- Email: [email protected]
Horizontal forces in steel structures tested in fire
Journal of Constructional Steel Research 65 (2009) 1896–1903
Contents lists available at ScienceDirect
Journal of Constructional Steel Research journal homepage: www.elsevier.com/locate/jcsr
Horizontal forces in steel structures tested in fire František Wald a,∗ , Zdeněk Sokol a , David Moore b a
Czech Technical University in Prague, Prague, Czech Republic
b
British Constructional Steelwork Association, London, United Kingdom
article
info
Article history: Received 13 November 2007 Accepted 21 April 2009 Keywords: Structural engineering Steel structures Fire test Full-scale test Fire design Eurocodes Connections Robustness Horizontal forces
abstract Fire tests carried out on the eight-storey steel framed building at the Building Research Establishment’s Cardington laboratory have shown that the connections are subject to large axial force. These forces are the result of thermal movements of the structure during heating and cooling and in some cases can result in failure of the structure. This paper describes a fire test carried out on the steel frame at Cardington on 16th January 2003 and a fire test carried out on a structure in Ostrava on 16th June 2006. In both cases the tests were designed to measure the forces generated in the connections. © 2009 Elsevier Ltd. All rights reserved.
1. Introduction Full-scale fire tests carried out on the eight-storey steel framed building at the Building Research Establishment’s Cardington facility have shown that the connections are subject to high axial forces as a result of thermal movements during heating and cooling, see [1]. Current design methods do not check the behaviour of the connections under these conditions which can be critical, particularly in the cooling phase of the fire. Determining these forces is not easy as they are dependent on the form of the structure, the boundary conditions, the time–temperature curve, the thermal movements and local failures during heating and the behaviour of the deformed structure during the cooling phase of the fire. Observations from the fire tests at Cardington also showed that a partially protected composite flooring system deforms to such an extent that it supports the applied load in a combination of bending and catenary actions transferring significant axial forces to the supporting connections. Although the axial capacity of a connection is not routinely checked under the fire situation the axial capacity of a connection is checked as a way of providing a structure with adequate robustness against disproportionate collapse in the event of an
∗ Corresponding address: Department of Steel Structures, Czech Technical University in Prague, Thákurova 7, 166 29 Prague 6, Czech Republic. Tel.: +420 224 354 757; fax: +420 233 337 466. E-mail addresses: [email protected] (F. Wald), [email protected] (Z. Sokol), [email protected] (D. Moore). 0143-974X/$ – see front matter © 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.jcsr.2009.04.020
accidental action such as a gas explosion or impact from a vehicle, see [2]. Early work on robustness required the connections in a steel framed structure to have at least two M 16 bolts in tension to ensure structural integrity., This was replaced by a estimation of the tie forces in the connections based on catenary behaviour of the supported beams. see [3–5]. There are many similarities between the catenary action of a flooring system under fire and that under accidental actions. It is postulated that the design of the connections for axial tying capacity will allow the connections to resist the axial forces generated as a result of the floor going into catenary under fire. If this could be proven no further checks at elevated temperature would be required. The aim of this paper is to determine the axial forces that connections are subject to under heating and cooling and to compare these with the tying forces used in design for progressive collapse. Annex A of EN 1991-1-7: 2006 [6], gives the following expression for calculating tie forces: Ti = min[k(gk + ψ qk )sL; 75 kN]
(1)
where k is the transformation factor; for internal ties k = 0.8; for perimeter ties k = 0.4, gk is the characteristic value of permanent action, ψ is the combination factor according to the accidental load combination, qk is the characteristic value of variable action, s is the spacing of the ties, and L is the span of the tie.
F. Wald et al. / Journal of Constructional Steel Research 65 (2009) 1896–1903
1897
Nomenclature fub fy gk k kE ,θ ky,θ s qk As L E Ft ,b Ft ,2 Ft ,3 Ft ,4 Ti Wy Wz
γM2 ψ εθ
is the ultimate strength of the bolt is the yield stress at ambient temperature is the characteristic value of permanent action is the transformation factor; k = 0.8 for internal ties; k = 0.4 for perimeter ties is the reduction factor for the slope of the linear elastic range at the steel temperature is the reduction factor for the yield stress at the temperature of the steel is the spacing of the ties is the characteristic value of the variable action is the tensile stress area of the bolt is the span of the tie is the elastic modulus of the steel is the design tension resistance of the bolt is the horizontal force to the column on the second floor is the horizontal force to the column on the third floor is the horizontal force to the column on the fourth floor is the tie force is the section modulus to y-axis, major axis is the section modulus to z-axis, minor axis is the partial safety factor for the bolt is the combination factor according to the accidental load combination is the strain at elevated temperature
2. Seventh large-scale fire test on a steel frame in Cardington A structural integrity fire test was carried out on the eightstorey steel framed building at Cardington on 16 January 2003, see [1]. The main purpose of this test was to collect data on the behaviour of typical beam-to-column and beam-to-beam connections subjected to a natural fire, see [7]. The test was carried out in a compartment on the fourth floor enclosing a plan area of 11 × 7 m, as shown in Fig. 1. The internal walls of the compartment were made of three layers of plasterboard (15 mm + 12.5 mm + 15 mm) with a thermal conductivity of 0.19–0.24 W m−1 K−1 . The external wall was a 0.9 m brick window sill and 1.95 m plasterboard wall. A 1.27 m high and 5.7 m wide opening simulated an open window and created ventilation for the compartment. The size of the opening was designed to produce a fire with a temperature exceeding 1200 ◦ C and a duration of 60 min. The steel structure within the compartment consisted of four columns (internal columns were 305 × 305 × 198UC section, edge columns 305 × 305 × 137UC section, steel grade S355), two primary beams (336 × 171 × 51UB section, steel grade S350), two secondary beams (305 × 165 × 40UB section, steel grade S275) and edge beam (356 × 171 × 51UB section), see [1]. Flexible end plates were used for beam-to-column connections and fin plates for the beamto-beam connections. In both cases, the plates were made from steel grade S275 and M20 bolts, grade 8.8, were used. To prevent the collapse of the structure, the columns were fire protected by a 20 mm thick layer of Cafco300 vermiculite-cement spray with a thermal conductivity of 0.078 W m−1 K−1 . In addition, protection was also applied to joints on external columns and parts of the primary beams (approximately at a distance 1.0 m from the joints), see Fig. 2. A lightweight concrete slab cast on the profiled metal decking was supported on the primary and secondary beams. 19 mm diameter shear studs were used on all the beams.
Fig. 1. The position of the fire compartment of the seventh large-scale fire test on the plan of the Cardington frame.
