Discussion of “Fire Resistance of Gypsum Board Protected Steel Columns with High Load Ratios” by Zheng Peng and Hossein Mostafaei

The original paper is an interesting and valuable contribution to the material on the fire protection of steel structures. The quite good agreement between the results of analyses using Abaqus and SAFIR and with the results of the fire tests shown in Figs. 14 and 15 is very encouraging.

However, the total lack of agreement with the results of the European Commission for Constructional Steelwork (ECCS) analysis is problematic. In a personal email, the authors said the ECCS curves were based on Table D3 of ECCS (1985). The background information to the table, in Chapter I of the report, makes it clear that the significant thermal mass of the insulation was not considered. The table is based on cases of lightweight insulation, which have low thermal conductivity, low specific heat, and low specific gravity, while gypsum board insulation has significantly higher thermal conductivity and much higher specific heat and specific gravity. This latter case is treated in an approximate manner in Chapter III of ECCS (1985), but the approach used cannot capture the significant delay in temperature increase in the steel in the first several minutes of fire exposure.

This problem was then explored using a series of one-dimensional heat-flow analyses described by Gamble (1989). The background to those analyses is described in ECCS Technical Committee 3 (1983) and Petterson et al. (1976). The 1989 paper described a series of relatively simple spreadsheets that computed the steel temperatures for bare steel, steel with insulation where the thermal capacity of the insulation could be ignored, and cases where the thermal capacity of the insulation is large compared to that of the steel section and must be included. Standard time-temperature curves for the fire can be considered, as well as any arbitrary time-temperature curve.

The Gamble (1989) analysis followed an ECCS Technical Committee 3 (1983) simplified rule to account for the rather large heat capacity of the gypsum board insulation, with results similar to those from ECCS (1985) Chapter III. The basic, not simplified, analysis method from ECCS Technical Committee 3 (1983) and Petterson et al. (1976) was both significantly more complex and led to some numerical problems for the first few time intervals. There was no obvious solution to the numerical problem at that time.

The numerical problem was later overcome, giving a way of taking the thermal mass of the insulation more directly into account. The resultant Fig. 1 shows the time-temperature curve computed using a newer spreadsheet for heavy insulation, that is, insulation with a high heat capacity, for the member shown in Fig. 14 of the original paper. The agreement with the curve in Fig. 1 and the measured and computed temperatures is quite good. Nothing predicted the gypsum board partial failure. The fire considered is the ISO 834 (ISO 1999) exposure, which is very similar to the ASTM E119 (ASTM 2012) time-temperature curve.

Fig. 2 shows the time-temperature curve computed using a spreadsheet for light insulation, that is, where the thermal capacity of the insulation is ignored, but with the much higher thermal conductivity appropriate to gypsum board. The agreement with the ECCS curve in Fig. 14 in the original paper is again quite good.

The original paper notes that ECCS (1985) also provides a simple treatment of sections with insulation with large thermal capacity, Eq. (5), and also a simpler Eq. (6) for insulation with low thermal capacity. Limited trials with both equations seem to give significantly shorter times to limiting temperatures in the 400°C to 600°C range than the other analyses just noted and thus may be quite conservative.