Date of Graduation


Document Type



This study investigated the relationship between industrial respirator half-facepiece fit and facial anthropometry, including three-dimensional parameters. The investigation included application of laser scanning technology to the measurement of faces. The goal was to identify facially-related variables that are worthy of further investigation. The relationships between facial features and respirator dimensions are of significance to the industrial respirator community (manufacturers, testers, and users) because the faceseal leakage affects the level of protection provided by the respirators to workers; factors that predict facepiece fit are therefore of interest for the design of respirators, the testing of respirators, and the selection of respirators for use in a workplace. The investigation was accomplished by: (1) obtaining facial dimensions from a panel of human subjects with a selected distribution of facial sizes, using two different techniques for facial measurement; (2) obtaining facepiece fit measurements for the same panel of subjects while wearing four different respirators, with three sizes each, with repeated donnings used to assess fit for each respirator; (3) analyzing the resulting data for relationships between the facial dimensions, respirator dimensions, and respirator fit. The measures of facial dimensions employed were (a) traditional manual anthropometric techniques (calipers and tape measure), and (b) an automated three-dimensional laser scanner. The respirator facepiece fit was evaluated using an ambient aerosol quantitative fit testing device. The study observed small but consistent differences between traditional and three-dimensional scan extracted facial measurements. Linear stepwise regressions of the measurements on the dependent variable (respirator fit) yielded somewhat inconsistent relationships when analyzed by respirator manufacturer and size, as well as by manufacturer only. The several non-linear three-dimensional parameters employed in this study (nose area and several facial angles) had lower R-square values for regressions compared to the R-square values from regressions based on linear measurements by traditional or 3D scanned image methods. Similarly, two principal component analysis factors also had lower R-square values.