Uncertainty analysis of NDA waste measurements using computer simulations
Uncertainty assessments for nondestructive radioassay (NDA) systems for nuclear waste are complicated by factors extraneous to the measurement systems themselves. Most notably, characteristics of the waste matrix (e.g., homogeneity) and radioactive source material (e.g., particle size distribution) can have great effects on measured mass values. Under these circumstances, characterizing the waste population is as important as understanding the measurement system in obtaining realistic uncertainty values. When extraneous waste characteristics affect measurement results, the uncertainty results are waste-type specific. The goal becomes to assess the expected bias and precision for the measurement of a randomly selected item from the waste population of interest. Standard propagation-of-errors methods for uncertainty analysis can be very difficult to implement in the presence of significant extraneous effects on the measurement system. An alternative approach that naturally includes the extraneous effects is as follows: (1) Draw a random sample of items from the population of interest; (2) Measure the items using the NDA system of interest; (3) Establish the true quantity being measured using a gold standard technique; and (4) Estimate bias by deriving a statistical regression model comparing the measurements on the system of interest to the gold standard values; similar regression techniques for modeling the standard deviation of the difference values gives the estimated precision. Actual implementation of this method is often impractical. For example, a true gold standard confirmation measurement may not exist. A more tractable implementation is obtained by developing numerical models for both the waste material and the measurement system. A random sample of simulated waste containers generated by the waste population model serves as input to the measurement system model. This approach has been developed and successfully applied to assessing the quantity of plutonium in a variety of waste types contained in 208-{ell} drums measured by the passive active neutron (PAN) radioassay system at the Idaho National Engineering and Environmental Laboratory (INEEL). Computer simulation of the PAN system performance uses the Monte Carlo N-Particle (MCNP) code to produce a neutron transport calculation for a simulated waste drum. A followup program was written to combine the MCNP output with other parameters generated by the modeling process to yield simulated measured plutonium mass values. The accuracy of the simulations is verified using surrogate waste drums with known contents.
- Research Organization:
- INEEL, Idaho Falls, ID (US)
- OSTI ID:
- 20093663
- Journal Information:
- Transactions of the American Nuclear Society, Vol. 82; Conference: 2000 Annual Meeting - American Nuclear Society, San Diego, CA (US), 06/04/2000--06/08/2000; Other Information: PBD: 2000; ISSN 0003-018X
- Country of Publication:
- United States
- Language:
- English
Similar Records
Uncertainty analysis of the SWEPP PAN assay system for glass waste (content codes 440, 441 and 442)
Uncertainty analysis of a nondestructive radioassay system for transuranic waste