Prepared by the LEVON Group, LLC and URS Corporation

ADDRESSING UNCERTAINT Y IN
OIL AND NATUR AL G AS INDUSTRY
GREENHO USE G AS INVENTO RIE S
TECHNICAL CONSIDER ATIONS
AND CALCU L ATION METH ODS

SEPTEMBER 2009
PILOT TEST VERSION

INTERNATIONAL PETROLEUM INDUSTRY

ENVIRONMENTAL CONSERVATION ASSOCIATION
INTERNATIONAL PETROLEUM INDUSTRY ENVIRONMENTAL CONSERVATION ASSOCIATION

The International Petroleum Industry Environmental Conservation Association (IPIECA)

was founded in 1974 following the establishment of the United Nations Environment
Programme (UNEP). IPIECA provides one of the industry’s principal channels of
communication with the United Nations.

IPIECA is the single global association representing both the upstream and downstream
oil and gas industry on key global environmental and social issues. IPIECA’s programme
takes full account of international developments in these issues, serving as a forum
for discussion and cooperation involving industry and international organizations.
IPIECA’s aims are to develop and promote scientifically-sound, cost-effective, practical,
socially and economically acceptable solutions to global environmental and social issues
pertaining to the oil and gas industry. IPIECA is not a lobbying organization, but provides
a forum for encouraging continuous improvement of industry performance.

5th Floor, 209–215 Blackfriars Road, London SE1 8NL, United Kingdom

Telephone: +44 (0)20-7633-2388

Fax: +44 (0)20-7633-2389
Email: info@[Link]
Website: [Link]

The American Petroleum Institute is the primary trade association in the United States
representing the oil and natural gas industry, and the only one representing all segments
of the industry. Representing one of the most technologically advanced industries in the
world, API’s membership includes more than 400 corporations involved in all aspects
of the oil and gas industry, including exploration and production, refining and marketing,
marine and pipeline transportation and service and supply companies to the oil and
natural gas industry.

API is headquartered in Washington, D.C. and has offices in 27 state capitals and provides
its members with representation on state issues in 33 states. API provides a forum for
all segments of the oil and natural gas industry to pursue public policy objectives and
advance the interests of the industry. API undertakes in-depth scientific, technical and
economic research to assist in the development of its positions, and develops standards
and quality certification programs used throughout the world. As a major research institute,
API supports these public policy positions with scientific, technical and economic research.
For more information, please visit [Link].

1220 L Street NW, Washington DC, 20005-4070 USA

Telephone: +1-202-682-8000
Website: [Link]

The Oil Companies' European Association for Environment, Health and Safety
in Refining and Distribution

Founded in 1963, CONCAWE engages in research on environmental, health and safety

issues related to the downstream oil industry with a view to improve the understanding
of these issues by the Industry, Authorities and Consumers, while upholding the three
principles of Sound Science, Transparency and Cost-Effectiveness. CONCAWE represents
nearly 100% of mineral oil refiners in the European Union, Norway and Switzerland and
is wholly funded by its members. It is based in Brussels and registered as a non-profit
Association under Belgian Law.

Bld du Souverain 165

B-1160-Brussels

Telephone: +32-2566-9160
Website: [Link]

© IPIECA 2009. All rights reserved. No part of this publication may be reproduced, stored
in a retrieval system, or transmitted in any form or by any means, electronic, mechanical,
photocopying, recording or otherwise, without the prior consent of IPIECA.
 Table of Contents
SECTION Page
FOREWORD vi
DOCUMENT AT A GLANCE viii
1.0 INTRODUCTION 1-1
1.1 Importance of Accurate and Reliable GHG Accounting 1-1
1.2 Overview of Uncertainty Terminology 1-2
1.3 Types of Errors 1-3
1.4 Determination of Uncertainty Intervals 1-3
2.0 SOURCES OF UNCERTAINTY 2-1
2.1 Overview of Emissions Inventory Uncertainty 2-1
2.2 Emissions Inventory Uncertainty in the O&G Industry 2-3
2.3 Sources of Measurement Uncertainty 2-5
2.4 Emission Estimation Approaches 2-7
2.5 Inventory Steps and Data Aggregation 2-9
3.0 OVERVIEW OF MEASUREMENT PRACTICES 3-1
3.1 Flow Measurement Practices 3-3
3.1.1 Measurements by Orifice Meters 3-3
3.1.2 Measurement of Flow to Flares 3-5
3.1.3 Example: “Custody Transfer” Measurements 3-7
3.2 Uncertainties of Flow Measurements for GHG Inventories 3-9
3.3 Uncertainties of Sampling and Analysis for GHG Estimation 3-13
3.3.1 Gaseous Samples Collection and Handling 3-13
3.3.2 Quantifying Sampling Precision 3-15
3.4 Carbon Content Measurement Practices 3-15
3.4.1 Laboratory Based Measurements 3-16
3.4.2 On-line Measurements 3-18
3.5 Heat Content Determination 3-19
3.5.1 Direct Measurements 3-19
3.5.2 Computational Methods 3-21
3.6 Laboratory Management System 3-22
4.0 STATISTICAL CONCEPTS AND CALULATION METHODS 4-1
4.1 Measurement Uncertainty 4-1
4.1.1 Precision and Bias 4-1
4.1.2 Confidence Intervals 4-3
4.2 Overview of Uncertainty Propagation 4-4
4.2.1 Propagation Equations 4-6
4.2.2 Correlation Coefficient 4-7

Pilot Version, September 2009 ii

 Table of Contents, continued
SECTION Page
4.3 Quantifying Emission Estimation Uncertainty 4-8
4.3.1 Simple Emission Estimation: EF x AF 4-8
4.3.2 Fugitive Emission Estimation 4-10
4.3.3 Emission Factor Uncertainty 4-13
4.4 Quantifying Measurement Uncertainty 4-16
4.4.1 Measurement Uncertainty – Multiple Measurements 4-21
4.5 Aggregating Uncertainty 4-29
4.5.1 Rounding off of Statistical Estimate 4-33
4.6 Assessing Data Correlations 4-33
4.6.1 Monte Carlo Simulation 4-34
4.6.2 Comparison of Error Propagation and Monte Carlo 4-35
5.0 UNCERTAINTY CALCULATION EXAMPLES 5-1
5.1 Introduction 5-1
5.2 Example 1: Onshore Oil Field with High CO2 Content 5-1
5.2.1 Background 5-1
5.2.2 Assigning Uncertainties to Activity Factors 5-5
5.2.3 Propagating Uncertainty in Calculated Activity Factors 5-7
5.2.4 Propagating Asymmetric Uncertainty Distribution 5-10
5.3 Example 2: Refinery 5-10
5.3.1 Background 5-10
5.3.2 Uncertainty Comparison for FCCU Emission
Estimation Methods 5-10
5.3.3 Uncertainty Comparison for Hydrogen Plant Emission
Estimation Methods 5-15
5.4 Strategic Reduction of Uncertainty 5-18
5.4.1 Opportunities to Improve the Uncertainty Estimation for
the Onshore Oil Field 5-20
6.0 REFERENCES 6-1
Appendices
A Glossary of Statistical and GHG Inventory Terms
B List of Industry Measurement Standards
C Operating Conditions, Inspection, Calibration and Expected
Uncertainties for Common Flow Meters
D Select Measurement Methods Summaries
E Units Conversion
F Uncertainty Estimation Details for an Example Inventory

Pilot Version, September 2009 iii

 List of Tables
SECTION Page
2-1 Overview of Methods Used to Estimate Emissions Uncertainty 2-2
3-1 Example of FFMS Combined Uncertainty 3-7
3-2 Summary of Alberta ERCB Accuracy Requirements 3-8
3-3 Compilation of Specifications for Common Flow Meters 3-10
3-4 Summary of Selected Carbon Content Measurement Methods 3-17
3-5 Summary of Selected Heating Value Measurement Methods 3-20
4-1 Component Counts and Uncertainties 4-11
4-2 Emission Factors and Uncertainties 4-11
4-3 Methane Weight Fractions for Production Operations by Service 4-12
4-4 Estimate Fugitive CH4 Emission 4-13
4-5 Uncertainty in a Single Measurement 4-19
4-6 Flow Measurements (in Mscf/day) 4-20
4-7 Uncertainty in Summation of Flow Measurements – Annual 4-21
4-8 Uncertainty in Summation of Flow Measurements – Monthly 4-21
4-9 Measured Composition Data 4-25
4-10 Average Composition Calculations 4-25
4-11 Measured Monthly Composition Data 4-27
4-12 Measured Monthly Emission Factors 4-29
4-13 Comparison of Annual Emission Estimates 4-32
5-1 Onshore Oil Field (High CO2 Content) Emission Sources 5-2
5-2 Onshore Oil Field (High CO2 Content) Fugitive Emission Sources 5-4
5-3 Gas Composition for Onshore Oil Field (High CO2 Content) 5-4
5-4 Onshore Oil Field (High CO2 Content) Emissions 5-8
5-5 Uncertainty Comparison for FCCU Estimation Methods 5-14
5-6 Composition Data 5-15
5-7 Hydrogen and Carbon Composition Data 5-17
5-8 Emission Uncertainty Ranking for Onshore Oil Production Example 5-19
D-1 Natural Gas Components and Range of Composition Covered D-1
D-2 ASTM D1945-03 Precision for Natural Gas Samples D-2
D-3 ASTM D1945-03 Precision for Reformed Gas Samples D-2
D-4 ASTM D4891-89 Range of Composition for Natural Gas D-5
Components
F-1 Gas Composition for Onshore Production Platform F-1
F-2 Operating Parameters for Boilers, Heaters and Reboilers F-2
F-3 Operating Parameters for Turbines F-3
F-4 Natural Gas Composition F-4
F-5 Operating Parameters for Diesel Engines F-6
F-6 Operating Parameters for Flares F-13

Pilot Version, September 2009 iv

 List of Tables, continued
SECTION Page
F-7 Operating Parameters for Fleet Vehicles F-16
F-8 Operating Parameters for Dehydration F-19
F-9 Operating Parameters for Storage Tanks F-21
F-10 Operating Parameters for Acid Gas Removal F-23
F-11 Operating Parameters for Pneumatic Devices and Chemical F-25
Injection Pumps
F-12 Operating Parameters for Maintenance/Turnaround Emissions and F-28
PRVs
F-13 Onshore Oil Field (High CO2 Content) Fugitive Emission Factors F-33
F-14 Fugitive Emission and uncertainty Estimates F-35
F-15 Onshore Oil Field (High CO2 Content) Emissions F-38

List of Figures
SECTION Page
4-1 Measurement Error Over Time of an Unbiased Estimate 4-2
4-2 Measurement Error Over Time of a Biased Estimate 4-3
4-3 Decision Diagram for Emission Factor Uncertainty 4-15
4-4 Decision Diagram for Measurement Uncertainty 4-17
4-5 Step C – Decision Diagram for Uncertainty Aggregation 4-31
5-1 Onshore Oil Field: Summary of Emissions 5-20
5-2 Onshore Oil Field: Summary of CO2 Equivalent Emissions and 5-21
Uncertainties

Pilot Version, September 2009 v

FOREWORD

The global oil and natural Gas (O&G) industry has been active in promoting consistency and
harmonization for industry greenhouse gas (GHG) emission inventories. Industry associations and their
members have been contributing to the development of guidance for accounting and reporting of GHG
emissions (API/IPIECA/OGP, 2003), and compiling methodologies that are appropriate for estimating
GHG emissions from industry operations (API, 2009). This guidance has been recently augmented with
guidelines to account for reductions associated with GHG projects (API/IPIECA, 2007).

The uncertainties inherent in the data used for emission inventories help inform and improve
understanding for the data’s use. The uncertainty of an O&G company’s GHG emission inventory, or of
its quantified emission reductions, is determined largely by the uncertainties in the estimates of its key
(largest) contributing sources. In turn, each of these uncertainties depends on the quality and availability
of sufficient data to estimate emissions. Our measured data robustness is receiving increased attention to
understanding determinations of GHG emissions and emission reductions.

