Based on Bayesian statistics and the Bayesian theory of measurement uncertainty, characteristic limits such as the decision threshold, detection limit and limits of a confidence interval can be calculated taking into account all sources of uncertainty. This approach consists of the complete evaluation of a measurement according to the ISO Guide to the Expression of Uncertainty in Measurement (GUM) and the successive determination of the characteristic limits by using the standard uncertainty obtained from the evaluation. This procedure is elaborated here for several particular models of evaluation. It is, however, so general that it allows for a large variety of applications to similar measurements. It is proposed for the revision of those parts of DIN 25482 and ISO 11929 that are still based on conventional statistics and, therefore, do not allow to take completely into account all the components of measurement uncertainty in the calculation of the characteristic limits.

It is shown how the decision threshold, the detection limit and the limits of a coverage interval - summarily called the characteristic limits - and, in addition, the best estimate and the associated standard uncertainty of a non-negative radiation measurand are to be calculated by using the Monte Carlo (MC) method in ionising-radiation measurements. The limits are mathematically defined by means of quantiles of the Bayesian distributions of the possible measurand values. The MC-induced uncertainties of the limits and typical problems connected with MC application are also treated. The paper can serve as a bridge between the ISO Guide to the Expression of Uncertainty in Measurement (GUM), Supplement 1 applying the MC method and ISO/FDIS 11929 (at present in preparation) dealing with the characteristic limits. As an illustration, a net count rate measurand, being the difference of a gross and a background count rate, is treated theoretically and numerically. More complex examples deal with the wipe test for surface contamination and with a linear multi-channel spectrum unfolding.

As part of the second phase of vibrational/earthquake investigations at the HDR (Heissdampfreaktor) Test Facility in Kahl/Main, FRG, high-level simulated seismic tests (SHAM) were performed during April-May 1988 on the VKL (Versuchskreislauf) in-plant piping system with two servohydraulic actuators, each capable of generating 40 tons of force. The purpose of these experiments was to study the behavior of piping subjected to seismic excitation levels that exceed design levels manifold and may result in failure/plastification of pipe supports and pipe elements, and to establish seismic margins for piping and pipe supports. The performance of six different dynamic pipe support systems was compared in these tests and the response, operability, and fragility of dynamic supports and of a typical U. S. gate valve were investigated. Data obtained in the tests are used to validate analysis methods. Very preliminary evaluations lead to the observation that, in general, failures of dynamic supports (in particular snubbers) occur only at load levels that substantially exceed the design capacity. Pipe strains at load levels exceeding the design level threefold are quite small, and even when exceeding the design level eightfold are quha tolerable. Hence, under seismic loading, even at extreme levels and in spite of multiple support failures, pipe failure is unlikely.

No abstract

No abstract