name="Fleischmann-Buchanan-4-242" />
 
== Limits and common misunderstandings ==
 
Steel does not fail at one universal temperature. Critical temperature depends on the member, steel type, loading, restraint and performance criterion. Values such as 538 °C belong to specific methods and should not be applied indiscriminately.<ref name="Kodur-Harmathy-1-163" /><ref name="Milke-4-209" />
 
A thicker or stronger member at ambient temperature is not automatically better in every fire. Section geometry, exposed area, protection and load paths affect heating and stability. Ambient and fire conditions must be evaluated together.<ref name="Milke-4-219" /><ref name="Cadorin-Franssen-430" />
 
An intumescent coating does not guarantee protection for every assembly. Adhesion, continuity and exposure determine performance, and local defects can dominate. Full assembly evidence is needed for critical applications.<ref name="Beyler-Hirschler-1-110" /><ref name="White-4-260">White (2002), p. 4-260.</ref>
 
Finally, steel's noncombustibility does not mean fire has no effect. Heat can change strength, stiffness, expansion and stability without any combustion of the steel itself. These are the central issues of steel in fire.<ref name="Milke-4-209" /><ref name="Fleischmann-Buchanan-4-242" />


== Assessment and review ==
Milke's description of A36 steel provides the mechanical context. At approximately 538 °C, its yield strength in the cited relationship is about 60% of the room-temperature value. The significance of that reduction depends on the stress already imposed on the member. A lightly loaded and a heavily loaded member need not reach their structural endpoints at the same temperature.<ref name="M_4-212" />


A steel-temperature calculation should identify the heated perimeter, exposed faces and any insulation or shielding. Those features determine how much energy enters the section and how quickly its temperature rises. Using only the section's ambient dimensions would omit the actual thermal boundary.<ref name="Milke-4-219" /><ref name="Cadorin-Franssen-430" />
A temperature endpoint and a structural endpoint therefore answer related questions. The thermal calculation can determine when a specified temperature is reached. A structural calculation evaluates the resulting stresses, deformation and capacity under the adopted loading and restraint. Where a temperature criterion is used, its connection to those structural conditions remains part of the method.<ref name="M_4-209" /><ref name="M_4-222" />


Temperature should also be evaluated through the section rather than only at the surface. Conduction redistributes energy and produces gradients, especially during changing exposure. A surface or average temperature can be useful, but it does not automatically describe every critical point or connection.<ref name="Rockett-Milke-1-29" /><ref name="Milke-4-209" />
== Expansion, restraint and structural response ==


Property selection should match the steel product. Hot-rolled, cold-drawn prestressing and high-strength alloy steels can have different strength and stiffness curves. Using a generic steel curve for all of them would misrepresent their elevated-temperature response.<ref name="Fleischmann-Buchanan-4-241" /><ref name="Fleischmann-Buchanan-4-242" />
Steel expands as its temperature rises. If supports or surrounding construction limit that movement, [[Restrained thermal expansion]] changes the forces and deformations within the system. The member then responds to more than its original mechanical loading. Milke consequently includes thermal expansion and end conditions in the structural assessment.<ref name="M_4-209_4-211">Milke (2002), pp. 4-209–4-211.</ref>


Critical-temperature criteria should be tied to the actual failure mode. A strength-based limit, a deflection limit and a connection or buckling limit can produce different acceptable temperatures. A single value cannot cover all of these without a structural justification.<ref name="Kodur-Harmathy-1-163" /><ref name="Milke-4-209" />
Test restraint is a defined condition rather than an informal description of a connection. Milke reproduces historical ASTM examples distinguishing different floor and framing arrangements. The classification considers how the surrounding structure resists movement. A restrained test result therefore requires interpretation in relation to the proposed construction and its actual boundaries.<ref name="M_4-210_4-211">Milke (2002), pp. 4-210–4-211.</ref>


Restraint should be evaluated at the level of the whole framing system. Connections and adjacent members can restrict expansion and introduce new forces, but they also provide alternative load paths. A member-only model can miss both the additional restraint and the redistribution that follows local weakening.<ref name="Fleischmann-Buchanan-4-242" /><ref name="Cadorin-Franssen-430" />
Temperature gradients add another complication. A beam beneath a slab can have a hotter lower flange and a cooler upper flange. A uniform-temperature model does not explicitly describe that difference. Where the gradient affects the structural question, Milke identifies the need for a more suitable analytical approach.<ref name="M_4-223_4-224" />