The American Petroleum Institute (API), the oil companies’ association for the Conservation of Clean Air
and Water in Europe (CONCAWE) and the International Petroleum Industry Environmental Conservation
Association (IPIECA) convened an international workshop on the topic on January 16, 2007, in Brussels,
Belgium. The goals of this workshop were to:

• Develop an understanding of the relative importance of the key factors that contribute to
uncertainty;

• Review and discuss emerging techniques for quantitative assessment of the uncertainty and
accuracy of GHG emissions estimates;

• Identify emission sources and methods where O&G industry efforts are needed to improve
accuracy and reduce uncertainty to acceptable levels; and

• Create a prioritized list of topics to be addressed by the O&G industry to minimize emissions
estimation uncertainty and improve data accuracy.

A summary report, as well as all the workshop presentations, is posted on the IPIECA website
(API/IPIECA, 2007).

The workshop served as the first step in the process of addressing uncertainty and accuracy issues and in
the ensuing industry discussion, a list of priority issues was prepared. This list is comprised of items that
industry experts ought to address in a systematic fashion. As presented in the workshop summary report,
the issues listed by industry members fall into three thematic areas:

1. Measurement methods;

Pilot Version, September 2009 vi

 2. Computational methods; and

3. External communications.

Industry recognizes the need for meeting regulatory mandates and stakeholders’ expectations, and follow-
up activities will be designed to provide opportunities for continued dialogue and collaborative activities
with stakeholders.

The goal of these guidelines is to augment existing industry guidance and provide technically valid
approaches that would be applicable for use by the global O&G industry to improve GHG emissions
estimation robustness and data quality. The aim is to summarize in a single document overarching
guidance for meeting data needs of a range of initiatives as well as the requirements of diverse GHG
regimes.

These technical considerations and statistical calculation guidance are not designed to be an industry
standard. They will merely provide a brief summary and an overarching discussion of possible
approaches for addressing each of the topics covered. These include: clarification of the sources of
uncertainty in entity GHG inventories; information on measurement practices and their associated
uncertainties; and explanation of statistical procedures that can be used to quantify uncertainties. This
document also contains several case studies that illustrate how to implement the recommended
approaches.

Pilot Version, September 2009 vii

DOCUMENT AT A GLANCE
This document is designed to provide a summary of technical considerations that are important for
understanding and calculating GHG emission inventory uncertainty. The document will provide needed
technical background and specific calculation methods to determine uncertainties with targeted
measurements and emission factors and determine how to aggregate these individual terms to derive
uncertainty ranges (at a pre-designated probability level) for entire GHG inventories, at any given level.

These emission inventories of typical oil and natural gas (O&G) operations are quite complex; they are
based on a combination of measured and estimated emissions data, according to local requirements and
available information. The overall range of uncertainty associated with an entity GHG inventory is
determined primarily by the uncertainty associated with the largest (“key”) sources of emissions. In turn,
the confidence interval associated with each individual source depends on the availability of sufficient
data to estimate emissions, or on the quality of that data, in order to properly account for emission
variability.

Uncertainty analysis is a potential tool to not only assess confidence intervals, but more importantly, to
allow the targeting of specific areas for enhanced data collection. Such an analysis will enable a user to
rank order the importance of different emission sources in terms of their overall contribution to the
emissions inventory and its overall uncertainty range.

This document is a companion to the API Compendium of Greenhouse Gas Emission Methodologies for
the Oil & Gas Industry (2009 API Compendium). It provides a range of background information on
industry practices and specific calculation methods that would enable inventory developers to quantify
and better understand the uncertainty associated with the resultant GHG emissions.

Section 1.0 introduces some basic concepts and terms that provide a foundation for understanding GHG
emissions inventory uncertainty. This terminology is used throughout the document. This section covers:
the importance of reliable GHG accounting; a terminology overview; definition of error types; and a
description of the determination of uncertainty ranges (also known as confidence intervals).

Section 2.0 discusses the major sources of uncertainty in GHG inventories. It moves from general
concepts to issues that are germane to GHG inventories in the O&G industry. It also describes factors
that could introduce errors into the emission measurements process and contribute to the range of
uncertainties of estimated emissions. It introduces the categories of emission estimation approaches and
their uncertainty implications, and concludes with a short description of emission inventory steps and data
aggregation.

Section 3.0 provides an overview of measurement practices, focusing on gas flow measurements and the
determinations of carbon content and heating values of combusted fuels. The section recognizes industry
recommended practices and standards that have traditionally applied to “custody transfer”. This section

Pilot Version, September 2009 viii

goes on to discuss data considerations when collecting information on activity levels and applicable GHG
emissions. It includes an overview of measurement practices that would result in high quality data when
properly implemented, while focusing on measurements that are applicable to the key contributing
sources, i.e. carbon dioxide (CO2) emissions from combustion devices. Topics discussed include: flow
measurement practices; uncertainties of flow measurements for GHG inventories; uncertainty of sampling
and analysis for GHG estimation; and laboratory management systems.

Section 4.0 provides calculation methods for quantifying GHG inventory uncertainty. It addresses the
statistical characterization of measurement uncertainty, and the uncertainty associated with published
emission factors, and concludes by providing specific statistical methods and industry relevant examples
for the propagation of uncertainty. This section links the statistical concepts introduced in earlier sections
to their application for measurement uncertainty and error propagation. Decision trees are provided to
guide the user through the steps needed to quantify uncertainty for each part of a GHG inventory.

Section 5.0 takes the guidelines, procedures, and equations for calculation of uncertainty that were
outlined in Section 4 and applies them to a hypothetical facility to demonstrate how to calculate total
uncertainty for a facility-level GHG inventory. The statistical approaches discussed previously are also
applied to two select refinery operations to examine and compare uncertainty estimates for different
emission estimation methods. The hypothetical example facilities and operations are taken directly from
the 2009 API Compendium containing details of how the inventory was generated or calculated. This
section also provides a detailed analysis of the resulting inventory data and its uncertainty in the context
of improving GHG data quality and reducing uncertainty.

Detailed technical information is organized in six appendices, as follows:

Appendix A – Glossary of statistical and GHG inventory terms;

Appendix B – A comprehensive list of applicable industry measurement standards;

Appendix C - Operating conditions, inspection, calibration and manufacturers’ reported measurement

errors for common flow meters;

Appendix D – Measurement method summaries for carbon content measurement methods, and heating
value measurement methods;

Appendix E – Unit conversions including energy units, common units of measure for fossil fuel heating
content values, and carbon content of selected fuels; and

Appendix F – Calculation details for uncertainty estimation for an example GHG inventory.

Pilot Version, September 2009 ix

1.0 INTRODUCTION

Policymakers use entity GHG inventories and reported facility-

Section Focus
level GHG emissions to develop strategies and policies for
This is an introductory section that
emission reductions and to track the progress of these policies. introduces some basic concepts and terms
that are the foundation for technical
Both regulatory agencies and corporations rely on inventories to
considerations related to understanding
better understand emission sources and trends. GHG inventory GHG emissions inventory uncertainty.
This terminology will be further expanded
data are associated with varying degrees of uncertainty, and such
throughout the next sections of the
actual uncertainties have both technical and policy implications. document. The subsections include:
ƒ Importance of accurate and reliable
“Uncertainty analysis” has been increasingly recognized as an
GHG accounting;
important tool for improving national, sectoral, and corporate
ƒ Overview of uncertainty terminology;
inventories of GHG emissions and removals (IPCC, 2000). This
ƒ Types of errors; and
increased attention to accurate inventories leads to the need to ƒ Determination of uncertainty intervals.
provide guidance to industry on the technical considerations and
calculation methods for consistent estimation of GHG inventory
uncertainty. This would typically consist of:

• Determination of the uncertainties associated with the individual measurements and factors used
in constructing the emissions inventory, and

• Propagation and aggregation of these individual terms to derive uncertainty intervals (at a pre-
designated probability level) for the whole inventory.

Clearly, the extent and scope of such analysis will depend on the likely uses of this information. For
example, the uncertainty analysis required for data that are merely used for relative ranking or comparison
of trends would be different than that required to demonstrate attainment of GHG emission limits, or
progress made towards meeting GHG emission reduction targets.

1.1 Importance of Accurate and Reliable GHG Accounting

Key areas that benefit from reliable GHG accounting include:

• Focused GHG emissions management;

• Reduced business risk and reputation management; and
• Participation in GHG emissions mitigation programs.

Pilot Version, September 2009 1-1

Since an understanding of the magnitude and sources of GHG emissions is critical to properly managing
these emissions, employing a consistent approach can significantly improve industry-wide, comparable
estimates of emissions, and emission reductions.

Higher quality GHG data leads to higher certainty of emission assessments, and improved confidence in
the data reported. This is true for national and government assessments, and is also important at the
entity, or facility level. To ensure that a company’s strategies and forward-looking actions are based on
the most robust data set and most appropriate computational methods, it is important that this data set and
method be based on four key factors (“The Four C’s”).

Comparability, Consistency, Certainty, Confidence

1.2 Overview of Uncertainty Terminology

The API in Chapter 13 of its Manual of Petroleum Measurement Standards (MPMS), provides detailed
guidance on statistical concepts and procedures for addressing the statistical procedures that should be
followed when estimating a true quantity from measurements–or models–and when deriving the
confidence interval of the results (API, 1985). That chapter also examines sources of error and
recommends how to develop a statement of the overall range of uncertainty of the results obtained. Some
of the key terms used in the API MPMS are presented in Exhibit 1-1.

EXHIBIT 1-1: SELECTED TERMINOLOGY

ƒ Accuracy – Ability to indicate values that closely approximate the true value of the measured variable.
ƒ Bias – Any influence on a result that produces an incorrect approximation of the true value of the variable being
measured. Bias is the result of a predictable systematic error.
ƒ Confidence interval (or range of uncertainty) – The range or interval within which the true value is expected
to lie with a stated degree of confidence.
ƒ Confidence level – The degree of confidence that may be placed on an estimated range of uncertainty.
ƒ Error – The difference between true and observed values.
ƒ Precision – The degree to which data within a set cluster together.
ƒ Random error – An error that varies in an unpredictable manner when a large number of measurements of the
same variable are made under effectively identical conditions.
ƒ Spurious error – A gross error in procedure (for example, human errors or machine malfunctions).
ƒ Systematic error – An error that, in the course of a number of measurements made under the same conditions
on material having the same true value of a variable, either remains constant in absolute value and sign, or
varies in a predictable manner. Systematic errors result in a bias.
ƒ Variance – The measure of the dispersion or scatter of the values of the random variable about the mean.

Source: API MPMS Chapter 13.1

Pilot Version, September 2009 1-2

The Intergovernmental Panel on Climate Change (IPCC) provided a conceptual basis for uncertainty
analysis as part of their “Good Practices Guidelines” for managing and estimating uncertainty in national
emission inventories (IPCC, 2000). The IPCC has introduced a structured approach to estimating GHG
inventory uncertainty by incorporating methods used to determine uncertainties of individual terms and
aggregating them to the total inventory. The IPCC also recognize that other uncertainties may exist, such
as those arising from inaccurate definitions or procedures, which cannot be addressed by statistical means.
Appendix A presents an expanded glossary of statistical terms with comments on how these terms are
used in the context of GHG emission inventories.

1.3 Types of Errors

The difference between the observed value of a variable and its true value includes all errors associated
with a given measurement or estimation process. Such errors are comprised of instrumentation errors,
errors resulting from faulty sampling procedures, changes in conditions during the measurement period,
or use of improper methods. Three basic types of errors should be considered:

− Spurious errors;
− Systematic errors; and
− Random errors.

One or all of these errors could be associated with individual measurements or input variables used for
deriving an emissions inventory. However, such individual error determinations should not be confused
with the overall inventory uncertainty.

