【正文】 ick Fireline Gyproc board manufactured by British Gypsum Limited and interiorly insulated by Isowool 1000, also manufactured byBritish Gypsum Limited. The gypsum plasterboards were fixed horizontally to the steel studs by screws at 300mm centres.All the steel ponents, including the lipped channel studs, the plain channel trances and the flat bar lateral bracing, were prehot dip galvanised to BSEN 10147 with a G275 coating. The steel grade is S350 GD+Z with a minimum yield strength of 350N/mm 2 . Tensile coupon tests were performed and Table 1gives the average measured yield strength and Young modulus for the lipped channels.. Test rig and instrumentationAmbient temperature tests: The ambient temperature tests were conducted in a purpose built test rig, shown in Fig. 2. The test specimen shown in Fig. 1 was inserted between two guiding unlipped channels, one at the top that was bolted to a moving beam (UC20320346) and one at the bottom that was connected to the supporting beam (UB40617860) of the reaction frame, the guiding channels being separated from the beams of the reaction frame by rollers. The two guiding channels were in direct contact with the top or bottom steel trances on the test specimen so as to allow the specimen to rotate out of plane but prevent the inplane rotation. Detailed measurements of deformation were taken at a number of locations of the panel. As shown in Fig. 2, nine displacement transducers were used to record the vertical inplane movement of the panel, six(transducers 4–9) at the panel top and three (transducers 19–21) at the transducers were also placed at three locations along each of the three channels to measure outofplane movements of the test panel (displacement transducers 10–18).Fire tests: To conduct the fire tests, the reaction frame described above and shown in Fig. 2 was placed in front of a big furnace that measures 443m 3 externally and about 3 internally. In addition to the same placement of transducers as in the ambient temperature tests, a total of 30 thermocouples were placed within each specimen to obtain temperature developments at five different locations as shown in Fig. 2.. Test procedureEach test specimen was placed in the reaction frame by putting into the support channels between the moving and supporting beams of the reaction frame. However,due to inaccuracy in fabrication and imprecision in placing the test specimen, it is inevitable that the specimen might not have been loaded exactly in the centre.Three hydraulic jacks, located directly above the three lipped channels of the test panel, applied the loads that were transmitted to the panel through the moving steel beam of the reaction frame on top of the test panel. The applied loads were recorded by load cells. During each ambient temperature test, the axial load in each jack was applied slowly using increments of at the beginning and smaller increments ofabout 1kN near the calculated failure load. The specimen was considered to have failed when it was not possible to pump up the oil pressure inside the hydraulic jacks.For each fire test, each jack slowly applied loads in increments of 2kN to the prescribed target load. The load was kept unchanged throughout the fire test. For the three panels using each type of lipped channels, three load ratios, being , and , were used respectively. The load ratio is defined as the ratio of the applied load in the fire test to the panel strength determined by the ambient temperature test.These three load ratios were considered to represent the low, medium and high loads in practical situations. The fire test was terminated when the initially applied load could not be maintained.3. Observations. Ambient temperature testsPanel with lipped channels : Only one test was performed. When the applied load in each individual jack reached about 23kN, lateral movement of the panel became noticeable. When the applied load approached 54kN, the load on the central channel could not be maintained but loads on the two side channels could be increased slightly more. As can be seen from Fig. 3, failure was by flexural buckling about the major axis of the channels, initiated by local buckling of the steel channels at the top service hole. The gypsum plasterboards stayed in position and effectively restrained the steel channels from torsional buckling and flexural buckling about the minor with lipped channels 10056152: When the applied load in each jack reached 50kN, bending of the test specimen could clearly be noticed. When the applied load approached 103kN, the channel on one side suddenly failed. The applied loads on the other two channels were increased to 112kN when the channels also experienced failure. The panel failure mode was also caused by local buckling near the top service hole. Again the gypsum plasterboards restrained minor axis buckling of the steel channels.In both tests, some screws were pulled through the gypsum plasterboards at the locally buckled section.. Fire testsTable 2 summarises the six fire tests. In all fire tests, the furnace temperature was regulated so that the average temperature recorded by the six control thermocouples inside the furnace followed the standard BS 476 fire temperature–time relationship. However, as will be shown later, since the furnace control software was not very refined at this stage, the average furnace temperature oscillated around the standard fire curve.. Panels using 10056152 studsLoad ratio=: When the applied load in each jack reached the target load of and before the furnace was ignited, the test panel had noticeably moved towards the furnace. After ignition, temperatures on the exposed side of the panel increased quickly and those on the unexposed side remained low, introducing severe temperature gradients through the panel thickness and causing the panel to bow towards the furnace. After 43min of fire test, the applied load on one side of the panel could