BS 7813-1995 Guide for statistical control of dynamic liquid hydrocarbon volumetric metering systems《液态烃类容量计系统统计控制指南》.pdf

《BS 7813-1995 Guide for statistical control of dynamic liquid hydrocarbon volumetric metering systems《液态烃类容量计系统统计控制指南》.pdf》由会员分享，可在线阅读，更多相关《BS 7813-1995 Guide for statistical control of dynamic liquid hydrocarbon volumetric metering systems《液态烃类容量计系统统计控制指南》.pdf（84页珍藏版）》请在麦多课文档分享上搜索。

1、BRITISH STANDARD BS 7813:1995 ISO 4124:1994 Guide for Statistical control of dynamic liquid hydrocarbon volumetric metering systemsBS7813:1995 This British Standard, having been prepared under the directionof the Sector Board for Materials and Chemicals, was published under the authority ofthe Stand

2、ards Board and comesinto effect on 15June1995 BSI 09-1999 The following BSI references relate to the work on this standard: Committee reference PTI/12 Draft for comment88/52423 DC ISBN 0 580 24139 4 Committees responsible for this British Standard The preparation of this British Standard was entrust

3、ed to Technical Committee PTI/12, Petroleum measurement and sampling, upon which the following bodies were represented: Department of Trade and Industry (National Engineering Laboratory) Department of Trade and Industry (Oil and Gas Office) Institute of Petroleum Salvage Association The following bo

4、dy was also represented in the drafting of the standard, through subcommittees and panels: GAMBICA (BEAMA Ltd.) Amendments issued since publication Amd. No. Date CommentsBS 7813:1995 BSI 09-1999 i Contents Page Committees responsible Inside front cover National foreword iii Section 1. General 1.1 Sc

5、ope 1 1.2 Definitions 1 1.3 Symbols and units 1 1.4 Central proving 2 1.5 On-line proving 2 Section 2. Statistical measurements 2.1 Principles of statistical measurement 3 2.2 Measurement procedure 6 Section 3. Central proving 3.1 Collection of data 9 3.2 Reliability of data collected and resulting

6、values 10 3.3 Performance charts 11 3.4 Control charts and tests 18 3.5 Worked examples 21 Section 4. On-line proving 4.1 Collection of data 45 4.2 Reliability of data collected 47 4.3 Performance charts 47 4.4 Control charts 48 4.5 Worked examples 52 Section 5. Secondary control 5.1 Comparison betw

7、een meter and tank 63 Annex A (informative) Statistical tables 67 Annex B (informative) t-distribution values for95% and99% probability (two-sided) 70 Annex C (informative) Normal (Gaussian) distribution 71 Annex D (informative) Outlier tests 73 Annex E (informative) Random uncertainty of polynomial

8、 approximation 75 Annex F (informative) Bibliography 75 Figure 1 Performance chart: K-factor versus flowrate showing scatter (range) of5 to10 consecutive runs 8 Figure 2 Control chart (general) 8 Figure 3 Performance chart No.1 13 Figure 4 Performance chart No.2 14 Figure 5 Performance chart No.3, E

9、 = f (Q) 14 Figure 6 Performance chart No.4 UCC polynomial 15 Figure 7 Control chart No.1 19 Figure 8 Control chart No.2 19 Figure 9 Control chart No.3 20 Figure 10 Control chart No.4 20 Figure 11 Double chronometer method of pulse interpolation 24 Figure 12 Performance chart, turbine No.310,1978 28

10、 Figure 13 Log (Q/v) vs. meter factor, turbine No.310,1978 29 Figure 14 Performance chart, turbine No.310,1979 33 Figure 15 Log (Q/v) vs. meter factor, turbine No.310,1979 34BS7813:1995 ii BSI 09-1999 Page Figure 16 Performance chart, turbine No.310,1980 38 Figure 17 Log (Q/v) vs. meter factor, turb

11、ine No.310,1980 39 Figure 18 Comparison of1978 data with1979 data, log (Q/v) vs. meter factor, turbine No.310 43 Figure 19 Comparison of1979 data with1980 data, log (Q/v) vs. meter factor, turbine No.310 44 Figure 20 Permissable change in flowrate during proving 46 Figure 21 Performance chart from s

12、eries of on-line provings 48 Figure 22 K-factor (mean of5 or10 consecutive runs) versus time 49 Figure 23 Combined performance/control chart 51 Figure 24 Moving-average meter factor versus time 52 Figure 25 Control chart for meter operating within linear range 56 Figure 26 Chart A Meter performance

13、curves Meter turbine provedwithpipe prover 59 Figure 27 Control chart Meter operating outside linear range 61 Figure C.1 Frequency histogram 72 Figure C.2 Bell-shaped distribution curve 72 Table 1 Performance chart No.5 Direct matrix 16 Table 2 Performance chart No.6 17 Table 3 Example5: test result

