Heat and Mass Transfer Model for Fire-Exposed Concrete — Part B: Validation of Mass Loss and Temperature Predictions

Abstract Explosive spalling of concrete is an essential phenomenon to be accounted for when analyzing the fire performance of concrete structures. Existing tools of fire simulation can provide the fire exposure on concrete surface, and to some extent, conjugate heat transfer too, but new tools are needed for modelling the concrete high-temperature reactions coupled with heat and mass transfer. In this second part of the two-part article, we characterize two concrete mixtures regarding their high-temperature reactions using TGA and DSC experiments and measure their porosities, permeabilities, and thermal conductivities. We measure sample temperatures and mass loss under radiative heating in two different scales and at two different moisture levels. The experimental data is then used to validate the numerical model with a kinetics-driven evaporation rate. The model predicts mass loss with mean absolute percentage error (MAPE) typically $$\le$$ <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:mo>≤</mml:mo> </mml:math> 0.3 % and $$\le$$ <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:mo>≤</mml:mo> </mml:math> 1.9 % in the worst case. For temperature increase, typical MAPE $$\le$$ <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:mo>≤</mml:mo> </mml:math> 15 % and maximum MAPE $$\le$$ <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:mo>≤</mml:mo> </mml:math> 25 %. The model bias and relative standard deviation are +8.4 % and 3.7 % for mass loss, and +5.0 % and 4.4 % for temperature. We then introduce an improved version of the model with pressure-constrained evaporation rate. The model improves the pore pressure predictions, but maintaining the accuracy of mass loss and temperature predictions would require re-calibration of the kinetic parameters. A sensitivity study is used to identify future topics of model improvement and data gathering.