Uncertainty Estimation In Natural Gas Flow Measurement and Practical Approach In Reducing It | Automation.com

# Uncertainty Estimation In Natural Gas Flow Measurement and Practical Approach In Reducing It

March 272011

By Chandu Bhatasana, Aker Solutions

Introduction: A statement of measurement uncertainty is crucial in evaluating the fitness for purpose of a gas measurement system for the given application. Generally, high accuracy metering system for fiscal measurement, medium accuracy system for allocation measurement and low accuracy system for operation measurement is employed. As the accuracy is qualitative term, the term uncertainty is used for quantitative evaluation of metering system for the given application.

In the domain of gas flow metering, the commercial impact of uncertainty in flow measurement depends on quantity of gas, commercial value of gas metered and uncertainty in measurement. For the high throughput gas metering system, marginal reduction in uncertainty helps to reduce the financial exposure.

The sole purpose of metering system is to measure quality and quantity of natural gas flown through the metering station. When measurement is made, the outcome depends on the measurement procedure, instruments, human factor, environment factors etc. Due to the effect of these factors, repeated measurements don’t tend to agree with one another but are dispersed in a certain band.

Uncertainty is defined as a parameter, associated with the result of a measurement that characterizes the dispersion of the values that could reasonably be attributed to the measurand. The parameter may be, for example, a standard deviation, absolute value or % of measurand along with level of confidence.

Uncertainty Estimation: It is not possible to conduct exhaustive statistical investigation on every natural gas metering system before it is built because of the differences due to manufacture and type of instruments, metering station piping layout, environment condition where the metering system is to be deployed and the operating parameters at which natural gas to be measured, etc. Due to economic constraints, uncertainty of a metering system must be evaluated by the Type B method during the design stage itself.

Uncertainty estimation should be performed during the basic engineering and the design stage of the metering system to understand the estimated uncertainly in the flow measurement. If the estimated uncertainty is beyond the acceptable limit, suitable actions, like changes in the design, selection of instrument, testing, etc can be taken.

Stages Of Uncertainty Estimation: Uncertainty estimation in the gas metering system can be performed in following stages:
1.    Formulation: a). defining the output quantity, b) identifying the input quantities on which output is calculated; c) developing a measurement model relating output to the input quantities’ and d) assigning probability distributions to the input quantities.
2.    Propagation – derive the effect of unit change in input on the output using: a) numeric method, b) analytic method, or c) Monte Carlo method.
3.    Summarizing: a.) this means summarizing the effect of input uncertainty to estimate uncertainty in output.

The natural gas is predominantly measured by DP measurement technique or gas ultrasonic flow meter (USM). As the DP measurement technique is matured and measurement by USM is emerging, the discussion in this paper is mainly based on gas measurement by USM.

Basic approach in estimation of uncertainty:
The parameters like line pressure, line temperature, gas composition, gross volume flow rate etc. are measured continuously to derive gas standard volume /mass flow rate. The uncertainty in gas volume at standard condition would depend on the uncertainty associated with measurement of these parameters.

In practice, these measurements may have many potential sources of uncertainty, including:
1)    Effect of environmental conditions on the measurement;
2)    Effect of process operating condition on the measuring instruments;
3)    Uncertainty in values of measurement standards, reference materials, constants;
4)    Approximations and assumptions in the measurement method;
5)    Incorrect installation and error in instruments;
6)    No representative sampling ; and
7)    Operator skill etc. etc

These sources are not necessarily independent, and some of the sources may contribute to other sources too. An unrecognized systematic effect cannot be taken into account in the evaluation of the uncertainty of the result of a measurement but contributes to its error. Hence the skill lies in identifying all potential contributors to measurement uncertainty for the given application.

Uncertainty budget for each of the main input parameter should be prepared. All factors that contribute in the uncertainty in measurement of input parameters should be accounted. Uncertainty budget helps in indentifying the biggest contributors to uncertainty in the final output.