Protection review should include continuity and local details. Joints, penetrations, fasteners and damaged coatings can create thermal bridges even when most of the member is protected. Those details can govern the time to a critical condition and should be represented in the model.<ref name="Milke-4-209" /><ref name="White-4-260" />
[[Structural fire engineering]] combines these effects rather than treating steel temperature as a complete outcome. It uses the exposure to calculate temperatures, then applies temperature-dependent properties within a structural model. Loads, geometry and end conditions determine how the member responds. The required result may involve deformation and stability as well as a nominal strength comparison.<ref name="M_4-209_4-210">Milke (2002), pp. 4-209–4-210.</ref>


Model comparison should use matching exposure and endpoint definitions. A furnace curve, a natural compartment fire and a localised exposure can produce different steel temperatures, and a critical-temperature criterion does not capture every structural failure. Validation therefore requires the complete thermal and structural chain.<ref name="Lie-4-201" /><ref name="Cadorin-Franssen-430" />
== Protection systems and their behaviour ==


A final sensitivity review should vary fire exposure, section factor, protection, material properties and restraint. This reveals whether the result is controlled by one uncertain assumption and whether additional full-scale evidence is needed. It also prevents a precise steel temperature from being mistaken for a complete fire-resistance conclusion.<ref name="VR-1.7" /><ref name="Milke-4-209" />
Protection reduces the rate of heat transfer into the steel. Milke describes insulating boards, spray-applied materials, concrete encasement and enclosing membranes. Their mechanisms and attachment details differ. The relevant evidence concerns the protected assembly, including the protection material and the way it remains attached during exposure.<ref name="M_4-212_4-214">Milke (2002), pp. 4-212–4-214.</ref>


The assessment should also distinguish thermal and mechanical limits. A section can reach a selected critical temperature without immediate collapse, or fail earlier through buckling, connection damage or restraint forces. The acceptable condition must be defined structurally rather than inferred from temperature alone.<ref name="Kodur-Harmathy-1-163" /><ref name="Milke-4-209" />
[[Intumescent material]] expands during heating to form an insulating layer. The change is part of the protective mechanism, so the layer cannot always be represented by its original thickness and properties. Van Rossum identifies swelling as one of several processes that change the heat-conduction path during fire. Milke includes intumescent products among spray-applied protection methods.<ref name="VR_5.2">Van Rossum (2027), §5.2.</ref><ref name="M_4-214">Milke (2002), p. 4-214.</ref>


For protected members, the protection's own condition should be evaluated. Intumescent layers and membranes can crack, detach or change conductivity, altering the thermal path during exposure. Those changes can create earlier local heating than an intact-layer model predicts.<ref name="Beyler-Hirschler-1-110" /><ref name="White-4-260" />
Board systems rely on suitable enclosure and attachment. Milke specifically identifies the means of securing boards as a factor affecting assembly performance. Spray-applied products require properties such as adhesion, durability and uniform application. These considerations connect the material's thermal performance to its condition on the actual member.<ref name="M_4-212_4-214" />


For structural systems, load redistribution can preserve some capacity after local weakening, but it can also increase demand on connections and adjacent members. Progressive and tensile mechanisms therefore require system-level analysis rather than a simple critical-temperature check for one section.<ref name="Fleischmann-Buchanan-4-242" /><ref name="Cadorin-Franssen-430" />
Protection can also store heat. Milke distinguishes simplified models that neglect its heat capacity from models that include it. Gypsum and concrete protection are examples where that contribution can matter. Omitting the protection's thermal storage is therefore an assumption to assess, rather than a property of every protected-steel calculation.<ref name="M_4-223" />


Post-fire assessment should use the cooling and damage history as well as the original member data. Connections, coatings and material properties may have changed, and residual heat can persist. Appearance and noncombustibility alone cannot establish residual structural capacity.<ref name="Rockett-Milke-1-29" /><ref name="Fleischmann-Buchanan-4-242" />
A protected member can receive heat through the insulation while the insulation itself is warming. Part of the incoming energy then increases the protection's temperature instead of immediately heating the steel. Milke includes this storage in the denominator of a protected-member temperature-increment expression. The effect differs from thermal resistance, which describes the rate of energy transfer through the layer.<ref name="M_4-223" />