The applied load was applied using sandbags distributed over an area of 18 m by 10.5 m on the floor above the fire compartment. This load represented the permanent action (including floor layers and partition walls) and 56% of the variable action. Wooden cribs with a moisture content of 14% provided a fire load of 40 kg/m2 . According to analytical and finite element simulations, failure of the concrete slab was expected during the fire test. Thermocouples, strain gauges and displacement transducers were used to measure various data during the test. A total of 133 thermocouples monitored the temperature of the connections and beams within the compartment, the temperature of the concrete slab and the gas temperature within the compartment. Additional 14 thermocouples were used to measure the temperature of the columns, see Fig. 3. Nine high-temperature strain gauges were used to measure the strains in the unprotected fin-plate and end-plate–minor-axis joints. A total of 47 ambient-temperature strain gauges were attached to the protected columns and to the concrete slab. See Fig. 4 for the location of the strain gauges. The vertical deformations of the concrete slab were measured by 25 displacement transducers installed on the fifth floor. Additional 12 transducers measured the horizontal movement of the columns and the slab. Ten video cameras recorded the fire and smoke development and the deformations, and two thermo-imaging cameras were used for measuring the temperature distribution in the steel elements. 3. Horizontal forces measured during the Cardington test High-temperature strain gauges were attached to the beams next to the connections to measure the strains induced by the fire. These strains were used to determine the axial forces in the connections during the fire. The high-temperature strain gauges are capable of measuring strains up to a temperature of 1200 ◦ C. The stress σθ at the elevated temperature was derived from the measured strain using modulus of elasticity reduced for the steel temperature (Ea,θ = kE ,θ E). The corresponding steel temperature was recorded by the thermocouple attached to the structure near the strain gauges. The calculations are given in Tables 1 and 2. The stress is limited by yielding of the steel. The resulting stress is calculated as
σθ = min(kE ,θ E ε; ky,θ fy )
(2)
where kE ,θ E
ε
ky,θ fy
θa
is the reduction factor for the slope of the linear elastic range at the corresponding steel temperature, see [8] is the elastic modulus of steel is the measured strain is the reduction factor for the yield stress at the corresponding steel temperature, see [8] is the yield stress at ambient temperature, 396 MPa based on coupon tests of beams for Cardington frame [7] is the steel temperature measured near the strain gauge.
1898
F. Wald et al. / Journal of Constructional Steel Research 65 (2009) 1896–1903
Fig. 2. (a) The fire load in the compartment; (b) the fire load around column D2; (c) the installation of the strain gauges on the column above the fire compartment; (d) and (e) the installation of the strain gauges on the column in the fire compartment; (f) the fire protection of the internal column E2 after the test; (g) the fire protection of the external column E1 after the test on the Cardington frame. Table 1 Measured strains on beam D1.5–E1.5 close to fin plate connection using high-temperature strain gauges. Time (min)
0 5 10 15 20 25 30 45 60 90 118 148 178 Max. Min.
C 444
HT 001
Temp. (◦ C)
Strain (µε × 10−3 )
Eθ (MPa)
18 24 42 66 114 212 331 636 810 702 476 299 205 834 18
0.000 0.051 0.198 0.248 −2.468 −2.554 −3.598 −6.425 0.294 0.400 0.393 0.351 0.317 0.458 −8.066
210 000 210 000 210 000 210 000 207 104 186 459 161 468 51 552 18 413 27 135 131 193 168 317 187 872 210 000 17 278
C 445
HT 003
Stress (MPa)
Temp. (◦ C)
Strain (µε × 10−3 )
0 11 42 52 −388 −335 −298 −120 5 11 52 59 59 68 −394
18 25 41 71 126 242 362 669 888 694 444 252 168 893 18
0.000
−0.068 −0.217 0.325 8.507 8.552 9.519 9.408 10.751 11.841 9.418 8.678 8.396 11.841 −0.217
The results are presented in Fig. 5a for the secondary beam D1.5–E1.5 connected to the primary beam (fin plate connection, located at the mid-span of the primary beam) and in Fig. 5b for the secondary beam connected to the columns (end plate connection). Due to a nonlinear stress–strain relationship Eq. (2) gives increased values at stains larger than the proportional limit. This deference, which reached up to 10% was neglected. The stress development shows the behaviour of the composite beam during the fire, see [9]. Local buckling of the lower flange in
C 446
HT 005
Eθ (MPa)
Stress (MPa)
Temp. (◦ C)
Strain (µε × 10−3 )
Eθ (MPa)
Stress (MPa)
210 000 210 000 210 000 210 000 204 522 180 229 155 035 39 165 14 722 29 453 137 821 178 062 195 659 210 000 14 486
0 14 46 −68 −391 −372 −359 −106 −25 −89 −320 −370 −384 46 −396
18 23 40 71 123 236 369 692 899 717 474 289 195 908 18
−0.046 −0.138 −0.084
210 000 210 000 210 000 210 000 205 204 181 518 153 551 30 464 14 218 25 895 131 525 170 299 190 085 210 000 13 783
0 10 29 18 −383 −320 −255 −51 −19 −50 −150 20 148 268 −388
2.574 2.419 2.496 2.269 2.827 3.232 3.261 1.139 −0.012 −0.077 −1.646 0.000
compression was observed during the test. There are two principal reasons for high compression forces in the flange:
• The beam is restrained from expanding longitudinally, which results in the development of axial compressive forces in the beam; • The rotational stiffness of the joint, neglected in most design models, allows the development of bending moments at the supports, which results in compression of the lower flange. The
F. Wald et al. / Journal of Constructional Steel Research 65 (2009) 1896–1903
1899
Table 2 Measured strains on beam D2–E2 close to end plate connection using high-temperature strain gauges. Time (min)
0 5 10 15 20 25 30 45 60 90 118 148 178 Max. Min.
C 462
HT 009
Temp. (◦ C)
Strain (µε × 10−3 )
Eθ (MPa)
18 21 135 63 114 182 281 545 774 726 474 295 204 847 18
0.000 −0.017 −0.102 −0.292 −0.756 −1.936 −1.483 −1,313 −0.787 −0.158 Failed Failed Failed 0.000 −1.936
210 000 210 000 210 000 210 000 207 003 192 736 171 922 98 624 21 081 25 129 – – – 210 000 16 696
Stress (MPa) 0
−4 −21 −61 −157 −342 −255 −123 −17 −4 – – – 0
−342
C 461
HT 011
HT 021
Temp. (◦ C)
Strain (µε × 10−3 )
Eθ (MPa)
18 20 33 61 116 187 273 492 706 726 513 334 235 781 18
0.000 −0.055 −0.333 −1.196 −1.718 Failed Failed Failed Failed Failed Failed Failed Failed 0 −7.855
210 000 210 000 210 000 210 000 204 895 – – – – – – – – 210 000 18 897
Stress (MPa) 0
−12 −70 −251 −352 – – – – – – – – 0
−354
Temp. (◦ C)
Strain (µε × 10−3 )
Eθ (MPa)
17 19 35 70 124 201 284 569 791 713 500 333 253 800 17
0.000 −0.041 −0.339 −1.058 −1.451 −1.988 −2.395 −1.197 −1.061 1.734 Failed Failed Failed 1.743 −2.398
210 000 210 000 210 000 210 000 204 895 188 732 171 403 84 043 19 625 26 183 – – – 210 000 18 897
Stress (MPa) 0
−9 −71 −222 −297 −331 −294 −99 −21 39 – – – 41
−339
Fig. 5a. Stress in the secondary beam at the beam-to-beam connection E1.5.
