Jin Lian Liu

Automated statistical analysis of high temperature and pressure vibrating tube densimeter data

Abstract A recently developed vibrating tube densimeter has been used to measure relative densities of aqueous sodium bromide from 373 to 523 K over a pressure range of 10–30 MPa. The instrument produces time period data consisting of both baseline and sample plateau regions. The number of samples injected into the instrument is known. The time period data have been analysed using a visual user defined baseline method and a computer program which performs an automated analysis of the data. The analysis program locates the transition from reference fluid baseline to sample solution plateau and back by testing the local slope against the null hypothesis. An adaptive algorithm is used to adjust the confidence level of the test to ensure that the correct number of samples are identified. A polynomial in temperature is fit to the baseline region surrounding each sample plateau datum to estimate the baseline at the time of measurement. Apparent molar volumes of aqueous sodium bromide have been calculated using both methods and compared to theoretical results obtained from Archer’s semi-empirical Pitzer Ion Interaction program. The statistical analysis program performs as well as the visual extrapolation method, completed by three different experimentalists, and outperforms individual experimentalists. Automated analysis allows the calculation of density errors which are shown to be better estimated than errors obtained from visual analysis. Furthermore, because the automated method can estimate errors from a single peak and the visual method requires two peaks for each sample, the need for replicate measurements is reduced.