Fire spalling of concrete is not a new phenomenon. To some degree there has always been a risk during rapid heating of concrete. Therefore, to a certain degree the effect of fire spalling is included in the bank of data from fire tests and fires on which our understanding of the fire resistance of concrete is based. However, the development and modern use of more dense concrete mixes have produced cases of very severe fire spalling which have increased the urgency to understand this phenomenon. In this context, the use of an addition of polypropylene (PP) fibres to the mix to limit the amount of spalling has been one topic of interest for this thesis.
During fire tests on a post-tensioned concrete structure made of spalling sensitive concrete, it has been shown that substantially lower amounts of PP fibres than 2 kg/m3, which is recommended in the Eurocode (1992-1-2:2004), can be used with successful results.
As part of this study, another important aspect has emerged, i.e. the impact the test method used can have on the fire spalling depths observed in concrete specimens. This has been known for many years but is seldom discussed in the scientific literature. In this thesis it has been shown that results from tests on unloaded cubes do not necessarily correspond to results seen on larger loaded slabs. In the results presented, none of the tested cubes spalled whereas some of the large slabs spalled to the degree that the reinforcement became fire exposed. Further, the difference in spalling depths between small and large post-tensioned slabs was shown to be substantial; although in general the ranking in severity from least to greatest spalling correlated between these two specimen sizes. The correlation to larger specimens was much vaguer in the case when the small slabs were not loaded in compression as there sometimes was no spalling in the small slabs.
From time to time the randomness of the fire spalling of concrete has been mentioned. To investigate this further, an analysis of 110 fire tests performed on small slab type specimens was performed. This analyse showed that the spalling behaviour had a good repeatability between two identical tests, which proved that the so called “random factor” relating to spalling depth was low for the chosen data set. It was also possible to make a multiple least squares fit of test parameters that could be used to predict the spalling behaviour which also underlines that a substantial stochastic factor was not present.
Regarding the influence of different factors, the results compiled on the influence of ageing show that for three of the tested Self-Compacting Concrete (SCC) mixes, the amount of spalling was reduced with age whereas for the fourth mix (which included the highest amount of limestone filler, 140 kg/m3) the spalling was not reduced with age. In this test series no systematic influence of the intensity of the fire, between standard fire exposure and the more severe hydrocarbon fire, on the spalling depth was detected for this type of specimen. The only major difference was that spalling started earlier during the more severe fire exposure.
Pressure measurements conducted as part of the work within this thesis, supported by results from the literature, indicate that there is no relationship between pressure rise due to moisture and fire spalling. Based on this and the fact that the spalling event in many cases happens at relatively low temperatures where the saturation vapour pressure is low two alternative factors to explain the function of PP fibres have been presented: (i) PP fibres reduce the moisture content in the critical zone close to the heated surface which affects the mechanical properties advantageously, and (ii) PP fibres amplify moisture movement leading to larger drying creep and shrinkage which locally relax the thermal stresses.
To investigate the influence of the presence of moisture on the compressive strength, specimens were tested after being boiled for varying periods of time in a water bath. The study showed a remarkable reduction of strength due to boiling of the mortar specimens. After boiling mortar in a water bath for 3, 10 or 20 minutes, i.e. approximately the same time span as the initiation of fire spalling during fully developed fires, the strength was only 64% of the corresponding value for a dry specimen. As no strength change was detected between the specimens boiled 3, 10 or 20 minutes, and that the corresponding saturation pressure for steam at 100ºC is negligible compared with the tensile strength of concrete, it was concluded that pore pressure is not a significant contributor to the measured reduction in strength. It appears that the presence of moisture itself rather than an increased pressure is the most important factor reducing strength. This is a clear indication that moisture plays a key role in the fire spalling of concrete but in a different way from previously assumed.
Stockholm: KTH Royal Institute of Technology, 2013. , xvi, 134 p.
2013-09-20, Kollegiesalen, Brinellvägen 8, entréplan. KTH, Stockholm, 10:00 (English)