Indicators such as the range, confidence interval, or other error bounds typically are used to quantify an
emission estimate uncertainty. Errors may be due to the inherent variability of the emission processes and
the bias–or imprecision–in the terms typically used to define them. Bias is the result of a systematic error
in some aspect of the emissions inventory process. In contrast, imprecision is due to random errors or
fluctuations in the measurement process.

1.4 Determination of Uncertainty Intervals

The uncertainty intervals associated with measured emission rates, activity data, or emission factors are
characterized by the dispersion of the respective values that are used in their derivation. Mathematically
these intervals are defined as either the standard deviation of the sample populations or the standard
deviation of the sample means. The standard deviation of the mean, known also as the standard error of
the mean, is the standard deviation of the sample data set divided by the square root of the number of data
points. While the standard deviation and variance of the data set do not change systematically with the

Pilot Version, September 2009 1-3

number of observations, the standard deviation of the mean decreases as the number of observations
increases.

Estimating uncertainties in emission inventories is based on the characteristics of the variable(s) of

interest (input quantities) as estimated from the corresponding data set. The statistical computations
could entail the determination of:

− The arithmetic mean (mean) of the data set;

− The standard deviation of the data set (the standard error, the square root of the variance);
− The standard deviation of the mean (the standard error of the mean);
− The probability distribution of the data; and
− Covariances of the input quantity with other input quantities used in the inventory calculations.

The limits of the confidence interval associated with GHG emissions from a source are directly dependent
on the probability distribution, or the probability function, used to represent that data set. For some
probability distributions, there are analytical relationships that relate the standard deviation to the required
confidence intervals. For example, when a normal distribution is assumed for the variable under
consideration, the confidence limits would be symmetric about the mean, and for a 95% confidence
interval the confidence limits are approximately 2 standard deviations above and below the mean.

Hence, the quantification of uncertainty intervals for calculated GHG emissions will depend both on the
accuracy and representativeness of measurement data used and the assumed distributions of other key
parameters used in the computations. The uncertainties associated with both emission factors and activity
data could be best described by probability density functions that are constructed from available data.
The applicable shapes of these probability density functions could be either determined empirically or by
expert judgment, following procedures described in many guideline documents and standards (ISO, 2005;
IPCC, 2000).

This section has introduced the basic concepts that are germane to estimating the uncertainty range of
GHG emissions. The applicable statistical calculation procedures are presented in greater detail in
Section 4.0.

Pilot Version, September 2009 1-4

2.0 SOURCES OF UNCERTAINTY
There are a myriad of sources that contribute to the
Section Focus
uncertainty of an emission inventory. Whether at the national,
This section discusses the major sources
entity, or facility level, the ability to quantify emissions and affecting the uncertainty of GHG inventories.
It moves from general concepts to issues that
understand their associated uncertainty hinges on two main
are germane to GHG inventories in the O&G
factors: industry. It also describes factors that could
introduce errors into the emission
− Readily available data for emission quantification; and measurements process and contribute to the
range of uncertainties of estimated emissions.
− Knowledge of input parameters for statistical The subsections address:
calculation of uncertainty.
ƒ Overview of emissions inventory
uncertainty;
The overall range of uncertainty associated with an entity ƒ Emission inventories uncertainties in the
GHG inventory usually is determined primarily by the O&G industry;
uncertainty associated with the largest (“key”) sources of ƒ Sources of measurements uncertainty;
emissions. Although very large confidence intervals may be ƒ Emission estimation approaches; and
associated with the data used to characterize some small ƒ Inventory steps and data aggregation.
sources, the overall impact on the range of uncertainty at the
entity, or installation level, may often be very small. In turn, the confidence interval associated with each
individual source depends on the availability of sufficient data to estimate emissions, or on the quality of
the data in order to properly account for emission variability.

2.1 Overview of Emissions Inventory Uncertainty

Uncertainties in inventories are the result of three error categories:

− Spurious errors, which may be due to incomplete, unclear, or faulty definitions of emission
sources that result from human error or machine malfunction;

− Systematic errors, which may be due to the methods (or models) used to quantify emissions for
the process under consideration; and

− Random errors, which may be due to natural variability of the process that produces the
emissions.

When assessing the process or quantity under consideration, uncertainties might be associated with one or
more factors such as: sampling, measuring, incomplete reference data, or inconclusive expert judgment.

Uncertainties due to models or equations are related to the proper application of estimation methodologies
to the respective source categories. These errors are typically eliminated as far as possible in advance,

Pilot Version, September 2009 2-1

 when planning the compilation of an emissions inventory and could address as part of emission inventory
assurance processes (API/IPIECA, OGP, 2003).

Adhering to appropriate sampling, measurement, and estimation procedures – with applicable quality
control and quality assurance measures – can help minimize uncertainties. Nonetheless, it is the nature of
the measurement process that makes it impossible to measure a physical quantity without error.

It is this collection of measurement errors and approximations that contribute largely to overall emission
inventory range of uncertainty. Emission inventory development and uncertainty analysis could be
viewed as components of a quality improved management process. Once estimates of uncertainty are
developed, the inventory preparer could review the inventory and target the most significant sources –
those exhibiting the largest range of uncertainty – for more research and refinement. Table 2-1 provides
an overview of selected methods recommended by the U.S. Environmental Protection Agency (EPA) for
qualitative and quantitative estimation of emissions ranges of uncertainty. The uncertainties associated
with natural variability inherent to the emission process and its underlying data can be assessed by
statistical analysis methods. Further discussion and specific procedures for statistical quantification of
uncertainty intervals are provided in Section 4.0.

Table 2-1. Overview of Methods Used to Estimate Emissions Uncertaintya

LEVEL OF
METHODOLOGY DESCRIPTION OF METHOD
EFFORT
Qualitative − Sources of uncertainty are listed and discussed. Low
Discussion − General direction of bias and relative magnitude of imprecision are given
if known.

Subjective Data − Subjective rankings based on professional judgment are assigned to each Low
Quality Ratings emission factor or parameter.

Data Attribute − Numerical values representing relative uncertainty are assigned through Medium
Rating System objective methods.
(DARS)

Expert − Experts estimate emission distribution parameters (i.e., mean, standard Medium
Estimation deviation, and distribution type).
Method − Simple analytical and graphical techniques are then used to estimate
confidence limits from the assumed distributional data.
− In the Delphi method, expert judgment is used to estimate uncertainty
directly.

Pilot Version, September 2009 2-2

 Table 2-1. Overview of Methods Used to Estimate Emissions Uncertaintya

LEVEL OF
METHODOLOGY DESCRIPTION OF METHOD
EFFORT
Propagation of − Emission parameter means and standard deviations are estimated using Medium
Errors Method expert judgment, measurements, or other methods.
− Standard statistical techniques of error propagation typically based upon
Taylor’s series expansions are then used to estimate the composite
uncertainty.

Direct − Monte Carlo, Latin hypercube, bootstrap (resampling), and other High
Simulation numerical methods are used to estimate directly the central value and
Method confidence intervals of individual emission estimates.
− In the Monte Carlo method, expert judgment is used to estimate the values
of the distribution parameters prior to performance of the Monte Carlo
simulation.
− Other methods require no such assumptions.

Direct or − Direct or indirect field measurements of emissions are used to compute High
Indirect emissions and emissions uncertainty directly.
Measurement − Methods include direct measurement such as stack sampling and indirect
(Validation) measurement such as tracer studies.
Method − These methods also provide data for validating emission estimates and
emission models.
a
Extracted from Table 4.1-1 of the Emissions Inventory Improvement Program (EIIP), Chapter IV: “Evaluating the Uncertainties of Emission
Estimates,” U.S. EPA, Research Triangle Park, NC, July 1996

2.2 Emissions Inventory Uncertainty in the Oil & Natural Gas Industry

The API Compendium (API, 2004) and its forthcoming revision (API, 2009) provide an extensive
compilation and tabulation of methods that are used by companies in all the sectors of the O&G industry
to calculate their GHG emissions in a consistent manner. The API Compendium provides a wide range of
emission factors and other emission estimation methods that are directly applicable to all sectors of
industry operations. It also exhibits the ranges of uncertainty (at the 95% confidence level) associated
with each of the case study examples featured in Section 8 of the 2009 API Compendium. Related
industry guidelines include those of the Interstate Natural Gas Association of America (INGAA), which
provides supplemental guidance for estimating GHG emissions from key emission sources associated
with natural gas storage and transmission sector of the industry (INGAA, 2005) and the AGA which has
also published specific guidelines for estimating GHG emissions from gas distribution operations (AGA,
2008).

O&G industry operations vary widely among its operating sectors due to the nature of the operation, its
geographical locations and local practices. Operations in some of the industry sectors are highly
centralized in large and complex facilities while other extend over large geographical areas, with some of

Pilot Version, September 2009 2-3

the operations not contained in traditional; “facilities”. Additionally, companies operations tend to
encompass many jurisdictions, which adds to the complexity of compiling an emission inventory.
Therefore, data availability may be different among industry sectors and regions due to a given sector’s
operational considerations and local requirements.

The uncertainty associated with CO2 emissions from combustion would be primarily attributable to
variation in the composition of combusted fuels and their respective consumption rates (or total volumes).
For quantification of combustion emissions, quality data are typically available for industry facilities in
all sectors, though significant effort may be required in order to collect data for smaller operating
installations that are spread out in multiple locations.

Moreover, since a large fraction of industry operations rely on self-generated, if is the knowledge of the
carbon content of such fuels that is at the root of determining their associated CO2 emissions. For the
exploration and production sector, the composition of these self-generated fuels may vary with the nature
of the producing formations, while for refining it will depend on the composition of the crude oil
processed and the slate of products manufactured. On the other hand, for natural gas transmission and
distribution operations, gas quality and its composition are expected to adhere to contract requirements
and would vary only within a narrow specifications range. Hence, the use of average fuel compositions
data has to be evaluated when compiling an emission inventory since it might result in wide uncertainty
ranges for some sectors and operations, while they might be perfectly acceptable for others.

A different set of parameters is important for understanding emissions associated with process vents and
fugitive emissions. For many of the large process units that are found in refineries and natural gas
processing plants, numerical models (equations) are available for estimating these emissions. For high-
pressure pipelines transmitting natural gas over long distances, the main GHG emissions are due to
reciprocating engines and turbines, venting due to gas blow-down, and fugitive emissions associated with
leaking piping components. For low-pressure gas distribution, most of the GHG emissions come from
compressors and leaks from gas distribution mains and associated equipment. Quantifying emissions, and
their associated uncertainty ranges, for venting and fugitive emissions in the exploration and production
sector poses a real challenge. These emissions could be quite significant for high-pressure uncontrolled
natural gas production operations or miniscule for controlled oil and associated gas production.

For vented emissions, operators in the U.S., as well as other jurisdictions, typically maintain required
records for reporting (and archiving) gas-venting incidents. However, U.S. reports of “lost and
unaccounted for gas” from natural gas pipelines typically account from both vented and fugitive
emissions. Therefore, disaggregating the data would be required in order to derive a separate average
emission factor for venting incidents only. When it comes to fugitive emissions from equipment leaks,

Pilot Version, September 2009 2-4

the uncertainty associated with current practices is significant. The most reliable emission factors still use
mid-1990 field measurement data, and when coupled with the difficulties of obtaining reliable equipment
counts for estimating such emissions, the result could exhibit large uncertainty ranges, although these
emissions may be negligible within the context of the overall inventory.

In summary, since the most prevalent emissions from fuel combustion is CO2, and from venting and
fugitive emissions, CH4, the main contributors to the uncertainty ranges of these respective GHGs in an
inventory generally are:

− For estimating CO2 emissions – Uncertainty is primarily attributable to variation in “self

generated” fuel gas composition and its associated consumption rates. Fuel gas composition
could vary from location to location or from batch to batch, and therefore using average
composition data may lead to a high degree of uncertainty if it is used to estimate emissions.
Measurements (or knowledge) of fuel gas volumes, the gas carbon content (or calorific values),
and careful review of the adequacy of the emission factors used, could help to improve data
quality and minimize this uncertainty.