Fig. 3. Comparison of the measured temperatures along the external column D1 to the gas, beam and internal column D2 temperatures during the Cardington test.
bending moment at the connection may develop after the lower flange comes into contact with the column. This occurs after significant rotation in the connection. • The stress in the lower flange increased rapidly to the yield stress, which is reached in the 17th min of the test.
The column fire protection limited the temperature rise in the columns and allowed the strains to be measurement using ambient-temperature strain gauges. The measurements were taken during the first 60 min of the fire on the third floor and during the whole experiment on the fourth floor, see Fig. 3. Selected results of strain measurements are presented in Fig. 6a (stresses in the edge columns on the fourth floor located 500 mm above the floor) and in Fig. 6b (stresses in the edge columns on the fourth floor located 500 mm below the ceiling). The stresses in
Fig. 4. Location (a) of the ambient-temperature strain gauges on columns and (b) the high-temperature strain gauges and the thermocouples on beams close to the connections on the Cardington frame.
1900
F. Wald et al. / Journal of Constructional Steel Research 65 (2009) 1896–1903
Fig. 7. The bending moments in the columns measured during the fire test.
Fig. 5b. Stress in the secondary beam at the beam-to-column connections D2 and E2.
Fig. 8. The horizontal forces calculated from the measured bending moments using a beam theory.
4. Fire test on a steel frame in Ostrava Fig. 6a. The stress in the external columns on the fourth floor, 500 mm above the floor.
Fig. 6b. The stress in the external columns on the fourth floor, 500 mm below the ceiling.
the columns were used to calculate the bending moments, see Fig. 7. The shear force in the columns was derived from the bending moment distribution in the columns and, finally, the horizontal forces in the beam-to-column connection were calculated. This simple calculation is based on the continuous beam model, see Fig. 8. The maximal calculated horizontal forces were Ft ,3 = −344 kN (tension), and −65 kN (compression) on the third floor, and Ft ,4 = −462 kN (compression) and +88 kN (tension) on the fourth floor. A global structural FE analysis of the frame was used to verify these values. The nonlinear procedure utilised 2D plastic beam elements and a simplified model of the composite slab. The model included nonlinear temperature-dependent material properties, nonlinear response of the composite beam-to-column joints, thermal expansion, large strain and large deformations. The results of the FE simulation confirmed the beam model approach see [10,11].
On 16 May 2006 a fire test was carried out on a threestorey administrative building which was attached to a singlestorey framed building of the Ammoniac Separator II in the Mittal Steel Plant in Ostrava, Czech Republic, see Fig. 9 [9]. The loadbearing structure, which was completed in 1992, consisted of steel columns and beams supporting a 130 mm (total thickness including ribs) concrete slabs. No shear connection was provided between the steel beams and the concrete slab. The beam-to-beam and beam-to-column connections were designed as simple end plate connections using two or six M20 bolts, see Fig. 10. The fire compartment was 3.80 × 5.95 m with a height of 2.78 m and was built on the second floor. The front wall was constructed from lightweight concrete bricks and the internal walls were made from hollow ceramic bricks. A single window 2.40 m wide and 1.40 m high was located in the front wall, see Fig. 11. The steel columns were partially encased in the walls with the flanges exposed. During the fire test, the exposed column flanges were protected by fibre-silicate boards. The beams and connections were unprotected during the test. The mechanical load was applied on the third floor. The total load, including selfweight of the structure, was 5.7 kN/m2 . Wooden cribs were used as the fire load, which created the fire load density of 1039 MJ/m2 . Thermocouples, strain gauges and displacement transducers were used to record the behaviour of the structure during the fire test. In total, 44 thermocouples were used to measure the gas temperature in the fire compartment, the temperature of the structural elements and the connections. Vertical deformations were measured using five transducers located at the mid-span of the primary and secondary beams. In addition, the relative horizontal displacement of the columns A2–D2, D1–D2 and D2–D3 were measured. Strains in the columns were measured using16 strain gauges attached to the column flanges on the first and third floors. Fire and smoke development was recorded by five video cameras and three thermo-imaging cameras. Two fire tests were carried out on the building. The first, a localised fire test, was performed on 15 June 2006, and was setup
F. Wald et al. / Journal of Constructional Steel Research 65 (2009) 1896–1903
1901
Fig. 9. (a) The building in Ostrava before demolition with the fire compartment on the second floor, (b) the applied fire load.
Fig. 10. (a) The concrete slab, (b) the beam-to-beam and (c) beam-to-column connections in the structure in Ostrava.
to measure the temperature of the steel column and beams close to the centre of the fire. The fire was located in the middle of the compartment below the primary beam A2–D2. The column was erected close to the fire. The second was a compartment test and was performed on 16 June 2006, and was designed to obtain the gas temperature in the fire compartment, the temperatures of the beams (measured at the mid-span lower flange), see Fig. 12, and at the connections. No collapse occurred during the tests; however, deformations and lateral–torsional instability of the beams were observed.
Fig. 11. Dimensions of the fire compartment in the Ostrava fire test.
5. Horizontal forces measured during the Ostrava test The horizontal forces were derived from the strains measured by the strain gauges attached to column flanges; see Fig. 13 [12]. The strain gauges for ambient temperature were located outside the fire compartment. The columns in the compartment were fire protected therefore the temperature of the columns at the location of the strain gauges did not exceed 120 ◦ C and correction for the modulus of elasticity of steel for the temperature was not necessary. The measured strains were transformed into stress increments. Average stress and section modulus of the column (Wy = 994 900 mm3 , Wz = 186 800 mm3 , Wz 0 = 233 500 mm3 ) were used to determine the bending moment in the column. The horizontal forces were obtained from a continuous beam model representing column D2, see Fig. 14. It is assumed that the bending moment diagram is linear along the column and the beam-tocolumn connections and column bases are simple connections (no rotational stiffness). The shear force diagram was derived from the bending moments by differentiation and finally, horizontal forces
Fig. 12. Comparison of the measured temperatures on the steel structure during the Ostrava test to the measured average gas temperature (from thermocouples TG1, TG2, TG3 and TG4).
were calculated. The FE simulation confirmed the accuracy of the simple beam prediction, see [9]. The maximum horizontal forces calculated from the measured values on the first and third floors were Ft ,3 = −215 kN (tension) and −50 kN (compression). The horizontal force on the second floor was Ft ,2 = −300 kN (compression) and +65 kN (tension). The average stress, bending moments and horizontal forces about
1902
F. Wald et al. / Journal of Constructional Steel Research 65 (2009) 1896–1903
Fig. 13. (a) and (b) The location of thermocouples on the first floor of column D2, (c) the column size and the position of the strain gauges during the Ostrava test.
