Contributions Of Human Errors To Uncertainties In Radiation Measurements And Implications For Training

INTRODUCTION Several major factors introduce uncertainties into the assessment of radon progeny exposure to miners using time-weighted average radon progeny concentrations: uncertainty in the measurement of radon progeny concentrations in specific areas, assignment of an individual miner's time to those areas, variation in radon progeny concentration between measurements and potential human errors involved in calculating concentrations and handling data. The currently available grab-sampling methods for determining working level were analyzed to determine the magnitude of the uncertainty due to each of these factors. For all measurement methods studied, the variation in the airborne concentration with time in operational areas of a mine is the dominant factor in the uncertainty in determining annual radon progeny exposures for individual miners. Uncertainties relating to accuracy of the method and precision of measurement were found to contribute a significantly greater portion of the total uncertainty than human errors. Under normal conditions, if the technicians performing the measurements are conscientious and well trained, human error contributes little to the total uncertainty of the radon progeny exposure determination. The primary goal of radiation monitoring is the reduction of radiation exposure to the lowest reasonably achievable level below regulatory limits. Monitoring personnel in mines should be trained not only to obtain accurate estimates of miner radiation exposures but also to recognize and, when possible, to implement correction of situations which result in unnecessarily high radon progeny exposures. ESTIMATION OF UNCERTAINTY DUE TO HUMAN ERRORS Human errors affecting the assignment of annual radon progeny exposure to individual miners can be placed in two categories: those related to the measurement of radon progeny concentration in specific mine areas and those related to estimation of occupancy time for individual miners and transcribing data to permanent records. The former are specific for the measurement method used; the latter are common to all methods. Errors in Determination of Working Level All systems for determining radon progeny concentration require measurement of several parameters, which include volume of air sampled, count rate and decay time. These quantities and appropriate constants are used in a basic equation, specific to the system, which estimates working level. An unintentional random mistake in measurement of any one of these parameters or in the selection of proper constants will contribute to the uncertainty in the determination of working level. In our analysis of human error we separated each measurement method into a sequence of independent operations, with each step subject to operator error. For each operation we estimated the probability of occurrence and the consequence of errors to obtain a resulting uncertainty. Certain types of errors result in specific consequences. For example, we assumed that an error of 5 seconds in timing of a 5-minute sample results in a fractional error of 1/60 (1.7%). Other types of errors can result in a range of uncertainty. Transposing digits read from a scaler can produce errors ranging from near zero to approximately 60%. In these cases we calculated the statistical variance for the distribution of errors. We assigned the square root of the variance divided by the mean as the consequence factor for that type of error. This is essentially the same as a coefficient of variation. The product of the probability of occurrence and the consequence factor is the fractional uncertainty in the measurement due to that particular error. The total uncertainty due to human errors is calculated by taking the square root of the sum of the squares of the uncertainties generated by all manual operations. Uncertainties Due to Human Error for the Kusnetz Method One of the techniques most commonly used to estimate working level in U.S. uranium mines is the Kusnetz method. A generalized way to express the equation used to compute WL by this method is: WL = (Net Alpha Counts)/(V)(ST)(CT)(E)(K) where: V = sample flow rate in liters/min ST = sampling time in min CT = counting time in minutes E = absolute counting efficiency K = Kusnetz conversion factor (dis/min-L per WL), as a function of decay time in minutes. The example of human error analysis presented here is based on the Kusnetz procedure having a timing sequence of 5 minutes sampling time, 40 minute decay time, 2 minute counting time. During the sampling procedure a stop watch is used to determine the timing interval. We assume that it is common to make small timing errors of a few seconds, but larger timing errors occur infrequently. Errors greater than 30 seconds are considered to be essentially non-existent since we assume that the