Comparisons for the 110 kW Corner Fire

Specific details of the temperature profile comparisons in Figures 9.20 – 9.36 for the 160 kW fire are discussed below.

Fire Room

The simulated temperature profiles in the lower layer generally match experimental profiles for trees 1 and 2. Lower layer temperature profiles for trees 3 and 4 are not as well simulated. Temperatures measured on tree 3, at heights of 600 mm and 900 mm of the experimental profiles, appear to be too high and are probably incorrect. This is more than likely a result of these thermocouples being affected by the experiments with the central fires (which were located directly under tree 3), as the temperatures at these heights are higher than any other tree. The simulated temperature profiles for the lower layer at tree 4 vary greatly, with the radiation simulation showing little or no constant temperature profile in the lower layer, whilst the simulation without radiation does. As mentioned earlier, temperatures measured in the lower layer are likely to be affected by radiation, thus increasing their temperature. Nevertheless, the simulations underestimate lower layer temperatures. Overall, the simulation with radiation predicts a higher temperature, giving a better result than the simulation without. The starting point for the temperature gradient of the corner fire between the upper and lower layer is not as clearly defined for some of the trees for the experimental profiles. It appears that the starting point for the temperature gradient for all trees between the upper and lower layer is underestimated by both simulations, with the radiation simulation predicting a slightly higher starting point for trees 1 and 3 than the non-radiation simulation. The reverse applies for trees 2 and 4. The profiles of the temperature gradients do not correspond well to the experimental profiles, particularly for the non-radiation simulations at trees 2 and 3. The experimental upper layer temperature profile of the corner fire is not constant as is seen with the centrally located fires (reasons are given in Section 5.5.2). In the upper section of the fire compartment (from about 1500 mm), both simulations predict an upper temperature profile where the curvature is opposite to that which is seen experimentally. Closer to the ceiling, experimental temperatures converge with the simulated, particularly for

the simulation with radiation. Here, the simulation without radiation produces a much higher temperature profile close to the ceiling.

Adjacent Room

Once again the simulated temperature profiles of the lower layer are constant, similar with what is observed for the experimental profiles. The temperatures are underestimated. Again, this is probably a result of the thermocouples absorbing radiation, and incorrectly increasing measured temperatures. There appears to be no significant temperature difference between simulations. Both simulations predict near identical starting points where the temperature gradient begins between the lower and upper layers. For all trees, except 6, the starting point of the temperature gradient is underestimated. Most of the profiles of the simulated temperature gradients correspond well with experimental observations; only the simulation profiles for tree 6 display a lower temperature gradient than what is observed experimentally. Generally the temperatures correspond well with experimental temperatures. However, for all trees temperatures very close to the ceiling are underestimated, more so by the radiation simulations than the non-radiation simulations.

Doorway

The temperature profiles for both simulations in the lower layer are constant with height and consistent with what is observed for the experimental profile. Temperatures are also very similar to the experimental results, with the small difference probably attributed to radiation increasing the experimental profile. For both simulations the temperature gradients between the lower and upper layers begin at almost identical heights, with the starting point for the temperature gradient for both simulations being slightly over predicted. The temperature gradient for both simulations follows a similar profile to that of the experimental, with little difference between the two simulations. The simulation without radiation predicts a higher temperature result nearer the soffit, with both simulations under predicting the temperature nearer the soffit.

9.5 Overall Discussion

Fire Room

For all simulations, the lower layer temperature of the fire room was underestimated. Without being able to quantify the effects of radiation on the thermocouples, the accuracy of the simulations can only be qualitatively discussed. The simulation including radiation performed better than the simulation without. The height at where the temperature gradient between the lower and upper layers begins is predicted well by both simulations, with modest differences between the two simulations and experimental results. The slope of the temperature gradient most tree locations is predicted with reasonable accuracy for by both simulations. Generally the temperatures along the temperature gradient between the upper and lower layers are close to what is seen experimentally, with the 110 kW corner fire simulations displaying the worst results. For the centrally located fires, experimental results revealed a constant temperature profiles, particularly for trees 1 and 2. This shows how the upper layer was uniform in temperature. Neither simulation predicts this well, with both simulations generating upper layer profiles with temperature gradients, often with very high temperatures close to the ceiling, indicating a ceiling jet. The simulation with radiation produced slightly better results in the upper layer. Overall, the simulation with radiation predicts the best profile in the adjacent room.

Adjacent Room

Temperature profiles by both simulations are predicted with more accuracy than the fire room. The lower layer temperature profiles show a constant temperature. This is close to what is seen experimentally, especially if the effects of radiation on the thermocouples are taken into account. The height where the temperature gradient between the lower and upper layer begins is predicted well by both simulations, often with little difference between them. The temperatures in the upper layer are also predicted well, however, temperatures are often under predicted closer to the ceiling.

Doorway

Temperature profiles in the lower layer are modelled well by both simulations. The height where the temperature gradient begins between the lower and upper layer

agrees with what is seen experimentally. The temperature gradient’s profile and temperatures for both simulations generally agree with the experimental profiles and temperatures. Nearer the soffit, the simulation without radiation over predicts the temperature for all but the 110 kW corner fire. Overall, the simulation with radiation produces the best results.

Overall

Capping off, running the simulations with the six-flux radiation sub-model simulates temperatures and temperature profiles more accurately than the simulation without the six-flux radiation model.