A complete report therefore links the fire exposure, thermal boundary, temperature-dependent properties and structural criteria. It identifies where those inputs are uncertain and whether additional assembly evidence is needed. This keeps steel in fire as a structural assessment rather than reducing it to a material-temperature rule.<ref name="Milke-4-209" /><ref name="VR-1.7" />
An intact thermal barrier also needs to survive handling and its surrounding environment. Milke includes bond strength, durability, weathering resistance and consistent application among relevant protection characteristics. These requirements concern maintaining the protective arrangement that the thermal model assumes. A conductivity value alone cannot establish whether the applied protection will retain that arrangement throughout service and exposure.<ref name="M_4-212" />


The exposure should also include the regions above, below and beside the member. A hot upper layer can radiate downwards, while a localised flame can produce a very different boundary condition on one face. Treating every side as uniformly exposed would miss those directional effects.<ref name="VR-5.4">Van Rossum (2027), §5.4.</ref><ref name="Cadorin-Franssen-430" />
== Evidence and modelling limits ==


A final check is whether the selected model represents the protection and restraint as they actually exist. Idealised insulation, simple supports and a generic fire curve can make the calculation easier, but they can also change the controlling failure mode. Those simplifications should be tested before the steel temperature is used for a decision.<ref name="Milke-4-209" /><ref name="Milke-4-210">Milke (2002), pp. 4-209–4-210.</ref>
Empirical protection formulas are linked to the tests from which they were developed. Milke cautions that their ranges can be restricted by member shapes, protection thickness and other conditions. A result outside those ranges has a different evidential basis from a result within them. The fire exposure, loading and span also limit the applicability of test-based correlations.<ref name="M_4-209" /><ref name="M_4-219" />


The report should also state whether the steel temperature was measured, calculated or adopted as a criterion. Those are different kinds of evidence, and each needs its own justification. A criterion can be useful without claiming that the member actually reached it in a test or in service.<ref name="Milke-4-209" /><ref name="Kodur-Harmathy-1-163" />
A more detailed thermal calculation does not remove uncertainty in its inputs. Heating conditions, temperature-dependent material properties and insulation behaviour still affect the result. A structural calculation introduces additional assumptions about loads and restraint. The accuracy of the combined prediction depends on the relationship between all these stages.<ref name="M_4-209_4-210" />


For final acceptance, the complete load and thermal history should be compared with the member's required function. Steel's noncombustibility does not answer that comparison, and a critical-temperature value alone does not complete it. The result must be expressed in terms of structural performance and exposure time.<ref name="Milke-4-209" /><ref name="Milke-4-210">Milke (2002), pp. 4-209–4-210.</ref>
Tests provide evidence about interactions that simplified models may omit. Milke emphasises their role in establishing material behaviour and checking protected assemblies. The resulting evidence describes the tested exposure and configuration. Its application to another member depends on whether the model represents the changes in geometry, heating and structural conditions.<ref name="M_4-210_4-211" />


== See also ==
== See also ==
Line 120: Line 121:
* [[Progressive collapse]]
* [[Progressive collapse]]
* [[Tensile membrane action]]
* [[Tensile membrane action]]
* [[Strength reduction at elevated temperatures]]
* [[Fire resistance]]


