PBD의 신뢰성 향상을 위한 FDS의 연소모델 및 입력조건에 대한 예측성능 검토 연구

Abstract

In this study, the prediction performances of combustion model and input value of Fire Dynamics Simulator (FDS) were evaluated to improvement reliability of Performance Based Design (PBD). Combustion models were investigated about Mixing Controlled Fast Chemistry 1-step (MCFC 1-step), Mixing Controlled Fast Chemistry 2-step (MCFC 2-step), Mixed 3-step, Eddy Dissipation Model 1-step (EDC 1-step) and Eddy Dissipation Model 3-step (EDC 3-step). And input value were investigated about design fire curve and B-RISK. The prediction performance for combustion models of FDS version 6 were investigated for propane gas fire in a compartment. Comparing numerical results with experimental data, it was found that all of combustion model reasonably predicted for experimental data for Heat Release Rate (HRR) and temperature distribution. Also numerical result well predicted for O2 concentration of experimental data but for differently predicted for CO2 and CO concentrations. For CO2 concentration, combustion models can well predict for experimental data except EDC 3-step. The reaction mechanism of EDC 1-step not consider about CO concentration so EDC 1-step can’t predict CO concentration. And EDC 3-step and MCFC 1-step respectively over-predicted and less-predicted than experimental data for CO concentration. But Mixed 3-step and MCFC 2-step can reasonably predict with experimental data. The prediction performance of design fire curve was validated for solid fuel fire in a building. Design fire curves used that Two-stage design fire curve (TDF) proposed in this study and Exponential design fire curve (EDF) and Quadratic design fire curve (QDF) suggested by Ingason. In the numerical study, window open slightly at 210 s and then totally open at 270 s because the window was broken by fire expansion during experiment. And MCFC 2-step was used for combustion model. Comparing with numerical results and experimental data, TDF and QDF sufficiently predict the experiment data for temperature distribution than EDF. However, numerical results were difficult to predict experimental data for species concentrations. This is not implies limitation of the MCFC 2-step because in the numerical study was consider about the window opening condition. The prediction performance of B-RISK was evaluated for fire propagation with experimental data for one- and two-clothing sets. The HRR for one-clothing set and design fire curve was used for input value of B-RISK to calculate for HRR for two-clothing set. Comparing with B-RISK results and experimental data for HRR, percentile of 100% of B-RISK results were over-predict for total HRR but less-predicted for fire growth rate than measured data. And when using experimental data for input value, B-RISK results over-prediction than experimental data for maximum HRR but when using design fire curve for input value, B-RISK results less-predict than experimental data for maximum HRR. And the prediction performance of FDS using input value for B-RISK results was evaluated. Comparing with FDS results with experimental data, it shows that FDS results less-predict for maximum temperature of experimental data. But comparing with FDS results for each input value, it was found that FDS results when used percentile of 70% of B-RISK results predict similarly with FDS results when used percentile of 100% of B-RISK results. This implies that the B-RISK result for percentile of 70% can sufficiently predicted for fire behaviour of B-RISK results for percentile of 100%.