Fig. 17. Horizontal forces in column D2, bending about the column’s major axis on the Ostrava frame.
Fig. 18. Relative displacement of columns A2–D2, bending about the columns’ major axes during the Ostrava test.
Fig. 14. Location of the strain gauges on column D2 and a model for evaluation of the horizontal forces on the Cardington frame.
Fig. 19. Average stress in column D2, bending about the column’s minor axis during the Ostrava test.
The tie force predicted according to Annex A of EN 1991-17: 2006 [6] for the load applied during the fire test is equal to 57 kN, however the value 75 kN should be used as a minimum value. Horizontal forces in the direction of the secondary beams (parallel to the minor axis of the columns, z-axis) were smaller. There are several reasons for this: Fig. 15. Average stresses in column D2, bending about the column’s major axis on the Ostrava frame.
• The secondary beams attached to column D2 were fire protected (encased in the partition walls), which prevented the thermal expansion of these beams. • The behaviour was influenced by the presence of the walls acting as a brace. The maximum forces on the first and third floor beams were
+8.4 kN (tension). The maximum force on the second floor was −10.5 kN (compression). The stress, bending moments, horizontal forces and relative displacement of the columns during the fire are plotted in Figs. 19–22. 6. Conclusion Fig. 16. Bending moments in column D2, bending about the column’s major axis on the Ostrava frame.
the major axis of the column (y-axis) are plotted in Figs. 15–17. Relative displacement of the columns A2–D2 is plotted in Fig. 18. The maximum displacement was 21 mm.
The full-scale fire test at Cardington performed on 16 January 2003 exhibited good structural integrity. Collapse of the structure was not reached despite the applied load reaching 145% of the predicted design resistance. The axial forces in the beam-tocolumn connections caused by the natural fire reached 460 kN in
F. Wald et al. / Journal of Constructional Steel Research 65 (2009) 1896–1903
Fig. 20. Bending moments in column D2, bending about the column’s minor axis during the Ostrava test.
1903
The horizontal forces obtained during these two fire tests may not be fully representative for all steel buildings. However, the forces indicate the magnitude of forces developed in an unprotected steel–concrete composite floor during the heating and cooling phases of a fire. The simple rule to use two M 16 bolts leads to a tension resistance 251 kN only (assuming bolt grade 8.8 and partial safety factor γM2 = 1) which is not sufficient for the Cardington frame (the tension force was 345 kN). The tie forces calculated according to Annex A of EN 1991-1-7: 2006 [6] are lower compared to forces measured during the compartment fires. The beam-to-column connections and the contact of the concrete slab to the column transferred the forces adequate for the transformation factor k = 1.15 for the Cardington 2003 test and k = 2.1 for the Ostrava 2006 test, respectively. Acknowledgements The experiments have been supported by the grants EU No. FP5 HPRI - CV 5535 and Grant Agency of Czech Republic No. 103/07/1142. This outcome has been achieved with the financial support of the Czech Ministry of Education, Youth and Sports, project No. 1M0579, within activities of the CIDEAS research centre.
Fig. 21. Horizontal forces in column D2, bending about the column’s minor axis during the Ostrava test.
Fig. 22. Relative displacement of columns D1–D2 and D2–D3, bending about the columns’ minor axes during the Ostrava test.
compression and 345 kN in tension. The test confirmed that the simple rules given in the Eurocode are suitable for the design of structural elements at high temperature, see Chapter 4 in [8], and verified the concept of unprotected beams and protected columns as a viable system for composite floors. The complementary test on the three-storey building in Ostrava confirmed the accuracy of the structural Eurocodes (gas temperature, heat transfer to the structure, temperature of the connections and behaviour of the structural elements exposed to compartment and localised fires) see [12]. The horizontal forces at three levels of the building were derived from the measured data. The horizontal forces reached 300 kN.
References [1] Lennon T. Cardington fire tests, survey of damage to the eight storey building. Building Research Establishment. Paper no. 127/97. Watford. 1997. [2] Advisory Desk: SCI answers to queries on steelwork design: AD027 – Accidental damage. Aug. 1998. vol. 2, no. 4; AD060 – Accidental Damage. Apr. 1990; AD 063 – Accidental damage – Tying. Sep. 1990. vol. 4; AD104 – BS 5950 – Tying Forces, Mar. 1992. vol. 6, no. 2. Steel Construction Today. [3] Advisory Desk: AD131 – Structural integrity – Tying to BS 5950, Part 1. New Steel Construction. Feb. 1993. vol. 1. no. 2. p. 29–30. [4] Way AGJ. Guidance on meeting the robustness requirements in approved document A. 2004 ed. Ascot: The Steel Construction Institute; 2005. [5] Owens GW, Moore DB. The robustness of simple connections. The Structural Engineer 1992;70(3):37–45. [6] Eurocode 1, EN 1991-1-7: 2006. Actions on structures Part 1–7: General actions – accidental actions. (Brussels): CEN; 1996. [7] Wald F, Simoes da Silva L, Moore D, Lennon T, Chladná M. Experimental behaviour of a steel structure under natural fire. Fire Safety Journal 2006; 41(7):509–22. [8] EN 1993-1-2. Eurocode 3: Design of steel structures. Part 1-2: General rules Structural fire design. (Brussels): CEN; 2005. [9] O’Connor MA, Martin MD. Behaviour of a multi-storey steel framed building subjected to fire attack. Journal of Constructional Steel Research 1998; 46(1–3):295. [10] Sokol Z, Wald F, Pultar M, Beneš M. Numerical simulation of Cardington fire test on structural integrity. In: Mathematical and computer modelling in science and engineering. Prague: Czech Technical University in Prague; 2003. p. 339–43. [11] Sokol Z, Wald F. Variations of forces in a real steel structure tested in fires. In: Urban habitat constructions under catastrophic events. In Proceedings of workshop. 2007. p. 80–5. [12] Wald F, Chlouba J, Kallerová P. Temperature of the header plate connection subject to a natural fire. In: Urban habitat constructions under catastrophic events, proceedings of workshop. 2007. p. 98–103.