14、s for turbine meter 25 Table 4 Performance chart, meter No.310 test report,1978 30 Table 5 Meter factor for turbine meter No.310,1978 31 Table 6 Random uncertainty of polynomial for turbine No.310,1978 32 Table 7 Performance chart, meter No.310 test report,1979 35 Table 8 Meter factor for turbine me

15、ter No.310,1979 36 Table 9 Random uncertainty of polynomial for turbine No.310,1979 37 Table 10 Performance chart, meter No.310 test report,1980 40 Table 11 Meter factor for turbine meter No.310,1980 41 Table 12 Random uncertainty of polynomial for turbine No.310,1980 42 Table 13 Example4 Normalizin

16、g meter factor 58 Table 14 Computed uncertainty for vertical cylindrical tanks at15 C 65 Table 15 Computed meter uncertainty at15 C 65 Table 16 Combined values of transfer uncertainty for tank and meter 66 Table A.1 Distribution of the range 67 Table A.2 Limiting values E 2(n, ), P =0,95 68 Table A.

17、3 Limiting values E 2(n, ), P =0,99 69 Table B.1 t-distribution values 70 Table D.1 For use with Dixons Test for outliers 73 Table D.2 For use with Grubbs Test for outliers 74 List of references Inside back coverBS7813:1995 BSI 09-1999 iii National foreword This British Standard has been prepared by

18、 Technical Committee PTI/12. It is identical with ISO4124:1994, Liquid hydrocarbons Dynamic measurement Statistical control of volumetric metering systems, published by the International Organization for Standardization (ISO). ISO4214 was prepared by Subcommittee2, Dynamic petroleum measurement, of

19、Technical Committee, Petroleum products and lubricants, in which the United Kingdom participated. A British Standard does not purport to include all the necessary provisions of a contract. Users of British Standards are responsible for their correct application. Compliance with a British Standard do

20、es not of itself confer immunity from legal obligations. Cross-references International Standard Corresponding British Standard BS6866 Proving systems for meters used in dynamic measurement of liquid hydrocarbons ISO7278-3:1986 Part3:1987 Methods for pulse interpolation (Identical) Summary of pages

21、This document comprises a front cover, an inside front cover, pagesi to iv, pages1to76, an inside back cover and a back cover. This standard has been updated (see copyright date) and may have had amendments incorporated. This will be indicated in the amendment table on the inside front cover.iv blan

22、kBS7813:1995 BSI 09-1999 1 Section 1. General 1.1 Scope In dynamic measuring systems the performance of meters for liquid hydrocarbons will vary with changes in flow conditions, viz. flowrate, viscosity, temperature, pressure, density of product, and with mechanical wear. This International Standard

23、 has been prepared as a guide for establishing and monitoring the performance of such meters, using appropriate statistical control procedures for both central and on-line proving. These procedures may be applied to measurements made by any type of volumetric or mass metering system. The procedures

24、to be followed for collecting data, on which the control limits are based, are described. An alternative method for establishing the reliability of these data is described in ISO7278-3. Methods are described for calculating the warning and action control limits for the charts covering the selected p

25、erformance characteristics, the application of these control charts to subsequent routine measurements, and their interpretation. Worked examples are given in the appropriate central and on-line proving sections. 1.2 Definitions For the purposes of this International Standard, the following definiti

26、ons apply. 1.2.1 proving; proof; calibration determination of the meter performance via the relationship between the volume of liquid actually passing through a meter and the reference volume of the pipe prover 1.2.2 K-factor relationship between the number of pulses (N) generated by the meter durin

27、g the proving run and the volume of liquid (V) displaced by the sphere or piston in the pipe prover between detectors normally, K = N/V; it is recommended that this value be corrected by the pulse interpolation technique described in ISO7278-3 1.2.3 meter factor ratio of the actual volume passed thr

28、ough a meter, as derived from the pipe prover, to the volume indicated by the meter totalizer 1.3 Symbols and units 1.3.1 General symbols h 1 high liquid level in tank metres h 2 low liquid level in tank metres E h gauging error millimetres E m meter volumetric error percent E t temperature error de

29、grees Celsius K K-factor pulses per unit volume %K change in K-factor pulses per unit volume MF meter factor dimensionless MF m mean meter factor dimensionless MF max maximum meter factor in a set of measurements dimensionless MF min minimum meter factor in a set of measurements dimensionless N numb