− For estimating CH4 emissions – Uncertainty is primarily associated with estimates of vented and
fugitive emissions. The records that installations are required to keep (and report) on vented or
released gas might not be similar under all regimes globally. Fugitive emission estimates exhibit
the highest degree of uncertainty due to the use of average emission factors per component,
device or type of operation, and due to improper conversions of existing factors that are expressed
in terms of volatile organic compounds (VOCs) to CH4. Since the CH4 to VOC ratio varies
among installations, or even within different parts of a processing plant, these average emission
factors, coupled with generic conversions from VOC to CH4 may not be the best representation of
CH4 emissions.

2.3 Sources of Measurement Uncertainty

The measurement process is comprised of different steps and each can introduce uncertainty into the final
results. These steps can be applied whether the measurement process involves measurements of activity
data (flow volumes), fuel carbon content speciation, screening for fugitive emission leaks, or direct
emissions testing.

The sources of uncertainty discussed below range from methods choice to physical constraints of the
measurement process itself and the processing of the data obtained.

Pilot Version, September 2009 2-5

a. Measurement methods Uncertainty
Uncertainty due to measurement methods is defined as those additional uncertainty sources that
originate from the techniques or methods inherent in the measurement process, especially if the
methods are not “fit for use.” These uncertainty sources that might impact the measurement system
can significantly affect the final results. Some common sources of measurement system uncertainties
are listed in Exhibit 2-1.

EXHIBIT 2-1: COMMON SOURCES OF

MEASUREMENT SYSTEM UNCERTAINTY
ƒ Assumptions or constants used in the calculations
ƒ Improper placement of monitoring device or extraction of unrepresentative samples
ƒ Spatial or profile uncertainty in the conversion from discrete measurement of a
representative emission factor for similar processes in other locations
ƒ Environmental effects on measurement instruments, such as heat transfer effects on a
temperature probe, or pressure considerations for flow measurements
−
ƒ Drift of an instrument between successive calibrations
ƒ Electrical interference with electronic components
ƒ Variation between the calibration and usage conditions

b. Calibration uncertainty
Manufacturers and companies typically calibrate measurement instruments before they are used in the
field or in a plant. However, instrument calibration needs to be checked periodically to detect
instrument drift and thus reduce measurement uncertainty. The calibration process could achieve that
goal if the process is traceable to a known reference standard and allowance is made for adjustments
of the measurement instrument if a bias is detected during the calibration process. .

c. Data acquisition uncertainty

Uncertainties in data acquisition systems depend on system design. For manual data collection – and
more specifically data entry – human error can be a factor. If hard copy data are used, misplaced
records could contribute to uncertainty. For instrumental data acquisition, uncertainty can arise from
the signal conditioning, and the sensors or recording devices used.

Quality control techniques and robust data management practices can minimize the effects of many of
these uncertainty sources. For example, comparing known input values with their measured or
computed results can provide an estimate of the data acquisition uncertainty. However, it is not
always possible to do this in practice. In these cases, it is necessary to evaluate each potential error
and aggregate these errors to assess the overall uncertainty.

Pilot Version, September 2009 2-6

d. Data processing uncertainty
Typical uncertainty sources in this category may stem from data scatter during the calibration process.
While the equation obtained from a regression analysis of the calibration data represents the best fit,
the data scatter about the curve indicates that, with more data, a slightly different equation would be
obtained in the same way, because the mean of a set of data will change as more values are obtained.
Thus, each coefficient in a regression equation will have an uncertainty associated with it, just as the
mean of a set of values does.

2.4 Emission Estimation Approaches

Emissions information is typically obtained either through direct on-site measurement of emissions or by
using an emission estimation equation or model. Emission estimates are used for facility permitting,
development of control strategies, compliance review, and demonstration of attainment of environmental
goals. The four basic approaches for estimating emissions are:

a. Emissions Factors
Emissions factors are at the core of calculating GHG emissions when compiling O&G industry
inventories. An emissions factor relates the rate of emission of a specific compound with an activity
rate associated with its release.

The general equation for using emission factors for calculating emissions is:

Emissions = Activity Rate x Emissions Factor (Equation 2-1)

In practice, for estimating GHG emissions from the O&G industry operations this means:

− For CO2

Emissions = Volumetric Gas Flow x Carbon Content (Equation 2-2)

– or –
Emissions = Fuel Energy Consumption x Carbon per Heating Value Unit (Equation 2-3)
− For CH4
Emissions = Component or Event Count x Emission Factor (Equation 2-4)

When calculating emissions and their associated uncertainties, it is important to note that the overall
uncertainty is based both on activity data (process flow, throughput, usage, or equipment count) and
on the emission factors used. Each contributor to uncertainty should be assessed independently and
then aggregated in the final analysis, as shown in Figure 4-3 in Section 4, and in the associated
examples provided.

Pilot Version, September 2009 2-7

 Section 3.0 provides an overview of measurement practices with information on measurement
uncertainty. In the absence of direct emission measurements in many cases, emission factors have
traditionally been easy to use, which frequently makes them the low cost method of choice to
calculate emissions.

Over the years, U.S. EPA and other emission factors repository databases have provided average
emission factors that can be used broadly across many industry source categories and operations. The
published emission factors are typically accompanied by a description of the group of processes and
the conditions they represent. Authoritative factors are generally published by EPA in AP-42 (U.S.
EPA, 1995 and further updates), or by the EU EMEP/CORINAIR Emission Inventory Guidebook
(EMEP/CORINAIR, 2007). The IPCC has also launched a new Emissions Factors Database (IPCC,
2006) for use with GHG emission calculations.

b. Continuous Emissions Monitoring

This technique involves continuously measuring flow and concentrations of species directly emitted
into the atmosphere from a specific source, such as a stack. It is accomplished by placing an
applicable monitor at the source. The error associated with these determinations varies for different
compounds and it is not necessarily a more reliable method for measuring emissions especially since
it is not available for monitoring all GHG emissions. Additionally, the acquisition and operating
costs of continuous emission monitoring are very high and thus this technique has limited application
to the evaluation of GHG emissions.

c. Source Testing

Like continuous emissions monitoring, source-testing data may be generated by either extracting a
sample or placing a monitor at a source, followed by analysis to characterize the emitted species. In
this application, the measurement campaign is limited to a specified number of hours. Facilities then
use the average emission rate calculated from source testing to estimate total annual emissions. For
characterizing GHG emissions over a longer period of time (such as a year), periodic sampling and
analysis can be used to determine emission variability.

This measurement approach is generally useful when appropriate test methods are used, and emission
and process data are collected at a frequency that allows good characterization of emission
variability. However, the cost of increased measurement frequency should be balanced with the
contribution of the tested source to the overall inventory and the incremental improvement in the
range of uncertainty that is attainable by this increased testing.

Pilot Version, September 2009 2-8

d. Material Balance

For some emission sources, a material balance (based on fuel flow and carbon measurements) may be
an appropriate means of estimating GHG emissions. Use of a material balance requires knowledge of
total flows or throughput rates, and the corresponding compositions of those streams. Material
balance approaches can also be used for assessment of evaporative losses or for simulation of process
emissions under defined conditions.

Clearly, various methods are applicable to quantifying GHG emissions. Emission estimation methods
range from simple activity measurements multiplied by applicable emission factors to more sophisticated
estimation algorithms. The advanced methods represent an integrated approach that relies on the use of
factors and other data, including generic process simulation models, source specific models, and species
profiles databases.

2.5 Inventory Steps and Data Aggregation

In GHG emission inventories, emission estimates are obtained from many intermediate and independent
results, each of which is calculated from a separate set of data that is characterized by a different range of
uncertainties. The compilation of an entity-wide GHG emissions inventory typically follows a sequence
of steps:

ƒ Establishing boundaries – Where the organizational and operational boundaries are defined (for a
first-time inventory), or examined (for recurring cycles), this step will be largely dictated by local
requirements or corporate policies. It might involve facility-by-facility assessment prior to
aggregation, or it could use other pertinent entity indicators and information;

ƒ Collecting and inputting data – Where the activities data are collected and archived based on the
boundaries established above. The data are then incorporated into appropriate calculation tools for
emission calculations. The level of ‘granularity’ of the data collected and the details of the
calculation methods are dictated by local requirements with industry guidance (API Compendium) as
a resource to provide relevant technical details.

ƒ Validating data compiled – Where various techniques are used to compare the new data with earlier
versions (if available) to identify potential large errors. These errors could include: either large
changes or unchanged activity data for given facilities; operations that are not accounted for; lack of
supporting data for measurements or emission factors used; erroneous units or unit conversions; and
more.

Pilot Version, September 2009 2-9

ƒ Assessing data uncertainty – Where the confidence intervals associated with the data available for
each of the emission sources are characterized independently, as discussed later in this document.
The uncertainty information could be based on documentation of data repositories (API, 2009), expert
judgment, or on measurements conducted during the inventory year.

ƒ Finalizing the inventory – Where the quality-checked and validated data are aggregated for
reporting based on company policy or local requirements. The preferable way of reporting the results
is in terms of the total emissions for each of the GHG species, along with the global warming
potential weighted sum of these emissions (also known as the CO2-E emission).

The overall uncertainty range for each GHG species, and CO2-E, should also be reported with the total
emissions in the format of:

Emissions = Average Value + % (at the 95% confidence limit).

Section 3.0 provides an overview of measurement practices that would result in high quality data when
properly implemented. It focuses on measurements that are applicable to the key sources that contribute
to the overall GHG emissions inventory, i.e., CO2 emissions from combustion devices.

Section 4.0 provides more detailed guidance on the calculation methods that are applicable for defining
single source uncertainties and for aggregating them at the facility (or entity) level. This section provides
a range of examples that are directly applicable to O&G facilities and their GHG reporting needs.

Pilot Version, September 2009 2-10

3.0 OVERVIEW OF MEASUREMENT PRACTICES

In defining aggregated uncertainty of measurement

Section Focus
ensembles used for developing emission inventories, the
This section provides an overview of
uncertainty for each measurement stream must be assessed
measurement practices focusing on gas flow
in a way that is applicable to that measurement method measurements and the determination of carbon
content or the heating values of combusted fuels.
and its implementation in practice. Random errors could
The section recognizes industry recommended
be a major factor in the uncertainty of an individual practices and standards that have traditionally
applied to “custody transfer” of fuel products
observation; however, their contribution to the overall
between companies, and upon entering the
emission inventory diminishes as more measurements are market.
obtained during the reporting period. This section goes on to link these practices to the
acquisition of GHG emissions and related
Note: Random errors tend to average out during the year, activity data, representing a move from reliance
on available data or engineering judgment. The
while systematic errors (or measurement bias) become
subsections address:
more important and tend to accumulate rather than ƒ Flow measurement practices;
diminish over longer periods of time such as a year. ƒ Uncertainties of flow measurements for
GHG inventories;
In fact, determining the true value of any measured ƒ Uncertainty of sampling and analysis for
variable is not practical due to the limitations of GHG estimation; and
measurement equipment and procedures, and the ƒ Laboratory management system.

possibility of human error. Hence, industry measurement

procedures and standards have been developed to emphasize practices that lead to collecting better quality
data, especially for critical measurements. Industry uses standards from several different standards
developing organizations resulting in equivalent measurements, based on the scope of the standard,
company preference, and type of devices used. Consensus industry standards, such as those developed by
ANSI, API, ASTM, ISO and other standard setting organizations have rigor in their development process,
and measurement standards are reviewed at least every five years to ensure that standards are in step with
technological changes and advancements. Most, if not all, of the measurement standards are developed
for measurements associated with ‘custody transfer’ and to define quantities that are essential for robust
financial transactions.

