scope:

INTRODUCTION

Cleanrooms utilize about 5 to 50 more times airflow rates than for general-purpose buildings. Utilizing high volume airflow has been mainly to meet the old federal standard FS-209 (versions A through E) and the recommendation by IEST's RP-CC012 (versions 1 and 2) since 1970s. Since then many published reports have indicated that cleanroom filtered air over-supply is a common practice which causes significant energy waste (Mills et al. 1996, Jaisinghani 2001).

The recommended guideline (tables) was based on old experience, in which air change rate was arbitrarily determined "only" based on room cleanliness class, disregarding a room's actual particle generation rate (internal generation and external intrusion) and other factors. Due to lack of an accurate theoretical model and related research, this obsolete guideline is still being used today. For cleanrooms with lower particle generation rates, lower-than-recommended air change rates (up to 20% reduction) have been practiced. However most of design and operating engineers still choose to obey the existing guideline to avoid being challenged. Establishment of a more accurate model supported with validations is a key to respond the challenge and to reduce cleanroom fan energy waste.

The fundamental airflow model in cleanrooms is the mathematical relationship between the air change rate and the room airborne particle concentration. In last a few decades, several mathematical models were proposed by Morrison (1973), Brown et al. (1986), Kozicki et al. (1991) and Jaisinghani (2001), however a common shortcoming of these previous models was over simplification due to ignoring many critical elements and lack of experimental validations, these models could only be used as qualitative indication, but not as a quantitative tool to calculate the required air change rate to meet a room air cleanliness class based on the room's specific airborne particle load, see Table 1. The main project objective was to establish a new model which is more descriptive (includes more variables and parameters), and more accurate than existing models.