== Notes ==
== Notes ==
Line 128: Line 129:
== References ==
== References ==


* Beyler, C. L., & Hirschler, M. M. (2002). Thermal decomposition of polymers. In P. J. DiNenno (Ed.), ''SFPE handbook of fire protection engineering'' (3rd ed.). National Fire Protection Association.
* Fleischmann, C., & Buchanan, A. (2002). Analytical methods for determining fire resistance of concrete members. In P. J. DiNenno (Ed.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 4-239–4-256). National Fire Protection Association.
* Cadorin, J.-F., Pintea, D., Dotreppe, J.-C., & Franssen, J.-M. (2003). A tool to design steel elements submitted to compartment fires: OZone V2. Part 2: Methodology and application. ''Fire Safety Journal, 38'', 429–451.
* Kodur, V. K. R., & Harmathy, T. Z. (2002). Properties of building materials. In P. J. DiNenno (Ed.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 1-155–1-181). National Fire Protection Association.
* Fleischmann, C., & Buchanan, A. (2002). Analytical methods for determining fire resistance of concrete members. In P. J. DiNenno (Ed.), ''SFPE handbook of fire protection engineering'' (3rd ed.). National Fire Protection Association.
* Milke, J. A. (2002). Analytical methods for determining fire resistance of steel members. In P. J. DiNenno (Ed.), ''SFPE handbook of fire protection engineering'' (3rd ed., pp. 4-209–4-238). National Fire Protection Association.
* Kodur, V. K. R., & Harmathy, T. Z. (2002). Properties of building materials. In P. J. DiNenno (Ed.), ''SFPE handbook of fire protection engineering'' (3rd ed.). National Fire Protection Association.
* Lie, T. T. (2002). Fire temperature-time relations. In P. J. DiNenno (Ed.), ''SFPE handbook of fire protection engineering'' (3rd ed.). National Fire Protection Association.
* Milke, J. A. (2002). Analytical methods for determining fire resistance of steel members. In P. J. DiNenno (Ed.), ''SFPE handbook of fire protection engineering'' (3rd ed.). National Fire Protection Association.
* Rockett, J. A., & Milke, J. A. (2002). Conduction of heat in solids. In P. J. DiNenno (Ed.), ''SFPE handbook of fire protection engineering'' (3rd ed.). National Fire Protection Association.
* Van Rossum, E. (2027). ''Introduction to Fire Safety Engineering''. Fire Safety Academy.
* Van Rossum, E. (2027). ''Introduction to Fire Safety Engineering''. Fire Safety Academy.
* White, R. H. (2002). Analytical methods for determining fire resistance of timber members. In P. J. DiNenno (Ed.), ''SFPE handbook of fire protection engineering'' (3rd ed.). National Fire Protection Association.


[[Category:Structural Fire Engineering]]
[[Category:Structural Fire Engineering]]

Revision as of 18:54, 2 November 2026

Steel in fire describes the heating, expansion and loss of mechanical performance of steel members exposed to fire. Steel can remain noncombustible while its strength and stiffness decrease. Structural performance therefore depends on temperature, loading, geometry, protection and support conditions, rather than combustibility alone.[1][2][3]

Assessment connects three stages: defining the fire exposure, calculating heat transfer into the steel and evaluating structural response. The gas temperature is an input to this sequence, not automatically the temperature of the member. Exposed surface, steel mass and protection determine how rapidly the member heats, while mechanical properties determine the consequences.[2][4]

A standard furnace exposure provides one basis for comparing protected and unprotected assemblies. A natural fire requires its own description of the thermal environment. Neither approach supplies one failure temperature for every steel product or structural system. The selected endpoint must remain connected to the load and conditions it represents.[5][6]

From fire exposure to steel temperature

The fire environment supplies heat to the surface of an exposed member. Radiation transfers energy from flames and hot surroundings, while convection transfers energy between the gases and the surface. Conduction then redistributes that energy within the steel. The member temperature follows this combined heat balance rather than simply equalling the surrounding gas temperature.[2][7]

The exposure history must therefore be defined before a heating calculation can begin. A furnace curve supplies one repeatable history, while a natural compartment fire develops according to the fuel, ventilation and enclosure. Heat release rate and gas temperature are different quantities within that description. Van Rossum treats standard temperature curves as prescribed exposures, rather than predictions of every room fire.[5]

The exposed faces also matter. A column surrounded by fire receives a different boundary condition from a beam connected to a concrete slab. The slab can shield the upper part of the beam and take heat from it. Milke explains that these effects can create appreciable temperature differences between the lower and upper flanges.[8]

A thermal model can treat the member as uniformly hot or calculate a temperature distribution. The first approach is simpler and can suit some exposed sections. The second becomes relevant when temperature gradients affect the structural response. Choosing between them depends on the exposure, section and purpose of the calculation.[4]

Geometry and section factor

Section factor relates the heated surface to the amount of steel being heated. For a member with a constant cross-section, the heated perimeter divided by the cross-sectional area expresses this relationship. More heated perimeter per unit area generally gives a greater heating response under otherwise equal conditions. The comparison concerns exposure and thermal storage together.[4]

Milke also uses the inverse form: mass per unit length divided by heated perimeter. Its SI unit is kilograms per square metre. A large value represents more steel mass behind each unit of heated surface and therefore greater thermal storage. Confusing this ratio with the perimeter-to-area section factor reverses the interpretation of which section heats faster.[9][4]

Two steel sections with different exposed perimeters and masses receiving the same nominal heat flux.
Section geometry affects steel heating. Based on Milke (2002), pp. 4-222–4-224.

