Development of techniques to improve fire dynamics informed loss estimation for cleanrooms in the semiconductor industry

Abstract

Our understanding of the behaviour of fire and smoke under common conditions as they occur e.g. in compartment fires has evolved in the last decades. This also took place due to progress made in computing capacity and Computational Fluid Dynamics (CFD) which is now also usable in the field of Fire Safety Engineering. However, this did not apply to cleanrooms where the unique pattern of recirculation airflow led to an incomparable spread of smoke. It is only recently that this technology can be applied to conduct fire research in these environments where real scale fire tests are too expensive and complex to carry out. The purpose of this thesis is to study cleanroom smoke spread by utilizing CFD and link these findings to loss estimations in the semiconductor industry. This approach would allow to improve the accuracy of these estimations and could also lead to the development of innovative smoke management strategies. A numerical 3D model was tested and rated beneficial in supporting smoke extraction by using cleanroom ventilation systems. Instead of studying the efficiency of “separate“ extraction systems, this approach focuses on the already installed clean air systems which usually facilitate the characteristic airflow. In the course of this thesis a novel method of smoke management was developed which is referred to as “buoyancy control“. Other than smoke exhaust venting equipment in common buildings which utilise smoke’s buoyancy for extraction, this new cleanroom strategy works reversely by deliberately decreasing buoyancy.