ISO/TS 29041:2008 pdf download – Gas mixtures — Gravimetric preparation — Mastering correlations in composition.
1 Scope In this Technical Specification, the gravimetric mixture preparation as given in ISO 6142 is investigated for influences of a priori existing, as well as correlations introduced by data processing. All calculations refer to an example which consists in the preparation of a synthetic natural gas of a target composition as follows: 1,4 mol % N 2 , 1,8 mol % CO 2 , 9,4 mol % ethane, 3,4 mol % propane, 1 mol % n- butane, and 83 mol % methane. All considerations given for this example concerning mixture feasibility, choice of preparation procedure, and weighing steps and sequences are the same as given in ISO 6142. This also applies to all estimates for basic uncertainty sources and the purity tables of the gases used for preparation. All calculations follow the principles, and use the tools and algorithms laid down in Annex A. For the sake of simplicity, procedural steps such as matrix transformation, inversion or matrix calculus are not detailed each time they are used in the calculations. 2 Normative references The following referenced documents are indispensable for the application of this document. For dated references, only the edition cited applies. For undated references, the latest edition of the referenced document (including any amendments) applies. ISO 6142, Gas analysis — Preparation of calibration gas mixtures — Gravimetric method
The following reasoning applies to possible correlations: it is assumed that u m is a random variable governed by a common distribution, and each realisation is a drawing from the distribution. Under this assumption, there is no reason for making allowances for correlation(s) since the realisations are independent. The uncertainty sources for the u B estimates are the same, namely pressure, temperature, and humidity. This causes correlation, but the common sources are not quantified in ISO 6142. There is no reason for assuming correlations for the u exp estimate, and u R occurs only once for methane. For m m , a clear correlation exists since the same mass pieces are used, but their combinations are unknown (except for some cases). Usually, these correlations are small and may be neglected in practice. Here, for demonstrating the principle of the method they are included where obvious. The corresponding variance-covariance matrix for the data table is given in Table 2.