Contents lists available at ScienceDirect
Journal of Constructional Steel Research journal homepage: www.elsevier.com/locate/jcsr
Horizontal forces in steel structures tested in fire František Wald a,∗ , Zdeněk Sokol a , David Moore b a
Czech Technical University in Prague, Prague, Czech Republic
b
British Constructional Steelwork Association, London, United Kingdom
article
info
Article history: Received 13 November 2007 Accepted 21 April 2009 Keywords: Structural engineering Steel structures Fire test Full-scale test Fire design Eurocodes Connections Robustness Horizontal forces
abstract Fire tests carried out on the eight-storey steel framed building at the Building Research Establishment’s Cardington laboratory have shown that the connections are subject to large axial force. These forces are the result of thermal movements of the structure during heating and cooling and in some cases can result in failure of the structure. This paper describes a fire test carried out on the steel frame at Cardington on 16th January 2003 and a fire test carried out on a structure in Ostrava on 16th June 2006. In both cases the tests were designed to measure the forces generated in the connections. © 2009 Elsevier Ltd. All rights reserved.
1. Introduction Full-scale fire tests carried out on the eight-storey steel framed building at the Building Research Establishment’s Cardington facility have shown that the connections are subject to high axial forces as a result of thermal movements during heating and cooling, see [1]. Current design methods do not check the behaviour of the connections under these conditions which can be critical, particularly in the cooling phase of the fire. Determining these forces is not easy as they are dependent on the form of the structure, the boundary conditions, the time–temperature curve, the thermal movements and local failures during heating and the behaviour of the deformed structure during the cooling phase of the fire. Observations from the fire tests at Cardington also showed that a partially protected composite flooring system deforms to such an extent that it supports the applied load in a combination of bending and catenary actions transferring significant axial forces to the supporting connections. Although the axial capacity of a connection is not routinely checked under the fire situation the axial capacity of a connection is checked as a way of providing a structure with adequate robustness against disproportionate collapse in the event of an
∗ Corresponding address: Department of Steel Structures, Czech Technical University in Prague, Thákurova 7, 166 29 Prague 6, Czech Republic. Tel.: +420 224 354 757; fax: +420 233 337 466. E-mail addresses: [email protected] (F. Wald), [email protected] (Z. Sokol), [email protected] (D. Moore). 0143-974X/$ – see front matter © 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.jcsr.2009.04.020
accidental action such as a gas explosion or impact from a vehicle, see [2]. Early work on robustness required the connections in a steel framed structure to have at least two M 16 bolts in tension to ensure structural integrity., This was replaced by a estimation of the tie forces in the connections based on catenary behaviour of the supported beams. see [3–5]. There are many similarities between the catenary action of a flooring system under fire and that under accidental actions. It is postulated that the design of the connections for axial tying capacity will allow the connections to resist the axial forces generated as a result of the floor going into catenary under fire. If this could be proven no further checks at elevated temperature would be required. The aim of this paper is to determine the axial forces that connections are subject to under heating and cooling and to compare these with the tying forces used in design for progressive collapse. Annex A of EN 1991-1-7: 2006 [6], gives the following expression for calculating tie forces: Ti = min[k(gk + ψ qk )sL; 75 kN]
(1)
where k is the transformation factor; for internal ties k = 0.8; for perimeter ties k = 0.4, gk is the characteristic value of permanent action, ψ is the combination factor according to the accidental load combination, qk is the characteristic value of variable action, s is the spacing of the ties, and L is the span of the tie.
F. Wald et al. / Journal of Constructional Steel Research 65 (2009) 1896–1903
1897
Nomenclature fub fy gk k kE ,θ ky,θ s qk As L E Ft ,b Ft ,2 Ft ,3 Ft ,4 Ti Wy Wz
γM2 ψ εθ
is the ultimate strength of the bolt is the yield stress at ambient temperature is the characteristic value of permanent action is the transformation factor; k = 0.8 for internal ties; k = 0.4 for perimeter ties is the reduction factor for the slope of the linear elastic range at the steel temperature is the reduction factor for the yield stress at the temperature of the steel is the spacing of the ties is the characteristic value of the variable action is the tensile stress area of the bolt is the span of the tie is the elastic modulus of the steel is the design tension resistance of the bolt is the horizontal force to the column on the second floor is the horizontal force to the column on the third floor is the horizontal force to the column on the fourth floor is the tie force is the section modulus to y-axis, major axis is the section modulus to z-axis, minor axis is the partial safety factor for the bolt is the combination factor according to the accidental load combination is the strain at elevated temperature
2. Seventh large-scale fire test on a steel frame in Cardington A structural integrity fire test was carried out on the eightstorey steel framed building at Cardington on 16 January 2003, see [1]. The main purpose of this test was to collect data on the behaviour of typical beam-to-column and beam-to-beam connections subjected to a natural fire, see [7]. The test was carried out in a compartment on the fourth floor enclosing a plan area of 11 × 7 m, as shown in Fig. 1. The internal walls of the compartment were made of three layers of plasterboard (15 mm + 12.5 mm + 15 mm) with a thermal conductivity of 0.19–0.24 W m−1 K−1 . The external wall was a 0.9 m brick window sill and 1.95 m plasterboard wall. A 1.27 m high and 5.7 m wide opening simulated an open window and created ventilation for the compartment. The size of the opening was designed to produce a fire with a temperature exceeding 1200 ◦ C and a duration of 60 min. The steel structure within the compartment consisted of four columns (internal columns were 305 × 305 × 198UC section, edge columns 305 × 305 × 137UC section, steel grade S355), two primary beams (336 × 171 × 51UB section, steel grade S350), two secondary beams (305 × 165 × 40UB section, steel grade S275) and edge beam (356 × 171 × 51UB section), see [1]. Flexible end plates were used for beam-to-column connections and fin plates for the beamto-beam connections. In both cases, the plates were made from steel grade S275 and M20 bolts, grade 8.8, were used. To prevent the collapse of the structure, the columns were fire protected by a 20 mm thick layer of Cafco300 vermiculite-cement spray with a thermal conductivity of 0.078 W m−1 K−1 . In addition, protection was also applied to joints on external columns and parts of the primary beams (approximately at a distance 1.0 m from the joints), see Fig. 2. A lightweight concrete slab cast on the profiled metal decking was supported on the primary and secondary beams. 19 mm diameter shear studs were used on all the beams.
Fig. 1. The position of the fire compartment of the seventh large-scale fire test on the plan of the Cardington frame.
