室內裝修材料之火災模擬

Abstract

本研究嘗試修正瑞典防火工程學者Dr. Bjorn Karlsson所發展出的火災模式，並且與兩種不同房間尺寸，五組天花板與牆面均為同一裝修材料之房間火災測試作比較，探討房間內壁裝材料之延燒現象。本研究以燃燒器後方牆面的細部引燃、修正天花板所受的熱幅射通量與上層氣體的有效熱傳係數，以及考慮房間熱層之幅射熱傳來改善Karlsson Model的缺失如初期熱釋放率的峰值、燃燒器輸出改變時的引燃延遲、天花板表面溫度和上層熱氣層溫度估算的誤差。將改善後的火災模式與實驗資料、Karlsson model的模擬值做一詳盡之比較。比較結果發現，在全尺寸房間火災測試上case 1 中程式模擬後期其火勢有衰退之現象，與實驗中火勢持續成長相反，這是因為木粒片水泥板會因高溫而變形，使火焰延燒至裝修板材背面造成額外之延燒並使火勢更加擴大，而最後甚至導致產生閃燃之現象，原因是因為程式本身僅考慮火焰在材料之單面延燒，並沒有考慮材料變形之機制所造成。而改善後的模式在case 2與case 3的模擬結果和Karlsson model 比較，除了在上層氣體溫度的模擬上前600秒會有低估之現象，與實驗值有較好之配合。在縮小尺寸之模型箱( case 4 與case 5 )的模擬上，則是發現在模擬上層氣體溫度有較Karlsson model有相當明顯之改善。

In this thesis, an extensive modification is made from the original Karlsson model [1]. The modifications include the division of material into many strips, utilization of the correlation for the convective and radiant heat transfer coefficient, increase of the incident heat flux to ceiling when the burner output is at the transition in full-scale fire tests, and incorporation of radiative heat exchange model in the calculation of hot gas temperature in upper layer. The predicted results are presented and compared with the experimental measurements of room fire tests [3~5], which consist of three full-scale room and two reduced-scale room fire tests. The shortcomings in Karlsson model, such as the initial peak and ignition delay appearing in heat release rate profiles and the inaccuracies of ceiling surface and upper layer gas temperatures, are improved by these modifications. From the comparisons between the fire tests and models, for full-scale tests both Karlsson and present models predict a decay of fire growth in later stage, whereas fire growth keeps growing in the experiment of case 1. It is because the particleboard was deformed, flame could spread into the backside of lining materials to increase the burning intensity and led to a flashover. There is no such mechanism in the present and Karlsson models, which only describe the flame spread along the surfaces. The present model shows a better prediction than that of Karlsson model in the cases lined with better fireproof materials, such as case 2 and 3, except the one in hot gas temperature before 600 sec.. In reduced-scale fire tests, the present model also performs better in case 4 and 5, especially in the prediction of hot layer gas temperature, whereas the corresponding one by Karlsson model is overshot.