API publishes one of the more comprehensive sets of custody transfer measurement standards, but it is
neither complete nor the only widely recognized source for such industry practices. API’s MPMS
includes over 140 titles, and API publishes approximately eight new or revised measurement standards
each year. Appendix B presents a comprehensive list of specific measurement standards (and their
respective editions) from several standards setting organizations, which could be used to support the

Pilot Version, September 2009 3-1

calculation of GHG emissions. The measurement standards cited in Appendix B are current as of the date
of this publication (i.e., June 2009). However, it is up to the user of these measurement standards to
reference the specific standards/editions used for a given measurement, and to incorporate the updated
measurement procedures, as applicable.

Custody transfer measurements are typically expected to meet a set of performance specifications. For
other measurements, such as those performed to support the development of a GHG emissions inventory,
data quality objectives should be established prior to initiating any data collection in order to ensure that
the uncertainty ranges of the measured quantities are consistent with the intended use of the data.
Throughout this section we provide references and describe a select subset of industry standards that are
most typically used for the respective measurements discussed. Even so, the full list of measurement
methods provided in Appendix B could be used to provide equivalent measurements to meet company
practices and available instrumentation.

For many O&G industry installations, CO2 emissions from combustion and flaring are the largest
contributors to overall GHG emissions. Therefore, this section focuses on measurements and methods
typically used for quantification of these CO2 emissions. The subsections below provide details on flow
measurement practices, and their associated uncertainties, as well as methods for the measurement of
carbon content and heating values of combusted fuels.

For some other O&G industry sectors, such as exploration, production, processing, transmission and
distribution operations, emissions from other GHGs (such as methane) can contribute significantly to a
facility’s GHG emissions. Available emission factors exhibit large uncertainties and might not be
representative of current operating practices. Therefore, for inventories where CH4 constitutes a larger
fraction of the emissions, the overall uncertainty would be expected to be substantially higher.

Industry is now more fully internalizing the potential impact of measurement errors and bias through the
full chain of emission calculations including measurement equipment, software calculations, simulation
models, and the limitations of reliance on existing emission factors on the uncertainty range of resultant
GHG emissions inventories. Considerations of the need for more representative measurement data, and
the assessment of equipment design and age, are gaining more prominence in the process of assembling
high certainty emission inventories. This section focuses on measurements that are pertinent to improved
characterization of combusted fuels and the quantities used, but might not be directly applicable to
measurement of leakages and fugitive emissions. The measurement practices highlighted here represent
an initial step in what could end up being a multi-year effort to improve measurements of GHGs and
quantify the activities that cause their emissions.

Pilot Version, September 2009 3-2

3.1 Flow Measurement Practices

Continuous handling of very large liquid and gas flow volumes is a characteristic of all the sectors of the
O&G industry. Therefore, an in-depth understanding of flow measurement is essential both for internal
process control and for transferring “custody” of intermediate streams or finished products. The accuracy
of measurements of “custody meters” is historically quite high, and practices follow rigorous industry
standards.

Industry has been instrumental in developing international voluntary standards such as ISO 5168 (see
Reference 8) establishing general principles and describing procedures for evaluating the uncertainty of
measuring fluid flow rate or quantity. Annex A of ISO 5168 provides a step-by-step procedure for
calculating and reporting these measurement uncertainties. Similarly, Chapter 14 of the API Manual of
Petroleum Measurement Standards (MPMS), contains detailed procedures and practices for all aspects of
natural gas (and similar gases) fluids measurement and calculation of their associated uncertainties, at the
point where oil or gas enters the marketplace (“custody transfer”) (API MPMS, 2006). Those same
practices are not as rigorously applied to internal accounting and process control during normal
operations.

3.1.1 Measurements by Orifice Meters

Orifice meters are by far the most prevalent flow meter type used in the O&G industry, and are used for
metering products during “custody transfer” as well as for process control and internal accounting. These
same flow meters are used to account for fuel volumes when estimating CO2 emissions. Most of these
flow meters are of the orifice type and are designed for long-term reliability and ruggedness under a
variety of component mixtures and conditions that are essential for consistent fluid blending and
processing. Although for these type of meters temperature and pressure calibrations can be done while
the units are operating, they generally have limited access to direct orifice plate inspections and
maintenance outside of planned shutdown (‘turnaround’) cycles.

Recommended practices for the installation, calibration and calculation of flows for these custody meters
are provided in Section 3 of Chapter 14 of API’s MPMS (API, 2005). This standard was developed by a
collaborative effort by members of API, AGA, and the Processing Gas Association and with contributions
from the Canadian Gas Association, American Chemical Council, the European Union, Norway, Japan
and others. It is designed to ensure global consistency for O&G transactions. The four-part standard for
square-edged, concentric orifice meters consists of:

Part 1 – General equations and uncertainty guidelines;

Part 2 – Specifications and installation requirements;

Pilot Version, September 2009 3-3

 Part 3 – Natural gas applications; and
Part 4 – Background, development, implementation procedures, and subroutine documentation.

The standard recognizes that many factors contribute to the overall measurement uncertainty associated
with many metering applications, as summarized in Exhibit 3-1.

EXHIBIT 3-1: FACTORS CONTRIBUTING TO MEASUREMENT

UNCERTAINTY FOR ORIFICE METERS

a) Tolerances in prediction of coefficient of discharge

−
Derivation of the basic flow equation for an orifice flowmeter is based on physical laws.
−
Any derivation is accurate when all assumptions used to develop the equation are valid.
− The empirical equation for the coefficient of discharge that is included in API 14.3 (Reference 16) was developed
from a large database with well-controlled and quantified independent variables.
b) Predictability in defining the physical properties of the flowing fluid
− All empirical equations and standards for concentric, square-edged orifice meters apply to steady state flow
conditions for fluids that are considered to be clean, single phase, and homogeneous, such as all gases – and most
liquids – in the petroleum, petrochemical, and natural gas industries.
− Fluid's flow rates are expressed in volume units at base (standard or reference) conditions, and the volumetric flow
rates that are measured at the operating flowing conditions are then converted to standard volume with respect to the
base conditions.
− Fluid properties are defined as a function of the operating pressure and temperature that are monitored by secondary
devices. Significant temperature variation between the thermal well and the orifice taps will affect the measurement.
c) Fluid flow conditions
−
Database is available for the empirical equations for coefficient of discharge for steady-state fully developed pipe
flow profile with negligible or no swirl flows and flow fluctuations.
− Deviations from these conditions are typically due to piping installation upstream of the flowmeter and they
introduce flow measurement uncertainty.
d) Construction tolerances in meter components
− Part 2 of the reapproved API MPMS Chapter 14.3 standard lists the changes recommended in the mechanical
tolerance requirements for the orifice meter components.
− The standard encompasses a wide range of diameter ratios for which experimental results are available and some of
the tolerances are more stringent than the tolerances in the previous standards.
e) Uncertainty of secondary devices/instrumentation
− The secondary devices are the instruments used to monitor the flowing fluid temperature, pressure, and the
differential pressure across the orifice plate.
− Parameters affecting the accuracy of the differential pressure device include: ambient temperature, static pressure,
linearity, repeatability, long-term stability, and drift, as well as the uncertainty of the calibration standard.
− The stated accuracy of most differential pressure-measuring devices is expressed as a percentage of the full-scale
reading, which leads to increased error bands with decreasing differential pressures.
f) Data reduction and computation
− Ultimate errors in flow rate computation depend on the accuracy of defining the physical properties of the flowing
fluid, as computed by the microprocessor-based flow computers.
− Computation of the physical properties, especially for gas flows, is dependent on the constituents of gas in the
flowing fluid.
− All fixed input and critical parameters affecting the flow rate computation should be verified to reduce bias error in
flow measurement.

Pilot Version, September 2009 3-4

All these factors should be assessed when estimating the overall range of uncertainty for flow
measurements using thin plate, concentric, square-edged metering systems.

In the reapproved 2006 version of the standard, several changes were incorporated to reduce the
uncertainty attributable to installation effects and to improve the rigor of the flow calculation routines.
The revised standard recognizes the lead-time necessary for upgrading existing installations, and leaves
this lead-time to the discretion of facility operators and their data quality targets for flow measurement
data.

However, it should be recognized that if orifice meter installations are not upgraded to conform to the
new recommendations, measurement bias error may occur. This bias might be due to improper upper and
lower distances from bends and points of flow turbulence that might lead to inadequate flow conditioning
prior to measurement. Additionally, even without changing equipment installations, the standard
recommends adopting new calculation procedures and techniques (explained in Part 1 and 3 of the
standard) that represent significant improvements over the previously adopted approach. It is important
to note that the expected uncertainty ranges for flow measurements quoted in Part 1 of the reaffirmed
standard may differ from those obtained in practice when the equipment installation differs.

3.1.2 Measurement of Flow to Flares

The measurement of flow to flares is distinctly different than other flow measurements. Flares are
designed as safety relief systems and typically are capable of handling highly variable flow rates of
widely varying gas compositions. Therefore, some of the practices that are generally applicable to
custody transfer or process control flows have to be modified when addressing flows to flares. API
published a measurement standard addressing gas or vapor flare flow measurements, which also includes
cautionary details about the effects of fouling (due to entrained liquid droplets, aerosol mists, or other
contaminations) on the measurement (API MPMS, July 2007).

Most flare headers are designed to operate during both non-upset conditions at near atmospheric pressure
and ambient temperature, and during flare episodes, at a wide range of pressure, temperature, and flow
velocities. During such episodes, flare gas compositions are also highly variable and could range from
molecular weights approaching that of hydrogen to molecular weights of C5+.

As with other flow measurements, the accurate determination of flow to flares is dependent on many
parameters such as the ability to predict – or measure – mixture composition, pressure, temperature,
and/or density. The accuracy of measurements associated with highly variable flare gas mixtures will
depend largely on the meter technology type and the ability of the flare flow measurement system
(FFMS) to achieve the targeted response time and analytical accuracy levels. Exhibit 3-2 below lists the

Pilot Version, September 2009 3-5

basic steps needed to conduct a simplified analysis for determining the uncertainty ranges of a given flow
measurement system. The actual approaches for the required calculations are provided in Section 4,
which provides examples demonstrating how to apply these methods.

EXHIBIT 3-2: STEPS FOR SIMPLIFIED UNCERTAINTY ANALYSIS

a) Determination of the equation that defines the meter output

− Governing equations are those applicable for the meter technology type utilized.
b) Determination of the combined sensitivity coefficient
− Numerical values of uncertainty are associated with pressure, temperature, and composition.
− Meter accuracy is estimated from calibration data, pipe size, or other installation effects.
− Sensitivity coefficients are obtained by dividing the calculated % uncertainty for an input variable by the
% change in that input variable.
c) Derive the combined uncertainty range (extended standard deviation)
− Combined uncertainty is calculated by summing the square of the errors for pressure, temperature,
composition, meter calibration, and installation effects.
− Range of Uncertainty is obtained from the square root of this combined sum.

Variability in flare composition may also be a significant factor in determining the measurement
uncertainty of an FFMS. Knowledge of flare composition may have a major effect on the calculation of
the actual volume, standard volume, or mass measured by the meters. For example:

− For a differential pressure meter – the output is a function of the square root of the flare gas
density.
− For thermal flow meters – knowledge of the actual volume (or standard volume) requires
consideration of the compositional effect on thermal conductivity and dynamic viscosity.
− For ultrasonic flow meters – sound speed is a function of gas composition.

Converting actual flow to mass flow requires the knowledge of gas composition in order to derive gas
density. When an analyzer is incorporated in the measurement system to correct for composition, care
must be taken to ensure that the response time of the system is short compared to the upset flow event
during flaring to ensure representative measurement during actual flaring.

An example of how all of these elements would be combined into the overall measurement uncertainty is
provided in Table 3-1.

