Home > News > Content


Behaviour of concrete in extreme fires

Concrete does not ‘melt’ in the majority of extreme fire conditions but it could do so in conditions such as created by, for example, a thermic lance (steel burning in a pure oxygen environment). However, this is exceptional and is not normally considered in the design of reinforced concrete subject to hydrocarbon fires.

The reported experience for this combination of materials and hydrocarbon fire is limited and there is little published information or guidance on this aspect compared to laboratory fire tests. One of the most helpful early reports was that on the consequences of a tanker explosion and fire while moored to a concrete jetty in Bantry Bay (Maling, 1987), Ireland. The limited reports showed a very rapid loss of cover concrete (50–75 mm) but that spalling beneath the reinforcement was limited. The records of the condition of the Channel Tunnel concrete linings after a vehicle fire (NCE, 1996) appear to show
localized complete penetration of the concrete lining. It is possible that the confined tunnel environment created more aggressive temperature conditions than those for the ‘open’ jetty at Bantry Bay.

Currently, although there are data available from both real extreme fires and tests, it is fragmented and is somewhat contradictory. However, it is possible to derive some general views and mechanisms that illustrate the likely performance under such conditions.

Rapid heat rise in concrete causes evaporation of free and physically bound water and, at higher temperatures, moisture loss by dehydration of cement hydrates. If the permeability of the concrete is insufficient to allow an adequate rate of dissipation then the vapour pressure in the pores of the concrete will rise. A contribution to the low apparent permeability-resisting vapour dissipation, is the vapour condensation further inside the concrete away from the fire. Once the vapour pressure rises to a critical level cracking and explosive spalling will occur.

This explosive spalling can occur after only a few minutes and rates are quoted in some reports of up to 3 mm/min for normal-weight aggregate concrete and up to 8 mm/min for lightweight aggregate concrete. More information is needed on these rates and the contribution from the various parameters.

Other types of spalling such as local spalling and sloughing-off (gradual reduction of a cross-section) that have been observed in cellulosic fires are possible but explosive spalling seems to be the dominant form in an extreme hydrocarbon fire.

The concrete subjected to high temperature will suffer loss of strength. The strength loss increases as the temperature increases and both the aggregate and the cement hydrates are affected. At the very high peak temperatures in a hydrocarbon fire the aggregate and cement hydrates may be completely destroyed.

Many factors have an influence on the performance of concrete in hydrocarbon fire but those with a primary influence are:

  • the rate of temperature rise in the concrete

  • the moisture content of the concrete

  • the permeability of the concrete

These factors are interlinked but there is some evidence to suggest that, at least for low-permeability concrete, sufficient vapour pressure for damage to occur can be generated by decomposition of cement hydration products alone, even where there is little or no free water within the concrete pores. This would mean that indoor concrete would never dry sufficiently for spalling not to be a problem and, that self-desiccation in concrete with a very low water/cement ratio would not alleviate the problem.

The definition of satisfactory performance will be dependent on individual circumstances. Nevertheless, it is likely that the very high rate of temperature rise in a hydrocarbon fire will cause explosive spalling and loss of section at a high rate, particularly in high-strength concrete and lightweight aggregate concrete. Reinforcement could thus be exposed to high temperatures in less than approximately 20 minutes depending on depth of cover and other factors.