This research examined a heat transfer model to predict the activation time of a concealed sprinkler. Concealed sprinklers consist of two stages of activation. They include the release of cover plates from a recess housing and the breakage of the glass bulbs or melting of the solder links. The research analysis is divided into two sections. The first section includes the prediction of cover plate activation time (stage one) and the second section includes the prediction of glass bulb activation time (stage two). Each prediction result is compared with the experimental data conducted by Annable (2006) and Yu (2007).

A lumped heat capacity method is introduced to predict the activation time of the cover plate. This method has been used for predicting the activation time of a standard pendent exposed sprinkler. It is reasonable to apply this method by assuming they are flush with the ceiling. The analysis results are compared based on the percentage of predicted and measured uncertainties. A recommendation is provided for which method is appropriate to apply to predicting the cover plate activation time.

The proposed of using FDS5 simulations is to simulate the heat transfer to the sensing element (glass bulb only) within the recessed housing. The constructed simulation models comprises of ceiling within a compartment. The simulations of various sprinkler heads are performed to investigate any parameters that can potentially affect the activation time of the sprinklers.

To simulate the glass bulb, combined thermal properties including glass and glycerine are modified to account for the differences in mass. Prior to stage two analysis, the FDS5 simulation was tested to predict the activation time of a standard pendent exposed sprinkler. The results showed positive progress to carry onto the next analysis. In stage two analysis, the simulations are constructed with and without the presence of vent holes within the recess housing.

The combined activation time for concealed sprinklers show lack of solid predictions compared to the experimental data especially Yu experimental data.