Pilot Version, September 2009 3-6

 Table 3-1. Example of FFMS Combined Uncertainty a
COMBINED SENSITIVITY
VARIABLE (S x U95) 2
AND ERROR (S x U95)
Pressure 2.0% 4.00
Temperature 0.1% 0.01
Flare composition 2.0% 4.00
Meter error (calibration) 1.4% 1.96
Installation effects 0.5% 0.25
Sum of squares 10.22
Square root of sum of squares 3.2%
a
Table 4 from API MPMS Section 14.10 (reference 17)

3.1.3 Example: “Custody Transfer” Measurements

“Custody transfer” measurements are defined as measurements that provide quantity and quality
information, which can be used as the basis for a change in ownership and/or a change in responsibility
for materials. In most O&G producing jurisdictions around the world national regulations and directives
have emerged to specify requirements for the expected accuracy and uncertainty ranges associated with
“custody transfer” and other critical measurements. For such precise metering applications, the
flowmeters and adjacent piping used in the measurement system are expected to meet the requirements of
the relevant, preferably the most stringent, specifications of the API and ISO standards that are cited in
many national regulations.

One such example of the measurement requirements promulgated for O&G operations are those of the
Alberta Energy Resources Conservation Board (ERCB), as shown in Exhibit 3-3.

Pilot Version, September 2009 3-7

 EXHIBIT 3-3: ALBERTA ENERGY RESOURCE CONSERVATION BOARD (ERCB)
[Directive 017, May 2007]

a.) Directive 017 (EUB/Board, May 2007), spells out the measurement requirements for custody transfer within the Upstream
Oil and Gas operations in Alberta, Canada. The ERCB standards are stated as “maximum uncertainty of monthly volume”
and/or “single point measurement uncertainty”, as listed in Tables 3-2 for Oil Systems and Gas Systems measurements,
respectively.
b.) The uncertainties are to be applied as “plus/minus” (e.g., ±5%), and only measurements at the delivery or sales points are
required to meet the highest accuracy standards since they would have a direct impact on royalty determination.

Table 3-2. Summary of Alberta ERCB Accuracy Requirementsa

Maximum Uncertainty Single Point
of Monthly Volume Measurement
OIL SYSTEMS MEASUREMENTS
(i) Total battery oil (delivery point measurement)
Delivery point measures >100 m3/d N/A 0.5%
Delivery point measures < 100 m3/d N/A 1%
(ii) Total battery gas (includes produced gas that is
vented, flared, or used as fuel)
> 16.9 103 m3/d 5% 3%
> 0.50 103 m3/d but < 16.9 103 m3/d 10% 3%
< 0.50 103 m3/d 20% 10%
(iii) Well oil (proration battery)
Class 1 (high), > 30 m3/d 5% 2%
Class 2 (medium), > 6 m3/d but < 30 m3/d 10% 2%
Class 3 (low), > 2 m3/d but < 6 m3/d 20% 2%
Class 4 (stripper), < 2 m3/d 40% 2%
(iv) Well gas (proration battery)
> 16.9 103 m3/d 5% 3%
> 0.50 103 m3/d but < 16.9 103 m3/d 10% 3%
< 0.50 103 m3/d 20% 10%

GAS SYSTEMS MEASUREMENTSa

(i) Gas deliveries (sales gas) N/A 2%
(ii) Hydrocarbon liquid deliveries
Delivery point measures >100 m3/d N/A 0.5%
Delivery point measures <100 m3/d N/A 1%
(iii) Plant inlet or total battery/group gas 5% 3%
(iv) Plant inlet or total battery/group condensate N/A 2%
(v) Fuel gas
> 0.50 103 m3/d 5% 3%
< 0.50 103 m3/d 20% 10%
(vi) Flare gas 20% 5%
(vii) Acid gas N/A 10%
(viii) Dilution gas 5% 3%
(ix) Well gas (well site separation)
> 16.9 103 m3/d 5% 3%
< 16.9 103 m3/d 10% 3%
(x) Well gas (proration battery) 15% 3%
(xi) Well condensate (recombined) N/A 2%
a
Note: Extracted from Section 1.8.1 and 1.8.2, respectively, in Reference 20

The directive makes it clear that other measurement points that play a role in the overall control and accounting process would
be subject to less stringent accuracy standards. These less stringent accuracy standards are designed to accommodate physical
limitations and/or the overall economics of achieving very stringent accuracy standards for each volumetric measurement.

Pilot Version, September 2009 3-8

3.2 Uncertainties of Flow Measurements for GHG Inventories
With the emergence of new mandatory reporting regulations and emission reduction compliance
obligations, new requirements are being promulgated for the accuracy of fuel flow measurements when
such flows are used to quantify GHG emissions. For example, according to the mandatory GHG
reporting regulations promulgated by the California Air Resources Board (CARB), flow measurement
uncertainties are expected to be + 5%. (CARB, 2008) The California requirements specifically apply to
all GHGs emitted from petroleum refineries, hydrogen plants, and cogeneration plants with the O&G
exploration and production sector required only to report CO2 emissions from large combustors (>
25,000 tonnes CO2E/yr).

In Europe, the European Union Emissions Trading System (EU-ETS) specifies a tiered approach for
emission calculations together with required uncertainty ranges, according to the Monitoring and
Reporting Guidelines (EU-ETS MRG, 2007). It sets up a matrix of uncertainty requirements for different
facility sizes and measurement approaches used.

− For facilities emitting between 50,000 to 500,000 tonnes of fossil CO2 – Uncertainty ranges
specified are ±7.5%, +5%, and ±2.5% when the facilities employ tiers 1, 2, and 3 calculation
approaches, respectively;

− For facilities emitting over 500,000 tonnes of fossil CO2 – Uncertainty ranges are expected to
be as low as ±1.5% for facilities required to employ Tier 4 approaches.

It is important to note that the EU-ETS requirements are applicable to a limited set of O&G industry
installations and that facility inventories are being tracked only for CO2 emissions from fuel combustion
and flaring. The requirement to quantify these sources within such tight uncertainty ranges is a reflection
of the fact that these are the sources for which appropriate emission calculation methods are available,
while they are also the largest emission sources and the key contributors to most installations’ GHG
emissions.

When measuring fuel flow rate, or its total volume, and using the information to calculate GHG
emissions, it must be determined whether the flow meters used are properly installed and calibrated, and
that they are capable of providing data that are within the uncertainty ranges required by the governing
climate program. Differences must be considered between the manufacturers’ specifications of flow
meters’ expected measurement errors and those that are attained when using the flow meters in the field.
It is common practice to test flow meters in a laboratory setting under controlled conditions, prior to field
installations. However, these laboratory bench tests typically do not simulate ”real world” variations in
fluid flow and other possible fluctuations, and drift of the entire measurement system. For any given

Pilot Version, September 2009 3-9

operating facility, there might only be a very limited number of “custody transfer” meters that are
equipped for testing and calibration under real operating conditions.

As an example, Table 3-3 provides a listing of different meter types, their applicable fluid medium, and a
brief description of their operating principles. The table also lists manufacturers’ specified instrument
errors, as provided in a survey conducted by the EU-ETS Support Group (ETSG, 2007). The information
provided is an indication of potential error ranges and not the expected uncertainty ranges obtained in
practice. Appendix B provides additional details about the manufacturers’ recommendations for the
installation, calibration, and maintenance of these flow meters. The error levels cited refer primarily to
random errors that are observed under ‘ideal’ laboratory conditions and that decrease with repeated
measurements. They do not properly account for systematic errors (or bias) where the errors are due to
improper installations, inadequate calibrations, or devices drift, as discussed above.

However, these manufacturers’ specified measurement errors might not be attainable due to the practical
operational limitations of the facility. In most O&G industry facilities, detailed inspections, maintenance,
and recalibration of process control flowmeters are possible only once every few years when process units
are shut down for scheduled turnaround.

Table 3-3. Compilation of Specifications for Common Flow Meters a

METER MEDIUM TECHNICAL DESCRIPTION MANUFACTURERS’
TYPE REPORTED ERRORS B
Rotary Gas The rotary flow meter is a type of positive 0-20% of the measurement
meter displacement (PD) flow meter that is widely used for range: 3%
utility measurements of gas flow.
(Expected Rotary flow meters have one or more rotors that are 20-100% of the
life span: used to trap the fluid. With each rotation of the rotors, measurement range: 1.5%
25 years) a specific amount of fluid is captured. Flow rate is
proportional to the rotational velocity of the rotors.
Rotary meters are used for industrial applications.
Turbine Gas Turbine flow meters have a rotor that spins in 0-20% of the measurement
flow meter proportion to flow rate. Many of those used for gas range: 3%
flow are called axial meters. Axial turbine meters have
(Expected a rotor that revolves around the axis of flow. Axial 20-100% of the
life span: meters differ according to the number of blades and the measurement range: 1.5%
25 years) shape of the rotors. Turbine meters are used as billing
meters to measure the amount of gas used at
commercial buildings and industrial plants.
Bellows Gas The bellow gas meter performs volumetric 0-20% of the measurement
meter measurement via its bellows. The measurements are range: 6%
based on the principle that the flexible bellows is 20-100% of the range: 4%
(Expected periodically filled and emptied.
life span: A major problem with the bellows system is the
25 years) residue in the pipe. The internal mechanisms fail to
perform their tasks due to such residue, causing the
meter to dysfunction and failing to perform a sound
measurement.

Pilot Version, September 2009 3-10

 Table 3-3. Compilation of Specifications for Common Flow Meters, a continued
METER MEDIUM TECHNICAL DESCRIPTION MANUFACTURERS’
TYPE REPORTED ERRORS B
Venturi Gas and Venturi meters are another example of differential 20-100% of the
meter Liquid pressure flow meters, as described under orifice measurement range: 1.5%
meters above.
(Expected In this case, the primary element is a Venturi flow
life span: 30 nozzle. Venturis are especially suited to high-
years) speed flows. They are also used for custody
transfer of natural gas.
Orifice Gas and Orifice meters belong to the category of 30-100% of the
meter Liquid differential pressure flow meters that consist of a measurement range: 1.5%
differential pressure transmitter, together with a
(Expected primary element, such as the orifice plates.
life span: 30 The orifice plates place a constriction in the flow
years) stream, and the differential pressure transmitter
measures the difference in pressure upstream and
downstream of the constriction. The transmitter
or a flow computer then computes flow using
Bernoulli’s theorem.
Orifice plates are the most widely used type of
primary elements. Their disadvantages are the
amount of pressure drop caused, and the fact that
they can be knocked out of position by impurities
in the flow stream. Orifice plates are also subject
to wear over time.
Ultrasonic Gas and There are two main types of ultrasonic flow 1-100% of the measurement
meter Liquid meters: transit time and Doppler. The transit time range: 0.5%
meter has both a sender and a receiver. It sends
(Expected two ultrasonic signals across a pipe at an angle:
life span: 15 one with the flow, and one against the flow. The
years) meter then measures the “transit time” of each
signal. The difference between the transit times
with and against the flow is proportional to flow
rate. Doppler flow meters rely on having the
signal deflected by particles in the flow stream
and the frequency shift in proportion to the mean
fluid velocity.
Coriolis Gas and Coriolis flow meters contain one or more vibrating 1-100% of the maximum
meter Liquid tubes. These tubes are usually bent, although measurement range: 1%
straight-tube meters are also available. The fluid
to be measured passes through the vibrating tubes.
(Expected It accelerates as it flows toward the maximum
life span: 10 vibration point, and slows down as it leaves that
years) point. This causes the tubes to twist. The amount
of twisting is directly proportional to mass flow.
Position sensors detect tube positions.