A multistage measurement model for the given application should be developed where the output quantities from previous stages become the input quantities to subsequent stages. This model helps to indentify the major contributors at each stage of measurement. It also helps in simplifying the estimate and avoids the gross mistake.

Generic model as in Fig 1 will be useful for understanding

Propagation of input uncertainty:
While combining uncertainties of input parameters, it is necessary to consider the effect each input quantity has on the final result as each input may not have identical effect on output. The sensitivity of an output parameter to an input parameter is defined by sensitivity coefficient. The sensitivity coefficients can be calculated either by the Analytical method, the Numerical method or Monte Carlo Analysis

As the measurement model (Equ 1) of effect of uncertainty in line pressure, line temperature and gas composition on compressibility factor is very complex, forming partial derivatives or numerical approximations is very difficult. Moreover uncertainty in composition of calibration gas is correlated as all components can not have positive error or negative error at the same. Some components may be having positive error while other balance will be having negative error. A suitable Monte Carlo Method (MCM) may be implemented for the propagation of uncertainty in gas composition in compressibility and energy value.

Good design and operating practices to reduce uncertainty:
Good design practices: Followings practices during design and engineering stage, besides guideline in various codes and standard, may help to reduce the uncertainty in primary measurements:
1.    Ambient temperature effect on pressure transmitter may be one of the substantial factors in error in pressure measurement. Where ambient temperature is expected to be substantially different than calibration temperature, installation provisions that reduce the ambient temperature effect should be considered.
2.    While calibrating pressure transmitter with dead weight tester, correction factors for critical parameters like air buoyancy, relative humidity and piston temperature etc should be considered besides local gravity correction factor to obtain the minimum uncertainty in pressure measurement.
3.    Class A RTD sensor may be employed as it has greater accuracy than class B. To further improve overall temperature measurement performance, the sensor should be calibrated. The exact values for the Callendar-van Dusen constants which are specific to each RTD sensor should be derived by calibrating each individual sensor. Using Callendar-van Dusen constants, temperature transmitter / flow computer to generate a sensor curve that describes the relationship between resistance and temperature for particular sensor.
4.    Field instruments that provide best performance in the harsh field environment like wide variation in ambient temperature, humidity etc and having long terms stability should be chosen.
5.    The uncertainty in USM measurement depends on the uncertainties in a) flow calibration laboratory, b) measurements of the meter body dimension, c) weighing factors /flow profile correction factor/calculation, d) transit time measurement. After wet calibration (calibration at high pressure natural gas in reputed flow laboratory) and adjustment, the errors in indicated flow rate, caused by these parameters, are compensated by laboratory calibration factor. Transferring the meter to the field, additional uncertainty due to the operating conditions and installation effect should be accounted.
Proper application and installation of USM are most significant parameters in the measurement chain. Laboratory testing of flow profiler and USM with different upstream disturbances as recommended in ISO/AGA is normally done at fully developed flow profile upstream to these elements. However practically in the field, upstream of the measurement section may be having series of piping elements like header, tees, elbows etc which may be different than when Flow profiler/USM were type tested in calibration laboratory. Moreover the purpose of flow profiler type testing is to prove that flow profiler limit the maximum additional error due to flow perturbations to +/- 0.3 %. Hence additional uncertainty due to field installation may be reduced by increasing the length of upstream straight pipe to the practically possible. Straight pipe length upstream of flow profiler helps to develop flow profile and reduce swirl. USM manufacture may be consulted to ensure that additional length is not creating any negative impact.
6.    Where practically feasible, based on commercial viability and application requirement, USM wet flow calibration may be carried out using actual upstream pipe work and process conditions similar to field application to compensates the velocity profile distortions caused by various combinations of upstream fittings, valves, bends etc.
7.    If it is practically not feasible to use the actual upstream pipe work during USM wet calibration, the USM should be calibrated along with the actual high performance perforated plate-type conditioner and meter tube as one package. This package should be treated as one unit and should not be dismantled after calibration for the purpose of transportation etc. to avoid additional uncertainty.
8.    Where possible, calibrate USM at a similar pressure and temperature to meter operating conditions. Differences in dimensions due to pressure and temperature differences between calibration and operation should be corrected during actual application.
9.    Where economically feasible, employ USM verification method by operating two meters in series. Compare USM performance online with foot-prints obtained during lab calibration and initial commissioning. Set the tolerance limit as low as practically possible.
10.    Gas chromatograph (GC) functions on the theory of separation of components of interest in the sample and then measurement using known reference. Uncertainty in analyzed gas composition mainly depends on repeatability in sample volume, its separation and uncertainty in the reference calibration gas. Certified reference gas (calibration gas) should be of primary standard prepared using gravimetric method. GC repeatability depends on ambient temperature variation. Installation provisions that reduce the ambient temperature effect on GC like temperature controlled analyser cabinet, 3 sided analyser shelter etc should be considered.
11.    Sample takeoff and sample conditioning are equally important as gas chromatograph analysis. Non representative sampling or alteration of sample during conditioning will add uncertainty in gas composition analysis. The sample system should maintain the temperature of the sample at least 30 DegF (17 DegC) above the Hydrocarbon Dew Point of the sample. The materials in the Sample System should not adsorb any of the components of interest in the sample. Stainless Steel components should be used for Natural Gas. It would be prudent to ensure sampling system complying with relevant codes and standards.