Protection can alter the relevant heated boundary. An insulating enclosure around a section need not follow every recess of the steel profile. Its thermal resistance and exposed perimeter form part of the protected-member model. The geometry used for an unprotected contour cannot automatically describe an enclosing board system.[10][11]

Shielding by adjacent construction likewise changes the surface receiving heat. Milke provides separate perimeter diagrams for different arrangements and connections to slabs. These diagrams distinguish steel area from the boundary exposed to the thermal environment. The resulting section factor describes a particular exposure configuration, rather than a permanent property of the section designation.[11]

The two geometry ratios can be converted through density. Milke's steel density of 7,860 kg/m³ divided by the example mass-to-perimeter ratio of 39.3 kg/m² gives 200 m⁻¹. This is the corresponding heated-perimeter-to-area ratio for that example. The conversion leaves the physical section unchanged while expressing the same thermal-storage relationship in another form.[12][9][13]

The conversion also provides a dimensional check. A perimeter divided by an area has units of inverse metres, whereas mass divided by heated area has units of kilograms per square metre. Values expressed in these units cannot be substituted directly for each other. The intervening density and the direction of the ratio determine their relationship.[4]

A simplified heating calculation

In a lumped-capacity model, the steel is assumed to have one temperature throughout its section. The heat entering over a short interval increases the energy stored in the steel. Dividing the incoming heat by the steel's heat capacity gives the temperature increment. Milke expresses the model using the mass-to-heated-perimeter ratio.[7]

ΔTs=htotcs(m′/P)(Tg−Ts)Δt

Symbol Description Unit
ΔTs Steel temperature increment K
htot Combined convection and linearised radiation coefficient W/(m²·K)
cs Steel specific heat J/(kg·K)
m′ Steel mass per unit member length kg/m
P Heated perimeter m
Tg,Ts Gas and steel temperatures °C
Δt Calculation interval s

The temperature difference in this equation has the same numerical value in kelvins and degrees Celsius. Radiation calculations used to determine the combined coefficient, however, require absolute temperatures. Milke separates the radiative and convective contributions and recommends short calculation intervals for this approximation. The equation is then applied repeatedly as the exposure and steel temperature change.[14]

A calculation of the convective contribution illustrates the role of the inputs. Milke gives a mass-to-perimeter ratio of 39.3 kg/m² for the contour of a W8 × 28 section. His A36 specific-heat expression gives 430 J/(kg·K) at 20 °C. Taking his example temperature level of 538 °C as an assumed gas temperature permits a simple initial-step comparison.[9][12][6]

With a convective coefficient of 25 W/(m²·K) and an interval of 10 s, the calculated convection-only rise is 7.7 K. The arithmetic uses the unrounded specific heat of 430.2 J/(kg·K). Radiation has deliberately been excluded from this component calculation, so 7.7 K is not a complete fire-heating prediction. The chosen gas temperature is an illustrative boundary condition, not a measured exposure for that section.[12][14]

Strength, stiffness and steel type

Strength reduction at elevated temperatures changes the load that a hot member can resist. Yield strength describes the onset of substantial permanent deformation in the adopted material model. The modulus of elasticity describes stiffness before that stage. A reduction in stiffness can increase deformation and affect stability even before the nominal yield-strength limit is reached.[3]

The response is not identical for all steel products. Fleischmann and Buchanan distinguish hot-rolled reinforcing steel, cold-drawn prestressing steel and high-strength alloy steel. Their chapter presents different strength-temperature relationships for these materials. It also distinguishes yield strength from tensile strength where the relevant design property differs.[15]

Schematic strength curves for hot-rolled, prestressing and alloy steel that fall at different temperatures and cross.
Steel products respond differently to elevated temperature. Based on Fleischmann and Buchanan (2002), p. 4-240.

