The increase in uncertainty throughout the viscosity scale being the principal disadvantage of capillary viscometry, the BNM-LNE decided to develop an absolute falling-ball viscometer making it possible to cover a wide range of viscosity while keeping a weak uncertainty. The measurement of viscosity then rests on the speed limit measurement of falling ball, corrected principal identified effects (edge effects, inertial effects, etc). An experimental bench was developed in order to reach a relative uncertainty of the order of 10 -3 to the measure of viscosity. This bench must allow to observe the trajectory of the ball inside a cylindrical tube filled with liquid for which the viscosity is to be measured, and to obtain the variations in speed throughout the fall in order to determine the area where the speed limit is reached.

Afin de réduire les incertitudes de mesure obtenues par viscosimétrie capillaire, nous avons développé un viscosimètre absolu à chute de bille. La caractérisation métrologique de ce banc expérimental est présentée pour une huile minérale de viscosité 30 Pa.s., en utilisant deux méthodes distinctes. Tout d'abord le calcul est effectué en suivant la démarche décrite par le GUM (Guide to the expression of Uncertainty in Measurement), c'est à dire, par propagation des variances. Ensuite, et parce que les techniques de propagations des distributions sont de plus en plus utilisées en métrologie, nous avons calculé les incertitudes de mesures en utilisant la simulation numérique de Monte-Carlo. Les deux résultats sont alors comparés pour un intervalle de confiances de 95 %. Finalement, les mesures de viscosité et les incertitudes obtenues avec le viscosimètre à chute de bille sont confrontées à celles obtenue avec un viscosimètre capillaire. Mots clés Viscosimètre à chute de bille, Incertitudes, Propagation des variances, Propagations des distributions, Monte-Carlo

Using techniques for the propagation of distributions to estimate uncertainties is becoming increasingly popular in metrology. Here we describe the calculation for the estimation of uncertainties in hygrometry using this method. This work is based on investigations on the two-temperature moist gas generator at Laboratoire National d'Essais.After a brief introduction on the calibration principles of hygrometers, models and calculation for estimating uncertainties using a Monte Carlo simulation are presented. Results concerning dew point temperature and relative humidity are shown, and compared with results obtained with the law of propagation of uncertainties (traditional method).The importance of the moist gas temperature measurement with regard to the saturator and the test chamber is confirmed. Moreover, results observed with the propagation of distributions and the traditional method are coherent.

Within the framework of the realization of a moist gas generator, a two-dimensional mass and heat transfer model has been established and then resolved. The impact of parameters such as flow rate of the gas and dew point temperature of the incoming gas was assessed. The realized moist gas generator was tested in a comparison made by means of a reference hygrometer.

The restrictions inherent in the use of the conventional reference in capillary viscometry, the viscosity of water at 20 C, have led to increased interest in absolute viscometers. The viscosity laboratory of the Bureau National de Métrologie, Laboratoire National d'Essais (BNM-LNE) has begun the study of an absolute falling-ball viscometer. The principal difficulties with this system are the measurement of the velocity of the ball and the corrections to compensate for the effects exercised on the ball. Results from a preliminary experiment indicate that uncertainties better than those of capillary viscometers can be obtained. The use of a velocity measurement video system should further reduce the uncertainties.