Good operating practices: It is important to reduce uncertainty in custody transfer flow measurement to the practical limit. Following general operational good practices may help to identify and reduce uncertainty in measurements:

• Employ extensive on line diagnostic practice to identify any cause of additional uncertainty during operation
• Calibration of all instruments should preferably be traceable to national standards via an unbroken shortest chain of measurements. It may be noted that in a successive chain of calibrations, the uncertainty increases at every step of the chain. Hence use the lowest calibration chain where possible.
• GC repeatability should be regularly ensured complying with relevant codes and standards. Detector repeatability may be checked using of working reference gas of primary standard having closely related composition to calibration gas. Correctness of valve timing in the three columns gas chromatograph should be verified by molecular weight v/s response factor curve on the logarithm graph. Ideally the graph should be linear.
• Regular verifying values of all the constants used in flow calculation, proper use of measurement units and unit conversion constants, online flow calculations by reputed offline software etc helps to avoid systematic uncertainty
• Keep good records of measurement raw data and calculations for verification

Caution:
Mistakes in recording or analyzing data can introduce a significant unknown error in the measurand. Gross mistakes may be identified by a proper review of the data while smaller ones could appear as random variations. Measures of uncertainty are not intended to account for such mistakes. An experienced and trained staff helps in reducing human errors.

Conclusion:
This article provides general guidelines on assessing uncertainty in natural gas measurement. Uncertainty can be estimated by Type B method and reasonably reduced by employing practical aspects suggested. The evaluation of uncertainty depends on detailed knowledge of the natural gas measurement system. The quality of the uncertainty estimation therefore depends on the assessor’s knowledge, understanding and critical analysis abilities. Possible reduction in uncertainty depends on practical feasibility and commercial viability on implementation of provided guidelines for particular application.

Disclaimer:
The suggestions and guidelines provided in this article should be considered general in nature and are not claimed to be authoritative and final. Readers are encouraged to refer to relevant codes and standards by legal metrology and other institutes.

Bibliography:
1.    Practical Approach in Estimating uncertainty in Oil and gas flow measurement – By Chandu Bhatasana
2.    ISO 5168:2005: Measurement of fluid flow -- Procedures for the evaluation of uncertainties
3.    ISO 17089-1:2010: Measurement of fluid flow in closed conduits — Ultrasonic meters for gas Part 1: Meters for custody transfer and allocation measurement
4.    ASTM D1945-03; standard test method for Analysis of Natural gas by gas chromatography
5.    AGA Report No. 9, Measurement of Gas by Multipath Ultrasonic Meters (2007)

Author: Chandu Bhatasana, PMP, PMI-RMP, has 19 years of experience in Metering & instrumentation. He has been working with Aker Solutions as Senior Metering Engineer.

VIEW ALL