Heating can also introduce creep, which is deformation that develops with time under sustained stress. Milke explains that creep depends on stress, heating rate and temperature. Some fire-design material relationships incorporate its effects implicitly. Combining data from incompatible test histories can therefore misrepresent the response even when the nominal steel temperature is the same.[16]

Room-temperature strength consequently supplies only part of the required information. A thermal analysis determines where and when the steel heats. Temperature-dependent mechanical properties then connect those temperatures to the structural calculation. The assessment still requires the applied load and the relevant support conditions.[2][3]

Critical temperature and structural endpoints

Critical steel temperature is a temperature associated with an unacceptable structural condition under stated assumptions. It depends on the load to be supported and the material's remaining capacity. The concept can simplify a thermal calculation when those structural conditions have already been established. It does not replace them with a universal failure temperature.[6][2]

Kodur and Harmathy describe historical North American examples of 538 °C for structural steel, 593 °C for reinforcing steel and 426 °C for prestressing steel. The structural-steel example is associated with particular ASTM E119-related assumptions, including full design loading in the discussion of columns. These figures illustrate method-dependent endpoints. They do not represent melting temperatures or limits applicable to every steel member.[6]

Critical-temperature examples 538, 593 and 426 °C for three steel uses.
Critical-temperature examples depend on steel use and method. Based on Kodur and Harmathy (2002), p. 1-163.

Milke's description of A36 steel provides the mechanical context. At approximately 538 °C, its yield strength in the cited relationship is about 60% of the room-temperature value. The significance of that reduction depends on the stress already imposed on the member. A lightly loaded and a heavily loaded member need not reach their structural endpoints at the same temperature.[16]

A temperature endpoint and a structural endpoint therefore answer related questions. The thermal calculation can determine when a specified temperature is reached. A structural calculation evaluates the resulting stresses, deformation and capacity under the adopted loading and restraint. Where a temperature criterion is used, its connection to those structural conditions remains part of the method.[2][7]

Expansion, restraint and structural response

Steel expands as its temperature rises. If supports or surrounding construction limit that movement, Restrained thermal expansion changes the forces and deformations within the system. The member then responds to more than its original mechanical loading. Milke consequently includes thermal expansion and end conditions in the structural assessment.[17]

Test restraint is a defined condition rather than an informal description of a connection. Milke reproduces historical ASTM examples distinguishing different floor and framing arrangements. The classification considers how the surrounding structure resists movement. A restrained test result therefore requires interpretation in relation to the proposed construction and its actual boundaries.[18]

Temperature gradients add another complication. A beam beneath a slab can have a hotter lower flange and a cooler upper flange. A uniform-temperature model does not explicitly describe that difference. Where the gradient affects the structural question, Milke identifies the need for a more suitable analytical approach.[8]

Structural fire engineering combines these effects rather than treating steel temperature as a complete outcome. It uses the exposure to calculate temperatures, then applies temperature-dependent properties within a structural model. Loads, geometry and end conditions determine how the member responds. The required result may involve deformation and stability as well as a nominal strength comparison.[19]

Protection systems and their behaviour

Protection reduces the rate of heat transfer into the steel. Milke describes insulating boards, spray-applied materials, concrete encasement and enclosing membranes. Their mechanisms and attachment details differ. The relevant evidence concerns the protected assembly, including the protection material and the way it remains attached during exposure.[20]

Intumescent material expands during heating to form an insulating layer. The change is part of the protective mechanism, so the layer cannot always be represented by its original thickness and properties. Van Rossum identifies swelling as one of several processes that change the heat-conduction path during fire. Milke includes intumescent products among spray-applied protection methods.[21][22]

Board systems rely on suitable enclosure and attachment. Milke specifically identifies the means of securing boards as a factor affecting assembly performance. Spray-applied products require properties such as adhesion, durability and uniform application. These considerations connect the material's thermal performance to its condition on the actual member.[20]

Protection can also store heat. Milke distinguishes simplified models that neglect its heat capacity from models that include it. Gypsum and concrete protection are examples where that contribution can matter. Omitting the protection's thermal storage is therefore an assumption to assess, rather than a property of every protected-steel calculation.[11]