30、er of pulses generated by meter during proving run dimensionless P pressure at line conditions kilopascals(1bar=100kPa) P 0 pressure at standard conditions(101,325kPa) kilopascals t temperature at line conditions degrees CelsiusBS7813:1995 2 BSI 09-1999 1.3.2 Statistical symbols 1.4 Central proving

31、With the method of central proving, the performance of a meter is established at a testing station by proving the meter over its entire operating range of flowrate, viscosity, temperature and oil density used in service. Meter performance charts are then prepared from the proving data, and are used

32、to establish the relationship between the meter factor and flowrate or flow and viscosity. Any large deviation in meter performance on site can be detected by secondary control procedures, which monitor the output of two meters in series or in parallel. Long-term deviations in meter factors can be e

33、stablished by statistical control charts. The latter method can also be used in on-line proving. 1.5 On-line proving With the method of on-line proving, the meter is proved under operating conditions with a portable or fixed installation pipe prover. Where significant changes in flowrate, viscosity,

34、 temperature or density occur, the meter can be reproved. Any marked deviation or abnormal trend in meter factor can be monitored by use of statistical control charts. By statistical analysis it is possible to establish whether the deviations are due to changes in flow conditions, random error or so

35、me other assignable cause. t 0 temperature at standard conditions(15 C or20 C) degrees Celsius T 1 elapsed time seconds Q volume rate of flow cubic metres per hour V p reference volume of pipe prover at standard conditions(15Cor20 C and101,325kPa) litres or cubic metres v kinematic viscosity of the

36、fluid millimetres squared per second centistoke (cSt) X true value of quantity mean value standard deviation x value of measurement mean of a set of measurements n number of repeated measurements m number of quantities s estimate of standard deviation w range of a set of measurements mean of a set o

37、f ranges t value of Students t-distribution r estimate of repeatability degrees of freedom x wBS7813:1995 BSI 09-1999 3 Section 2. Statistical measurements 2.1 Principles of statistical measurement 2.1.1 Introduction Measurements taken via central or on-line meter proving provide information on the

38、random variability of the parameters of hydrocarbon flow through the meter (for example meter factor, flowrate, temperature, Reynolds number). Using this information, it is possible to assign a level of probability to a deviation observed in practice, and thereby differentiate between a normal or “a

39、llowable” deviation and one that has been caused by an external and systematic influence, such as meter component wear. The true value of the meter characteristic in question, and its range of variability, can be represented diagrammatically on a control chart (see2.2.5). This will indicate the devi

40、ation (warning limit) which should be taken as an early indication of malfunction, and the deviation (action limit) at which it is almost certain that meter failure has occurred. It is standard practice to assign a probability of95% to warning limits, and99% to action limits. This means, for example

41、, that there is only a1% chance that a measurement falling outside the action limits did so as a result of normal variation when the process is under statistical control. Once a control chart is established, the measurements from subsequent meter provings can be entered periodically onto the control

42、 chart, from which it is possible to monitor trends in meter performance over a period of time. In order to establish control through this means, reliable estimates should be obtained of the statistics to be used. The initial period in which data is collected, and against which the performance of th

43、e meter is to be monitored, is called the “learning period”. This should be long enough to provide a reliable assessment of the true value of the meter characteristic in question. Before considering the steps to be followed in the creation, use and maintenance of control charts, it is first necessar

44、y to understand the statistical treatment which is to be applied. 2.1.2 Distribution of measurements The measurement of any physical quantity, be it direct (for example temperature by thermometer) or indirect (for example meter factor) is always subject to error. The error is sometimes systematic an

45、d assignable to a definite cause, for example a large change in temperature may result in a large change in meter factor. If that is not the case, however, data scatter can be regarded as random, and is thus amenable to statistical treatment. Random errors often vary in magnitude with the quantity b

46、eing measured (in which case they are expressed as percentages) or with some other external factor. The error in K-factor, for example, will change in magnitude according to the flowrate (see performance chart in Figure 1). For this reason it is vital that operating conditions are controlled while m

47、easurements are being taken (see2.2.2). In practice, the distribution of errors approximates a Gaussian (normal) distribution, and this is fully defined if its two parameters are known. The parameters in this case are mean value, represented by , and standard deviation, represented by . The Gaussian

48、 distribution is described in more detail in Annex C. Each of the parameters of a distribution of measurements is assumed to have a true value, and is represented algebraically by a Greek or capital Roman letter. Estimates of the parameters, or statistics, are represented algebraically by small Roma

49、n letters. When necessary these will be qualified algebraically by the use of brackets. For example the standard deviation estimate of a measurement x will be shown as s(x) (see2.1.4). The statistics which are of primary interest are mean, standard deviation, range of a set of measurements, and uncertainty. 2.1.3 Estimate of true quantity Given a set of measurement x i , for i=1 to n, the estimate of the true quantity which is most likely