Abstract
The fluctuation-induced error in estimates of time-weighted-averages (TWA) of fluctuating concentrations is analyzed theoretically for a specific type of diffusive monitor in which the concentration is close to zero at the sorbent. Sampler calibration which results in unbiased TWA estimates is discussed. The error is found to have a lower, though no upper, bound. The variance of the error averaged over a variety of sampling start times (with fixed sampling period) is presented in terms of a sampler response function and the autocorrelation function of the concentration. A logarithmic dependence on pulse width is found for the variance of the error in the special case of concentration pulse sampling and indicates that significant error can be present in estimates of averages of rapidly varying concentrations. Similar conclusions are reached from numerical computation of the error and its variance using time-dependent field concentration data. The results indicate that excessive errors in exposure estimates can be expected in short time Palmes tube sampling of fluctuating concentrations, such as in the measurement of substances with short term exposure limits (STEL). A simple sampler modification is suggested for reducing or eliminating this error due to concentration fluctuations. As conclusions are based on calculation alone, suggestions as to their experimental verification are given.