A protected member can receive heat through the insulation while the insulation itself is warming. Part of the incoming energy then increases the protection's temperature instead of immediately heating the steel. Milke includes this storage in the denominator of a protected-member temperature-increment expression. The effect differs from thermal resistance, which describes the rate of energy transfer through the layer.[11]

An intact thermal barrier also needs to survive handling and its surrounding environment. Milke includes bond strength, durability, weathering resistance and consistent application among relevant protection characteristics. These requirements concern maintaining the protective arrangement that the thermal model assumes. A conductivity value alone cannot establish whether the applied protection will retain that arrangement throughout service and exposure.[16]

Evidence and modelling limits

Empirical protection formulas are linked to the tests from which they were developed. Milke cautions that their ranges can be restricted by member shapes, protection thickness and other conditions. A result outside those ranges has a different evidential basis from a result within them. The fire exposure, loading and span also limit the applicability of test-based correlations.[2][9]

A more detailed thermal calculation does not remove uncertainty in its inputs. Heating conditions, temperature-dependent material properties and insulation behaviour still affect the result. A structural calculation introduces additional assumptions about loads and restraint. The accuracy of the combined prediction depends on the relationship between all these stages.[19]

Tests provide evidence about interactions that simplified models may omit. Milke emphasises their role in establishing material behaviour and checking protected assemblies. The resulting evidence describes the tested exposure and configuration. Its application to another member depends on whether the model represents the changes in geometry, heating and structural conditions.[18]

See also

Notes

  1. ↑ Kodur and Harmathy (2002), p. 1-157.
  2. ↑ 2.0 2.1 2.2 2.3 2.4 2.5 2.6 Milke (2002), p. 4-209.
  3. ↑ 3.0 3.1 3.2 Milke (2002), pp. 4-211–4-212.
  4. ↑ 4.0 4.1 4.2 4.3 4.4 Milke (2002), pp. 4-222–4-224.
  5. ↑ 5.0 5.1 Van Rossum (2027), §8.6.
  6. ↑ 6.0 6.1 6.2 6.3 Kodur and Harmathy (2002), p. 1-163.
  7. ↑ 7.0 7.1 7.2 Milke (2002), p. 4-222.
  8. ↑ 8.0 8.1 Milke (2002), pp. 4-223–4-224.
  9. ↑ 9.0 9.1 9.2 9.3 Milke (2002), p. 4-219.
  10. ↑ Milke (2002), pp. 4-213–4-214.
  11. ↑ 11.0 11.1 11.2 11.3 Milke (2002), p. 4-223.
  12. ↑ 12.0 12.1 12.2 Milke (2002), p. 4-211.
  13. ↑ Milke (2002), p. 4-224.
  14. ↑ 14.0 14.1 Milke (2002), pp. 4-222–4-223.
  15. ↑ Fleischmann and Buchanan (2002), p. 4-240.
  16. ↑ 16.0 16.1 16.2 Milke (2002), p. 4-212.
  17. ↑ Milke (2002), pp. 4-209–4-211.
  18. ↑ 18.0 18.1 Milke (2002), pp. 4-210–4-211.
  19. ↑ 19.0 19.1 Milke (2002), pp. 4-209–4-210.
  20. ↑ 20.0 20.1 Milke (2002), pp. 4-212–4-214.
  21. ↑ Van Rossum (2027), §5.2.
  22. ↑ Milke (2002), p. 4-214.

References

  • Fleischmann, C., & Buchanan, A. (2002). Analytical methods for determining fire resistance of concrete members. In P. J. DiNenno (Ed.), SFPE handbook of fire protection engineering (3rd ed., pp. 4-239–4-256). National Fire Protection Association.
  • Kodur, V. K. R., & Harmathy, T. Z. (2002). Properties of building materials. In P. J. DiNenno (Ed.), SFPE handbook of fire protection engineering (3rd ed., pp. 1-155–1-181). National Fire Protection Association.
  • Milke, J. A. (2002). Analytical methods for determining fire resistance of steel members. In P. J. DiNenno (Ed.), SFPE handbook of fire protection engineering (3rd ed., pp. 4-209–4-238). National Fire Protection Association.
  • Van Rossum, E. (2027). Introduction to Fire Safety Engineering. Fire Safety Academy.
Retrieved from "https://fire-safety.academy/w/index.php?title=Steel_in_fire&oldid=2699"