The applied load was applied using sandbags distributed over an area of 18 m by 10.5 m on the floor above the fire compartment. This load represented the permanent action (including floor layers and partition walls) and 56% of the variable action. Wooden cribs with a moisture content of 14% provided a fire load of 40 kg/m2 . According to analytical and finite element simulations, failure of the concrete slab was expected during the fire test. Thermocouples, strain gauges and displacement transducers were used to measure various data during the test. A total of 133 thermocouples monitored the temperature of the connections and beams within the compartment, the temperature of the concrete slab and the gas temperature within the compartment. Additional 14 thermocouples were used to measure the temperature of the columns, see Fig. 3. Nine high-temperature strain gauges were used to measure the strains in the unprotected fin-plate and end-plate–minor-axis joints. A total of 47 ambient-temperature strain gauges were attached to the protected columns and to the concrete slab. See Fig. 4 for the location of the strain gauges. The vertical deformations of the concrete slab were measured by 25 displacement transducers installed on the fifth floor. Additional 12 transducers measured the horizontal movement of the columns and the slab. Ten video cameras recorded the fire and smoke development and the deformations, and two thermo-imaging cameras were used for measuring the temperature distribution in the steel elements. 3. Horizontal forces measured during the Cardington test High-temperature strain gauges were attached to the beams next to the connections to measure the strains induced by the fire. These strains were used to determine the axial forces in the connections during the fire. The high-temperature strain gauges are capable of measuring strains up to a temperature of 1200 ◦ C. The stress σθ at the elevated temperature was derived from the measured strain using modulus of elasticity reduced for the steel temperature (Ea,θ = kE ,θ E). The corresponding steel temperature was recorded by the thermocouple attached to the structure near the strain gauges. The calculations are given in Tables 1 and 2. The stress is limited by yielding of the steel. The resulting stress is calculated as
σθ = min(kE ,θ E ε; ky,θ fy )
(2)
where kE ,θ E
ε
ky,θ fy
θa
is the reduction factor for the slope of the linear elastic range at the corresponding steel temperature, see [8] is the elastic modulus of steel is the measured strain is the reduction factor for the yield stress at the corresponding steel temperature, see [8] is the yield stress at ambient temperature, 396 MPa based on coupon tests of beams for Cardington frame [7] is the steel temperature measured near the strain gauge.
1898
F. Wald et al. / Journal of Constructional Steel Research 65 (2009) 1896–1903
Fig. 2. (a) The fire load in the compartment; (b) the fire load around column D2; (c) the installation of the strain gauges on the column above the fire compartment; (d) and (e) the installation of the strain gauges on the column in the fire compartment; (f) the fire protection of the internal column E2 after the test; (g) the fire protection of the external column E1 after the test on the Cardington frame. Table 1 Measured strains on beam D1.5–E1.5 close to fin plate connection using high-temperature strain gauges. Time (min)
0 5 10 15 20 25 30 45 60 90 118 148 178 Max. Min.
C 444
HT 001
Temp. (◦ C)
Strain (µε × 10−3 )
Eθ (MPa)
18 24 42 66 114 212 331 636 810 702 476 299 205 834 18
0.000 0.051 0.198 0.248 −2.468 −2.554 −3.598 −6.425 0.294 0.400 0.393 0.351 0.317 0.458 −8.066
210 000 210 000 210 000 210 000 207 104 186 459 161 468 51 552 18 413 27 135 131 193 168 317 187 872 210 000 17 278
C 445
HT 003
Stress (MPa)
Temp. (◦ C)
Strain (µε × 10−3 )
0 11 42 52 −388 −335 −298 −120 5 11 52 59 59 68 −394
18 25 41 71 126 242 362 669 888 694 444 252 168 893 18
0.000
−0.068 −0.217 0.325 8.507 8.552 9.519 9.408 10.751 11.841 9.418 8.678 8.396 11.841 −0.217
The results are presented in Fig. 5a for the secondary beam D1.5–E1.5 connected to the primary beam (fin plate connection, located at the mid-span of the primary beam) and in Fig. 5b for the secondary beam connected to the columns (end plate connection). Due to a nonlinear stress–strain relationship Eq. (2) gives increased values at stains larger than the proportional limit. This deference, which reached up to 10% was neglected. The stress development shows the behaviour of the composite beam during the fire, see [9]. Local buckling of the lower flange in
C 446
HT 005
Eθ (MPa)
Stress (MPa)
Temp. (◦ C)
Strain (µε × 10−3 )
Eθ (MPa)
Stress (MPa)
210 000 210 000 210 000 210 000 204 522 180 229 155 035 39 165 14 722 29 453 137 821 178 062 195 659 210 000 14 486
0 14 46 −68 −391 −372 −359 −106 −25 −89 −320 −370 −384 46 −396
18 23 40 71 123 236 369 692 899 717 474 289 195 908 18
−0.046 −0.138 −0.084
210 000 210 000 210 000 210 000 205 204 181 518 153 551 30 464 14 218 25 895 131 525 170 299 190 085 210 000 13 783
0 10 29 18 −383 −320 −255 −51 −19 −50 −150 20 148 268 −388
2.574 2.419 2.496 2.269 2.827 3.232 3.261 1.139 −0.012 −0.077 −1.646 0.000
compression was observed during the test. There are two principal reasons for high compression forces in the flange:
• The beam is restrained from expanding longitudinally, which results in the development of axial compressive forces in the beam; • The rotational stiffness of the joint, neglected in most design models, allows the development of bending moments at the supports, which results in compression of the lower flange. The
F. Wald et al. / Journal of Constructional Steel Research 65 (2009) 1896–1903
1899
Table 2 Measured strains on beam D2–E2 close to end plate connection using high-temperature strain gauges. Time (min)
0 5 10 15 20 25 30 45 60 90 118 148 178 Max. Min.
C 462
HT 009
Temp. (◦ C)
Strain (µε × 10−3 )
Eθ (MPa)
18 21 135 63 114 182 281 545 774 726 474 295 204 847 18
0.000 −0.017 −0.102 −0.292 −0.756 −1.936 −1.483 −1,313 −0.787 −0.158 Failed Failed Failed 0.000 −1.936
210 000 210 000 210 000 210 000 207 003 192 736 171 922 98 624 21 081 25 129 – – – 210 000 16 696
Stress (MPa) 0
−4 −21 −61 −157 −342 −255 −123 −17 −4 – – – 0
−342
C 461
HT 011
HT 021
Temp. (◦ C)
Strain (µε × 10−3 )
Eθ (MPa)
18 20 33 61 116 187 273 492 706 726 513 334 235 781 18
0.000 −0.055 −0.333 −1.196 −1.718 Failed Failed Failed Failed Failed Failed Failed Failed 0 −7.855
210 000 210 000 210 000 210 000 204 895 – – – – – – – – 210 000 18 897
Stress (MPa) 0
−12 −70 −251 −352 – – – – – – – – 0
−354
Temp. (◦ C)
Strain (µε × 10−3 )
Eθ (MPa)
17 19 35 70 124 201 284 569 791 713 500 333 253 800 17
0.000 −0.041 −0.339 −1.058 −1.451 −1.988 −2.395 −1.197 −1.061 1.734 Failed Failed Failed 1.743 −2.398
210 000 210 000 210 000 210 000 204 895 188 732 171 403 84 043 19 625 26 183 – – – 210 000 18 897
Stress (MPa) 0
−9 −71 −222 −297 −331 −294 −99 −21 39 – – – 41
−339
Fig. 5a. Stress in the secondary beam at the beam-to-beam connection E1.5.