Pilot Version, September 2009 3-11

 Table 3-3. Compilation of Specifications for Common Flow Meters, a continued
METER MEDIUM TECHNICAL DESCRIPTION MANUFACTURERS’
TYPE REPORTED ERRORS B
Vortex meter Gas Vortex flow meters are one of the few types of 10-100% of the
meters, besides differential pressure, that can measurement range: 2%
(Expected life accurately measure the flow of liquid, steam, and
span: 10 gas. Vortex flow meters operate on the von
years) Karman principle of fluid behavior, where the
presence of obstacles in the fluid path generates a
series of vortices called the von Karman street.
To compute the flow rate, vortex flow meters
count the number of vortices generated.
Gas Meter Gas An electronic device designed for the primary For 0.95-11 bar and -10 –
with purpose of converting a volume of gas measured 40°C: 0.5%
Electronic at one set of conditions to a volume of gas
Volume expressed at another set of conditions. The device
Conversion incorporates integral (internal or external)
Instrument temperature and/or pressure measurement
(EVCI) transducers. It may be directly mounted onto a
single meter (with mechanical drive or magnetic
(Expected life drive coupling) or connected to a remotely located
span: 10 meter from which it is fed volumetric pulses. The
years) device may perform additional functions such as
super compressibility correction, meter accuracy
curve correction (linearization), and energy
calculations.
Notes:
a
Based on material presented in Appendix I of the ETSG, July 2007 survey summary document and sources cited.
b
The error levels specified are those reported by the manufacturers when instruments are calibrated under laboratory conditions.

As indicated in Section 3.1.2 (Exhibit 3-2), a few key steps that should be followed to determine the
combined uncertainty associated with individual flow measurements. An expanded list of factors that
ought to be considered when evaluating the uncertainty of flow measurements that are used for GHG
emission calculations are provided in Exhibit 3-3.

Pilot Version, September 2009 3-12

 EXHIBIT 3-3: FACTORS TO CONSIDER WHEN EVALUATING UNCERTAINTY
OF FLOW MEASUREMENTS USED FOR GHG EMISSION CALCULATIONS
a) Confidence range of the measurement instrument
−
Manufacturers’ anticipated measurement errors for common flow meters could be used (Table 3-3) if on-site
calibration data are not available
b) Errors associated with “context-specific” factors
Such factors may include the following considerations:
− Are measurement instruments installed according to the manufacturer’s requirements?
− Is the measurement instrument designed for the medium (gas, liquid, solid substance) for which it is being
used?
− If manufacturer’s data are not available, are the instruments operated according to the general requirements
applicable to that measurement principle?
− Are there any other factors that can have adverse consequences on the uncertainty of the measurement
instrument? (i.e., how the measurement instrument is used in practice).
c) Pressure and temperature corrections for gas meters
− Pressure and temperature corrections are only applicable to the determination of the amount of gas and not to
the measurement of liquids or solid substances.
−
The actual amount of gas flow has to be corrected for pressure and temperature to the specified standard
conditions in order to avoid major systematic errors.
d) Determination of total uncertainties
− Individual uncertainties determined in a), b), and c) above ought to be summed up to determine the total
uncertainty of the individual quantity measured.

The discussion above pertains to single flow measuring device uncertainty. Additional analyses are
required to convert this individual measurement uncertainty to an assessment of the uncertainty range of
annual measurements. Detailed guidance on the calculation steps is provided in Section 4.

3.3 Uncertainties of Sampling and Analysis for GHG Emissions

The emission measurement process comprises either direct measurement at the source level of collection
of samples and their analysis in the laboratory to determine mass emissions. Sampling and analysis are
part of the same measurement process and their combined contribution to emissions estimation
uncertainty is obtained by their combined variances, as detailed in Section 4.0. Emission measurement
uncertainties for processes of sampling and analysis depend on random errors, measurement precision,
and systematic errors or bias.

3.3.1 Gaseous Samples Collection and Handling

Proper collection and handling of natural gas samples could have a major impact on the accuracy and
representativeness of the analytical measurements based on these samples. Analyses of gas samples are
used for multiple purposes and are applied to a variety of calculations including determination of heating

Pilot Version, September 2009 3-13

values, gas density and viscosity, hydrocarbon dew point, and compressibility. These analyses are
essential for obtaining information about the composition, including contaminants in the gas stream.
Calculations based on these analyses are key to optimization of process conditions, determination of
adherence to contractual specifications, or estimation of GHG emissions when such a stream is
combusted.

The API MPMS provides specific details for collecting and handling natural gas samples for critical
measurements such as custody transfer (API, February 2006). Exhibit 3-4 provides guidance on general
considerations of inaccuracies that might be introduced in the measurement system when collecting
samples that are used for carbon content and/or heating values determinations for GHG emissions. At the
same time, designing a sampling and analysis system should – in all cases – take into account regulatory
requirements and contractual obligations.

EXHIBIT 3-4: KEY FACTORS IMPACTING GAS SAMPLING AND ANALYSIS UNCERTAINTY
a) Inappropriate sampling techniques or equipment
− All sampling methods would require the use of a sample container for transporting the sample from the field
location to the laboratory
− Whenever practical, samples should be collected on a flow proportional or flow weighted basis, since spot
samples – by their nature – may not fully represent a gas stream of varying composition.
− Gaseous samples of interest are a mixture of organic and inorganic gases, and their integrity will be
compromised.
b) Inappropriate sample conditioning and handling
− Bias could be introduced if any components of a sample are either depleted or augmented during the
sampling, transport, or laboratory handling phases prior to analyses.
− Condensation and revaporization of hydrocarbons can cause significant distortions in the gas sample.
− Care should be taken to sample above the hydrocarbon dew point and/or to prevent retrograde condensation
when pressure is reduced during sampling.
c) Collection of samples from non-representative locations and/or under non-representative operating
conditions
− Sampling systems that are used in conjunction with on-line analyzers, such as chromatographs or
gravitometers, are typically designed to extract, condition, and deliver a representative sample to the analyzer.
− Sampling lines are kept as short as possible in conjunction with proper heating and insulation to avoid
condensation.
− The flow rate of the sampling system is adjusted to allow for close to real time data, while at the same time
not increasing the flow to a level that might lead to turbulence.
d) Inappropriate analytical methods
− The threshold sensitivity of the analytical methods used are typically those that are well documented by
industry recommended practices for sample families
− Analyses are typically limited to the range of mixture concentrations and species previously identified
− For complicated sample matrices, the potential for interferences is usually noted.

Pilot Version, September 2009 3-14

3.3.2 Quantifying Sampling Precision

Quantifying sampling precision requires that primary samples be collected according to a defined
protocol, but randomized in some way for each sample (in either space or time). For example, a pre-
selected percentage of the total number of samples can be collected in duplicate and the repeatability of
the measurement determined from these duplicate samples. Additionally, duplicate gaseous samples can
also be analyzed in duplicate and thus a full record of both sampling and analysis system variations can be
obtained. In this case, the sampling component of the variance represents the natural variability of the
sampling target as well as any errors in the sample collection and preparation.

When in situ measurement techniques are used (e.g., infrared gas analyzer), both the sampling and
analysis are addressed at the same time. The determination of sampling bias, or the difference between
the mean of several measurements and the true value, is more difficult. Biases could arise from several
causes such as sample fouling, inappropriate handling, or unrepresentative sampling. The “true” value for
the concentration of unknown species in a sample is never known since it is impossible to construct a true
reference laboratory standard for an unknown mixture of gases.

The importance of sampling uncertainty is a relatively new concept whose importance is slowly
beginning to be recognized. Recent research has shown that sampling uncertainty is often far greater than
analytical uncertainty. Therefore, combining sampling and analytical uncertainties to provide an estimate
of measurement uncertainty is an important component of quantifying the overall uncertainty of GHG
estimations that rely on sampling gaseous fuels of varying composition.

3.4 Carbon Content Measurement Practices

Different types of gas chromatography (GC) systems are normally used to analyze the carbon content of
gaseous streams. The GC systems might be laboratory based or set up as an online device for automated
collection of samples and their analysis. The systems are typically set up to analyze the individual
components in the sampled gas and provide detailed reports of properties including composition, calorific
value, and density.

The results of the determination of individual – or groups – of carbon-containing species are then used to
assess the total emissions of CO2 upon combustion of such a fuel. Several key considerations include:

− If the method is capable of determining CO2 content with the rest of the carbon containing
species, no further correction of the carbon content data is required in order to properly account
for all CO2 emissions;

Pilot Version, September 2009 3-15

 − If the method is set up to provide information only on hydrocarbon species, the CO2 content
should be obtained by an independent measurement and added to the fuel carbon content data;

− If the method is capable of a quantitative determination of CH4 content, these data can be used
separately for calculating evaporative and processing leaks along with venting losses; and

− All carbon content measurement data should be used in conjunction with the applicable fuel flow
measurements when calculating emissions.

3.4.1 Laboratory-Based Measurements

Several ASTM and ISO methods are available for determining the composition and carbon content in the
natural gas range as well as for liquid and solid fuels. Although the terms “repeatability” and
“reproducibility” are applicable to all measurement situations (see Appendix A for terminology and
definitions), they are used here in the context in which they are defined in ASTM standards:

− Repeatability is defined as the difference between two successive results obtained by the same
operator with the same apparatus under constant operating conditions on identical test materials,
and

− Reproducibility is the difference between two results obtained by different operators in different
laboratories on identical test materials.

When using a natural gas measurement method for refinery fuel gas, care should be taken to ensure that the
range of compositions of individual components is within the bounds specified by the method. Table 3-4
lists selected commonly used laboratory measurement methods that are frequently used for the
determination of fuel carbon content. Additional details about these methods and further discussions of
their main features are provided in Appendix D.

Pilot Version, September 2009 3-16

 Table 3-4. Summary of Selected Carbon Content Measurement Methods
METHOD TITLE BRIEF DESCRIPTION MEASUREMENT
PRECISION a
ASTM 1945-03 Analysis of Natural Gas − Covers the determination of the − Repeatability ranges from
by Gas Chromatography chemical composition of natural 0.01- 0.10 mole%
gases and similar gaseous − Reproducibility ranges
mixtures within specified ranges from 0.02-0.12 mole%
of applicable composition for
individual components
ASTM 1946-90 Analysis of Reformed − Used for determination of − Repeatability ranges from
Gas by Gas chemical composition of reformed 0.05 - 0.5 mole%
(Reapproved 2006) Chromatography gases and similar gaseous − Reproducibility ranges
mixtures from 0.1 - 1.0 mole%
− Applicable to mixtures containing:
hydrogen, oxygen, nitrogen,
carbon monoxide, carbon dioxide,
methane, ethane, and ethylene
ASTM UOP539-97 Refinery Gas Analysis by − Used for determining the − Quantification from 0.1-
Gas Chromatography composition of refinery gas 99.9 mole% for a single
samples or expanded liquefied component or composite
petroleum gas (LPG) samples − For hydrogen sulfide,
obtained from refining processes quantitative results
or natural sources between 0.1-25 mole%
ISO 6974 Natural Gas - − Describes a gas chromatographic − Relative repeatability:
Determination of method for the quantitative 2% for species < 1.0mole%;
(Six parts) composition with determination of the content of
0.8% for species 1-50mole%
Defined Uncertainty by hydrogen, helium, oxygen,
Gas Chromatography nitrogen, carbon dioxide and C1 to − Relative reproducibility:
C8 hydrocarbons in natural gas 4% for species < 1.0mole%
samples 1.6% for species 1-50mole%
− Uses either two or three packed or
capillary columns combinations
ASTM D2650-04 Standard Test Method − Applicable for the quantitative − NOT applicable for
for Chemical analysis of gases containing constituents < 0.1
Composition of Gases by specific combinations of the mole%.
Mass Spectrometry following components: hydrogen; − Developed on a specific
hydrocarbons (up to 6-Csper type of MS
molecule); CO; CO2; mercaptans
(1-2 Cs per molecule); H2S; and − Users have to modify for
air (N2, O2, and Ar) their instrument.

ASTM D5291 - 02 Standard Test Methods − Applicable to samples such as − NOT recommended for
for Instrumental crude oils, fuel oils, additives, and the analysis of volatile
(2007)
Determination of residues for carbon and hydrogen materials such as
Carbon, Hydrogen, and and nitrogen analysis gasoline, gasoline-
Nitrogen in Petroleum − Tested in the concentration range oxygenate blends, or
Products and Lubricants of at least 75 - 87 wt% carbon, at gasoline type aviation
least 9 - 16 wt% hydrogen, and turbine fuels
0.1 - 2 wt% nitrogen
a
Measurement precision data is provided as an indication of attainable precision if all steps of the methodology are adhered to as described in the methods
procedures.

