Nuclear data uncertainty information

The assessment of the uncertainties in the prediction of reactor properties is an important requirement, both for economic and safety reasons. Allowances must be made in the design and operation of reactors to cover uncertainties by introducing suitable margins. A very high level of confidence in the safety aspects of reactors must be achieved. Uncertainties in nuclear data contribute to the overall uncertainty. Consequently the uncertainties in the data and the sensitivity of calculated reactor parameters to these uncertainties must be estimated. The estimation and representation of uncertainties in evaluated differential cross-sections is a complex problem. However, since the required information is the imcertainty in reactor neutron spectrum averaged values of cross-sections, or in ratios of such averages, the required 

---

The nuclear equipment failure rate database has not changed markedly since the RSS and chemical process data contains information for non-chemical process equipment in a more benign environment. Uncertainty in the database results from the statistical sample, heterogeneity, incompleteness, and unrepresentative environment, operation, and maintenance. Some PSA.s use extensive studies of plant-specific data to augment the generic database by Bayesian methods and others do not. No standard guidance is available for when to use which and the improvement in accuracy that is achieved thereby. Improvements in the database and in the treatment of data requires, uhstaiui.il indu.sinal support but it is expensive. 

The significance of these predictions is shown in Table XV. If the nuclear data are adjusted in line with this experimental information, the design calculations show that at least 50% of the BeO, which was included in the core to soften the neutron spectrum and to increase the Doppler coefficient, can be removed and still meet the Doppler and sodium loss criteria of the reference core. The fissile inventory required decreases by about 3%, and the breeding ratio increases from 1.18 to about 1.25. This results in a decrease in the fuel cycle cost of about 0.1 mill/kW-hr. If one assumes a favorable combination of nuclear data within the limits of uncertainties reported by Greebler (5) and, furthermore, if the safety criteria are relaxed to allow a calculated Doppler effect (T dkjdT) of — 0.004 (with sodium out) and a positive total sodium loss reactivity effect between 1 and 2, all of the BeO can be removed... 

Although some experiments can be used as benchmarks for both areas of application, in general the requirements of the two categories are quite different. Nuclear data benchmarks consist of simple-geometry, simple-configuration experiments dependent on as few components (cross sections) as possible. Such experiments are often designed to accentuate sensitivities to cross-section features of interest. Their accuracy requirements are very high, since to provide usefUl information, the experimental uncertainties must be lower than the combined uncertainties of the nuclear data and methods of analysis. 

The nuclear data libraries used in different countries result in different uncertainties in reactor parameters caused by nuclear data. But the magnitudes of these uncertainties are to some extent similar for most current nuclear data libraries because they are based on similar experimental information. So the comparison of target accuracies and uncertainties, provided by the ABBN data set, is of general interest. Some results of an analysis based on ABBN [4.21] are given in Table 4.1. 

Estimation of Risk. The estimation of risk contains uncertainties, based on the lack of specific data (such as exposure information) and/or the lack of understanding of the mechanism of toxic action of a compound. Between the extremes of acturial risk, which is based on enough information that "time has removed the uncertainty," such as the probability of death as cited in an insurance table, and theoretical risk, which is based on probabilistic calculations of events which have never actually occurred (e.g., nuclear "winter" (7)) lies a wide continuum into which most estimates of human health effects fall. In real-life situations, many assumptions are made in evaluating risk in order to make a conclusion, and these assumptions lead to uncertainties in the final result. These uncertainties should be understood as limitations to the best guess science can presently make. Although one response to this uncertainty, in the face of an outcome as fearsome as cancer, is to deny that there is a lack of certainty, the more reasonable response is to try to estimate the uncertainty, making it clear that any estimate is bracketed by these possible errors. 

---