Fig. 3. Comparison of the measured temperatures along the external column D1 to the gas, beam and internal column D2 temperatures during the Cardington test.
bending moment at the connection may develop after the lower flange comes into contact with the column. This occurs after significant rotation in the connection. • The stress in the lower flange increased rapidly to the yield stress, which is reached in the 17th min of the test.
The column fire protection limited the temperature rise in the columns and allowed the strains to be measurement using ambient-temperature strain gauges. The measurements were taken during the first 60 min of the fire on the third floor and during the whole experiment on the fourth floor, see Fig. 3. Selected results of strain measurements are presented in Fig. 6a (stresses in the edge columns on the fourth floor located 500 mm above the floor) and in Fig. 6b (stresses in the edge columns on the fourth floor located 500 mm below the ceiling). The stresses in
Fig. 4. Location (a) of the ambient-temperature strain gauges on columns and (b) the high-temperature strain gauges and the thermocouples on beams close to the connections on the Cardington frame.
1900
F. Wald et al. / Journal of Constructional Steel Research 65 (2009) 1896–1903
Fig. 7. The bending moments in the columns measured during the fire test.
Fig. 5b. Stress in the secondary beam at the beam-to-column connections D2 and E2.
Fig. 8. The horizontal forces calculated from the measured bending moments using a beam theory.
4. Fire test on a steel frame in Ostrava Fig. 6a. The stress in the external columns on the fourth floor, 500 mm above the floor.
Fig. 6b. The stress in the external columns on the fourth floor, 500 mm below the ceiling.
the columns were used to calculate the bending moments, see Fig. 7. The shear force in the columns was derived from the bending moment distribution in the columns and, finally, the horizontal forces in the beam-to-column connection were calculated. This simple calculation is based on the continuous beam model, see Fig. 8. The maximal calculated horizontal forces were Ft ,3 = −344 kN (tension), and −65 kN (compression) on the third floor, and Ft ,4 = −462 kN (compression) and +88 kN (tension) on the fourth floor. A global structural FE analysis of the frame was used to verify these values. The nonlinear procedure utilised 2D plastic beam elements and a simplified model of the composite slab. The model included nonlinear temperature-dependent material properties, nonlinear response of the composite beam-to-column joints, thermal expansion, large strain and large deformations. The results of the FE simulation confirmed the beam model approach see [10,11].
On 16 May 2006 a fire test was carried out on a threestorey administrative building which was attached to a singlestorey framed building of the Ammoniac Separator II in the Mittal Steel Plant in Ostrava, Czech Republic, see Fig. 9 [9]. The loadbearing structure, which was completed in 1992, consisted of steel columns and beams supporting a 130 mm (total thickness including ribs) concrete slabs. No shear connection was provided between the steel beams and the concrete slab. The beam-to-beam and beam-to-column connections were designed as simple end plate connections using two or six M20 bolts, see Fig. 10. The fire compartment was 3.80 × 5.95 m with a height of 2.78 m and was built on the second floor. The front wall was constructed from lightweight concrete bricks and the internal walls were made from hollow ceramic bricks. A single window 2.40 m wide and 1.40 m high was located in the front wall, see Fig. 11. The steel columns were partially encased in the walls with the flanges exposed. During the fire test, the exposed column flanges were protected by fibre-silicate boards. The beams and connections were unprotected during the test. The mechanical load was applied on the third floor. The total load, including selfweight of the structure, was 5.7 kN/m2 . Wooden cribs were used as the fire load, which created the fire load density of 1039 MJ/m2 . Thermocouples, strain gauges and displacement transducers were used to record the behaviour of the structure during the fire test. In total, 44 thermocouples were used to measure the gas temperature in the fire compartment, the temperature of the structural elements and the connections. Vertical deformations were measured using five transducers located at the mid-span of the primary and secondary beams. In addition, the relative horizontal displacement of the columns A2–D2, D1–D2 and D2–D3 were measured. Strains in the columns were measured using16 strain gauges attached to the column flanges on the first and third floors. Fire and smoke development was recorded by five video cameras and three thermo-imaging cameras. Two fire tests were carried out on the building. The first, a localised fire test, was performed on 15 June 2006, and was setup
F. Wald et al. / Journal of Constructional Steel Research 65 (2009) 1896–1903
1901
Fig. 9. (a) The building in Ostrava before demolition with the fire compartment on the second floor, (b) the applied fire load.
Fig. 10. (a) The concrete slab, (b) the beam-to-beam and (c) beam-to-column connections in the structure in Ostrava.
to measure the temperature of the steel column and beams close to the centre of the fire. The fire was located in the middle of the compartment below the primary beam A2–D2. The column was erected close to the fire. The second was a compartment test and was performed on 16 June 2006, and was designed to obtain the gas temperature in the fire compartment, the temperatures of the beams (measured at the mid-span lower flange), see Fig. 12, and at the connections. No collapse occurred during the tests; however, deformations and lateral–torsional instability of the beams were observed.
Fig. 11. Dimensions of the fire compartment in the Ostrava fire test.
5. Horizontal forces measured during the Ostrava test The horizontal forces were derived from the strains measured by the strain gauges attached to column flanges; see Fig. 13 [12]. The strain gauges for ambient temperature were located outside the fire compartment. The columns in the compartment were fire protected therefore the temperature of the columns at the location of the strain gauges did not exceed 120 ◦ C and correction for the modulus of elasticity of steel for the temperature was not necessary. The measured strains were transformed into stress increments. Average stress and section modulus of the column (Wy = 994 900 mm3 , Wz = 186 800 mm3 , Wz 0 = 233 500 mm3 ) were used to determine the bending moment in the column. The horizontal forces were obtained from a continuous beam model representing column D2, see Fig. 14. It is assumed that the bending moment diagram is linear along the column and the beam-tocolumn connections and column bases are simple connections (no rotational stiffness). The shear force diagram was derived from the bending moments by differentiation and finally, horizontal forces
Fig. 12. Comparison of the measured temperatures on the steel structure during the Ostrava test to the measured average gas temperature (from thermocouples TG1, TG2, TG3 and TG4).
were calculated. The FE simulation confirmed the accuracy of the simple beam prediction, see [9]. The maximum horizontal forces calculated from the measured values on the first and third floors were Ft ,3 = −215 kN (tension) and −50 kN (compression). The horizontal force on the second floor was Ft ,2 = −300 kN (compression) and +65 kN (tension). The average stress, bending moments and horizontal forces about
1902
F. Wald et al. / Journal of Constructional Steel Research 65 (2009) 1896–1903
Fig. 13. (a) and (b) The location of thermocouples on the first floor of column D2, (c) the column size and the position of the strain gauges during the Ostrava test.
