ISO 10723:2012 pdf download – Natural gas -Performance evaluation for analytical systems.
The instrumental method may measure such a component as an individual chromatographic peak whichis typically backflushed through the system, and all components elute at the same time through thedetector. Alternatively, in systems where valve switching is not possible, the heavier hydrocarbonselute in a forward fashion through the columns and the component is simply measured as the sum ofindividual peaks.However, the system may be set up to measure all hexanes (Cgs) individually and thesummed peak Cz+ may be specified. This is often the case should the Cg+ amount be significant and amore detailed breakdown of this component be required to minimize errors on the measurement. Thisprinciple can be extended such that the system is set up to measure in a C6+,C7+,C8+, Cg: or even C10+mode. Users of this International Standard shall decide which of these components are to be included inthe evaluation of the instrument’s performance based upon the significance of the amounts of each ofthe components specified in the instrument set-up.
6.1.3Component content ranges
Once it is clear which measured components are going to be included in the evaluation, the user shalldetermine, for each of these, over what range of amount fractions the instrument is expected to be used.Such ranges shall generally be greater than that which is expected to be measured by the instrument inregular duty. If the data from the performance evaluation is used subsequently to update the responsefunctions assumed by the instrument, then it is vital that the component content ranges used in theevaluation extend beyond the specified operating range. Should this not be the case,considerablemeasurement errors might result from extrapolation outside the determined response function.
6.2 Response function types
6.2.1 Assumed functional descriptions
The instrument data system will assume a relationship between response and content of a component inthe gas.This is the assumed analysis function of the instrument, x =Gasm(y).Many instruments assumea simple first-order polynomial function in the form x= by,where by is often referred to as the responsefactor (RF) for that component. In this case a single calibration gas mixture is used and a first-orderresponse function is assumed, passing through the origin.Alternatively, the instrument may assume ahigher-order polynomial functional description or even an exponential or power function.
ln some cases the response, particularly for a minor component, may be calculated as relative to that ofanother (reference) component.Such a relative response factor shall have a response function similar tothat of the reference component.
The assumed analysis function for each component,x =Gasml(y), shall be noted and used for subsequentcalculation of the instrument’s performance characteristics described in 6.6.
The function types considered for the treatment of the performance evaluation data shall be matched tothose used by the instrument’s data system.
NOTE 0ccasion ally´╝îfunctional types other than polynomials´╝îsuch as exponential relationships, areimplemented by an instrument’s data system. Ilf the instrument uses functional types other than polynomials,it is appropriate to use these in the determination of the analysis functions.However, for the purposes of thisInternational Standard, only polynomial functions up to third order are considered.
6.2.2Selection of function types
The type of function to be used in practice is chosen according to the response characteristics of themeasuring system and that assumed by the instrument’s data system.
Polynomial functions describing the true response/amount fraction relationship can be derived ineither domain.A mathematical description of instrument response as a function of amount fraction istermed the calibration function, whereas that describing amount fraction as a function of response istermed the analysis function.
The response functions above are shown in a form up to third order.However, simpler forms up tosecond order or simply first order may be considered.Choose the form of the response functions withthe following considerations:
a)the simplest form that gives an adequate fit to the data should be used to avoid over-parameterizing the response function;
b)the number of calibration points, and hence the number of reference gases required to satisfactorily describe a polynomial, increases with the order of the function (see 6.4.2).