Computer simulation of particle overlap in fiber count samples

Fibrous aerosols are of great importance to industrial hygienists because o f the severe health risks that may be associated with inhaling such particl es. Previous studies on measurement error due to overloading of fibers and nonfibrous particles on the collected sample indicate that a 100-1300 fiber /mm(2) filter area is the best filter loading density to reduce bias in fib er counts. The present study investigated the upper fiber and particle conc entration limits for reliable counting and identification and the possibili ty of a procedure for correcting observed fiber counts to account for fiber masking due to overlapping particles or fibers. A computer-generated grid was used to simulate the light microscope graticule field. The resolution o f 2000 x 2000 was found to accurately represent the shape of the fibers and nonfibrous particles. Bivariate lognormal distributions were used to descr ibe the length and width distributions of the fibers. The capability of dis tinguishing particle-overlapped fibers (defined as the resolution index), t he coverage of the graticule field, the filter surface loading density, siz e distributions of fibers and particles, and the fiber-to-particle concentr ation ratio were the primary parameters in this study. The counting efficie ncy was found to consistently decrease with increasing filter surface loadi ng density and decreasing resolution index. The recommended upper limit of filter surface fiber density depended not only on the number concentration ratio but also on the filter surface loading densities and size distributio ns of fibers and particles. The advantage of using a thoracic preseparator on counting efficiency was calculated and found to improve counting efficie ncy significantly when the count median diameter of nonfibrous particles wa s close to or larger than the thoracic 50% cutoff point of 10 mum.