Fig. 17. Horizontal forces in column D2, bending about the column’s major axis on the Ostrava frame.
Fig. 18. Relative displacement of columns A2–D2, bending about the columns’ major axes during the Ostrava test.
Fig. 14. Location of the strain gauges on column D2 and a model for evaluation of the horizontal forces on the Cardington frame.
Fig. 19. Average stress in column D2, bending about the column’s minor axis during the Ostrava test.
The tie force predicted according to Annex A of EN 1991-17: 2006 [6] for the load applied during the fire test is equal to 57 kN, however the value 75 kN should be used as a minimum value. Horizontal forces in the direction of the secondary beams (parallel to the minor axis of the columns, z-axis) were smaller. There are several reasons for this: Fig. 15. Average stresses in column D2, bending about the column’s major axis on the Ostrava frame.
• The secondary beams attached to column D2 were fire protected (encased in the partition walls), which prevented the thermal expansion of these beams. • The behaviour was influenced by the presence of the walls acting as a brace. The maximum forces on the first and third floor beams were
+8.4 kN (tension). The maximum force on the second floor was −10.5 kN (compression). The stress, bending moments, horizontal forces and relative displacement of the columns during the fire are plotted in Figs. 19–22. 6. Conclusion Fig. 16. Bending moments in column D2, bending about the column’s major axis on the Ostrava frame.
the major axis of the column (y-axis) are plotted in Figs. 15–17. Relative displacement of the columns A2–D2 is plotted in Fig. 18. The maximum displacement was 21 mm.
The full-scale fire test at Cardington performed on 16 January 2003 exhibited good structural integrity. Collapse of the structure was not reached despite the applied load reaching 145% of the predicted design resistance. The axial forces in the beam-tocolumn connections caused by the natural fire reached 460 kN in
F. Wald et al. / Journal of Constructional Steel Research 65 (2009) 1896–1903
Fig. 20. Bending moments in column D2, bending about the column’s minor axis during the Ostrava test.
1903
The horizontal forces obtained during these two fire tests may not be fully representative for all steel buildings. However, the forces indicate the magnitude of forces developed in an unprotected steel–concrete composite floor during the heating and cooling phases of a fire. The simple rule to use two M 16 bolts leads to a tension resistance 251 kN only (assuming bolt grade 8.8 and partial safety factor γM2 = 1) which is not sufficient for the Cardington frame (the tension force was 345 kN). The tie forces calculated according to Annex A of EN 1991-1-7: 2006 [6] are lower compared to forces measured during the compartment fires. The beam-to-column connections and the contact of the concrete slab to the column transferred the forces adequate for the transformation factor k = 1.15 for the Cardington 2003 test and k = 2.1 for the Ostrava 2006 test, respectively. Acknowledgements The experiments have been supported by the grants EU No. FP5 HPRI - CV 5535 and Grant Agency of Czech Republic No. 103/07/1142. This outcome has been achieved with the financial support of the Czech Ministry of Education, Youth and Sports, project No. 1M0579, within activities of the CIDEAS research centre.
Fig. 21. Horizontal forces in column D2, bending about the column’s minor axis during the Ostrava test.
Fig. 22. Relative displacement of columns D1–D2 and D2–D3, bending about the columns’ minor axes during the Ostrava test.
compression and 345 kN in tension. The test confirmed that the simple rules given in the Eurocode are suitable for the design of structural elements at high temperature, see Chapter 4 in [8], and verified the concept of unprotected beams and protected columns as a viable system for composite floors. The complementary test on the three-storey building in Ostrava confirmed the accuracy of the structural Eurocodes (gas temperature, heat transfer to the structure, temperature of the connections and behaviour of the structural elements exposed to compartment and localised fires) see [12]. The horizontal forces at three levels of the building were derived from the measured data. The horizontal forces reached 300 kN.
References [1] Lennon T. Cardington fire tests, survey of damage to the eight storey building. Building Research Establishment. Paper no. 127/97. Watford. 1997. [2] Advisory Desk: SCI answers to queries on steelwork design: AD027 – Accidental damage. Aug. 1998. vol. 2, no. 4; AD060 – Accidental Damage. Apr. 1990; AD 063 – Accidental damage – Tying. Sep. 1990. vol. 4; AD104 – BS 5950 – Tying Forces, Mar. 1992. vol. 6, no. 2. Steel Construction Today. [3] Advisory Desk: AD131 – Structural integrity – Tying to BS 5950, Part 1. New Steel Construction. Feb. 1993. vol. 1. no. 2. p. 29–30. [4] Way AGJ. Guidance on meeting the robustness requirements in approved document A. 2004 ed. Ascot: The Steel Construction Institute; 2005. [5] Owens GW, Moore DB. The robustness of simple connections. The Structural Engineer 1992;70(3):37–45. [6] Eurocode 1, EN 1991-1-7: 2006. Actions on structures Part 1–7: General actions – accidental actions. (Brussels): CEN; 1996. [7] Wald F, Simoes da Silva L, Moore D, Lennon T, Chladná M. Experimental behaviour of a steel structure under natural fire. Fire Safety Journal 2006; 41(7):509–22. [8] EN 1993-1-2. Eurocode 3: Design of steel structures. Part 1-2: General rules Structural fire design. (Brussels): CEN; 2005. [9] O’Connor MA, Martin MD. Behaviour of a multi-storey steel framed building subjected to fire attack. Journal of Constructional Steel Research 1998; 46(1–3):295. [10] Sokol Z, Wald F, Pultar M, Beneš M. Numerical simulation of Cardington fire test on structural integrity. In: Mathematical and computer modelling in science and engineering. Prague: Czech Technical University in Prague; 2003. p. 339–43. [11] Sokol Z, Wald F. Variations of forces in a real steel structure tested in fires. In: Urban habitat constructions under catastrophic events. In Proceedings of workshop. 2007. p. 80–5. [12] Wald F, Chlouba J, Kallerová P. Temperature of the header plate connection subject to a natural fire. In: Urban habitat constructions under catastrophic events, proceedings of workshop. 2007. p. 98–103.