Pilot Version, September 2009 3-17

3.4.2 On-line Measurements
On-line determination of fluid stream compositions is quite challenging due to possible variations in these
compositions. This is especially notable for self-generated fuel gas such as refinery fuel gas or other
processing plant gas. Conversely, for commercial products such as natural gas, liquid fuels, coal and
coke, or for the analysis of associated gas in exploration and production operations, the challenges are
more related to the ability to analyze multiple streams rapidly and ascertain that they all are within a
desired property range.

Instrumentation in this field has been developed to provide a measurement of stream components in order
to achieve optimum control and assure product quality. The configurations of such analyzers are
customized to accommodate typical site parameters and operating practices. Most such analyzers are
designed with ASTM and ISO standards in mind and their calibration routines are designed to provide
both reported data and its associated uncertainty. As mentioned previously, the two primary applications
are for refinery gas analyzers and natural gas analyzers, and these are discussed briefly below.

a. Refinery Gas Analyzer

Refinery gas samples are delivered to the sample inlet of the GC after passing through a sample
conditioning system that selectively removes any liquid fractions and particulate matter from the
sample. This ensures that only the gas phase sample is delivered to the analyzer. An internal vacuum
pump draws this conditioned sample into each injector, which then injects the mixture onto each of
the columns for analysis. Typically, a complete analysis of hydrogen, saturated and olefinic
hydrocarbons (C1-C5, and C6+ grouped peaks), plus fixed gases (O2, N2, CO, and CO2) is performed.
Precise “retention times” information and component areas translate into accurate component
identification and quantification of the relative magnitude of individual components present in
refinery gas.

b. Natural Gas Analyzer

These types of analyzers are applicable to natural gas samples from wellhead to pipeline-quality gas.
Samples are introduced using sample cylinders, Tedlar bags, or by direct connection to the pipeline or
wellhead sampling points. Usually, two chromatographic modules are used to quickly separate and
measure the individual components in natural gas. The analyzer separates and measures the
permanent gases and hydrocarbons present via an optimized, dual-channel portable gas
chromatograph. Wellhead samples may often contain significant amounts of H2S. Many instruments
take this into account and there are no interferences, which means that H2S can be measured from 50
PPM to 30-mole%.

Pilot Version, September 2009 3-18

3.5 Heat Content Determination

The heating value or calorific value of a substance is the amount of heat released during the combustion
of a specified amount of the substance. The calorific value is a characteristic for each substance and is
measured in units of energy per unit of the substance, such as: kcal/kg, kJ/kg, J/mole, Btu/m³. The heat
of combustion for fuels is expressed as:

− HHV – higher heating value (or gross calorific value or gross energy or upper heating value).
This value is determined by bringing all the products of combustion back to the original pre-
combustion temperature, and in particular condensing any water vapor produced.

− LHV – lower heating value (or net calorific value) is determined by subtracting the heat of
vaporization of the water vapor from the higher heating value and treating any water formed as a
vapor. The energy required to vaporize the water therefore is not realized as heat.

In the O&G industry, the two most prevalent modes of determining heating values of gaseous fuels is
either by measuring it directly, which can be accomplished either by stoichiometric combustion or by
calorimetric techniques, or by computational methods that are based on standardized calculation
procedures using gas sample composition data. A brief summary of these two types of practices is
provided below.

3.5.1 Direct Measurements

The heating value indicates the amount of energy that can be obtained as heat by burning a unit of gas.
The heating values of a gas depend not only upon the temperature and pressure, but also upon the degree
of saturation with water vapor.

As mentioned above, general practices for determining fuel gas heating values rely on either calorimetric
techniques or stoichiometric combustion practices. Table 3-5 provides a listing and a brief description of
some selected methods for heating value determination for gaseous, liquid, and solid fossil fuels. Further
discussion of the main features of these methods can be found in Appendix C.

Pilot Version, September 2009 3-19

 Table 3-5. Summary of Selected Heating Value Measurement Methods

METHOD TITLE BRIEF DESCRIPTION MEASUREMENT

PRECISION (*)
ASTM D4891 – 89 Test Method for Heating − Used for the determination of − Repeatability: 0.76 Btu/scf,
(Reapproved 2006) Value of Gases in Natural heating value of natural gases and 95% confidence: 2.1
Gas Range by similar gaseous mixtures within a Btu/scf.
Stoichiometric Combustion specified composition range − Reproducibility: 1.67
− Provides an accurate and reliable Btu/scf, 95% confidence:
procedure for regulatory 5.1 Btu/scf.
compliance, custody transfer, and − Average bias: within 0.1%
process control from reference value
ASTM D1826 – 94 Standard Test Method for − Used for the determination of the − Weekly standardization with
(Reapproved 2003) Calorific (Heating) Value of total calorific (heating) value of methane
Gases in Natural Gas Range fuel gas produced or sold in the − Errors < 0.5% within a week
by Continuous Recording natural gas range from 900-1200 after standardization
Calorimeter Btu/scf
− Higher errors expected for
− Provides a reliable method for longer standardization
measurement on a continuous basis periods
with a recording calorimeter
ASTM D7314 – 08 Standard Practice for − Used to determine the heating − No generic precision data
Determination of the value of gaseous fuels with at-line apply here since this is a
Heating Value of Gaseous and in-line instruments practice and not a method
Fuels using Calorimetry − Suitable for periodic operation on a − The installation and
and On-line/At-line continuous basis operation of particular
Sampling systems vary with process
− Suitable for monitoring systems
for tracking properties when using type, performance and
or producing gaseous fuels in regulatory requirements
industrial processes
ASTM D4809 – 06 Standard Test Method for − Covers the determination of the − Strict adherence to all
Heat of Combustion of heat of combustion of hydrocarbon details of the procedure is
Liquid Hydrocarbon Fuels fuels essential
by Bomb Calorimeter − Can be used for a wide range of − The error contributed by
(Precision Method) volatile and nonvolatile materials each individual
where slightly greater differences measurement that affects the
in precision can be tolerated precision shall be < 0.04 %,
− Under normal conditions, the insofar as possible
method is directly applicable to
such fuels as gasoline, kerosenes,
Nos. 1 and 2 fuel oil, Nos. 1-D and
2-D diesel fuel and Nos. 0-GT, 1-
GT, and 2-GT gas turbine fuels
ASTM D5865 – 07a Standard Test Method for − Pertains to the determination of the
Gross Calorific Value of gross calorific value of coal and
Coal and Coke coke by either an isoperibol or
adiabatic bomb calorimeter
(*) Measurement precision data are provided as an indication of attainable precision if all steps of the methodology are adhered to as described in the
methods procedures.

Pilot Version, September 2009 3-20

3.5.2 Computational Methods
Heating value may be determined from gas compositional analysis in accordance with a standard practice
established by the ASTM for calculating heating values for natural gas and similar mixtures from
compositional analysis. ASTM D3588 – 98 (Reapproved 2003), is the standard recommended practice
for calculating heating values, compressibility factors, and relative densities of gaseous fuels (ASTM
D3588-98, 2003). This practice covers procedures for calculating these quantities at base conditions
(14.696 psia and 60°F or15.6°C) for natural gas mixtures from compositional analysis. It applies to all
common types of utility gaseous fuels, i.e., dry natural gas, reformed gas, oil gas (both high and low Btu),
propane-air, coke oven gas, and other gaseous fractions for which suitable methods of analysis are
designated.

The ideal gas heating value and ideal gas relative density are calculated from the molar composition and
the respective ideal gas values for the components; these values are then adjusted by means of a
calculated compressibility factor.

The ASTM approach recognizes that the calorific value of a mixture, such as in refinery fuel gas systems,
is a function of the mol% composition of the individual components of that mixture. For refinery fuel
gas, this mixture could contain both carbon containing species, which upon combustion will contribute
directly to CO2 emissions, as well as non-carbon containing species. Hydrogen, for example, is an
important contributor to refinery fuel gas heating values but it does not contribute to CO2 emissions when
combusted.

To implement this practice, the user would first determine the molar composition of the gas in accordance
with any applicable ASTM or GPA method. For a precise calculation, at least 98 % of the sample
constituents should be determined as individual components, with no more than a total of 2 % in terms of
groups of components (e.g. butanes, pentanes, hexanes, and so forth).

An ideal combustion reaction for fossil fuels (that may contain hydrogen) in the ideal gas state can be
generally represented as:

⎛ b c⎞ ⎛b ⎞
Ca H b + cH2 + ⎜ a + + ⎟ O2 = aCO2 + ⎜ + c ⎟ H 2 O (Equation 3-1)
⎝ 4 2⎠ ⎝2 ⎠

The ideal net heating value is the heating value that is observed when all the water remains in the ideal
gas state, while the ideal gross heating value is observed when all the water formed by the reaction
condenses to liquid. The difference between them is the enthalpy of vaporization of the water formed
during the combustion process. Therefore, the ideal gross heating value for a mixture can be expressed
as:

Pilot Version, September 2009 3-21

 n

∑x
j =1
j × M j × H mj
Hm = n
(Equation 3-2)
∑x
j −1
j ×Mj

where
xj = the mole fraction of Component j;
Mj = the molar mass of Component j;
n = the total number of components; and
Hmj = the ideal gross heating value per unit mass for Component j (at 60°F or 15.6°C), as tabulated
in ASTM D3588. Values of Hm are independent of pressure, but they vary with temperature.

Errors that should be considered when computing heating values include errors in the heating values of
the components and in the composition data. The uncertainty ranges of the heating values for the
components cited in the ASTM practice are about 0.03%. These errors may affect the agreement between
calculated and measured heating values; however, those are negligible when compared to the errors
associated with the determination of the composition of individual species. Appendix D contains a listing
of common energy and fossil fuel unit conversion factors that could be used for these and other
calculations discussed in these document.

3.6 Laboratory Management System

An additional consideration for uncertainty of GHG measurements is the credibility and technical veracity
of the laboratory performing the requested tests, as specified by different programs. For example, in the
EU-ETS program the MRGs require the demonstration of laboratories management systems (EU-ETS
MRGs, 2007; Section 13.5). Any laboratory used to determine an emission factor, calorific value, carbon
content, or composition data should be accredited according to ISO 17025:2005. If non-accredited
laboratories are used, the EU regulations provide specific requirements for the validation and testing of
such laboratories.

ISO/IEC 17025:2005 (ISO/IEC 17025, 2005) is not intended to be used for overall laboratory
certification, and it does not address compliance with regulatory and safety requirements. It merely
emphasizes the need for a well-developed and communicated laboratory management system that
addresses areas such as: quality, administrative procedures, and technical systems that govern the
operations of the laboratory. The ISO standard highlights the need for laboratories to ensure that their
personnel are aware of the relevance and importance of their activities to the achievement of the
management system’s objectives.

The standard specifies the general requirements for demonstrating competence to carry out specific tests
and calibrations, including field sampling. It covers testing and calibration procedures that are performed

Pilot Version, September 2009 3-22

using standard methods, non-standard methods, and laboratory-developed methods. This standard may be
applied to all organizations performing tests and/or calibrations, including both internal companies’
laboratories as well as external contract laboratories.

ISO/IEC 17025:2005 is applicable to all laboratories regardless of the number of personnel or the extent
of the scope of testing and/or calibration activities. When a laboratory does not undertake one or more of
the activities covered by ISO/IEC 17025:2005, such as sampling and the design/development of new
methods, the requirements of those clauses would not apply.

As part of compliance with ISO/IEC 17025:2005, laboratories that want to be accredited to this standard
are mandated to seek feedback, both positive and negative, be it from in-house users or from external
customers. The information gathered is expected to help laboratories improve their management systems,
their testing and calibration activities, and customer service. The ISO standard seeks to improve
laboratory measurement proficiency and accuracy by continual improvement of the effectiveness of
laboratory management systems and the implementation of quality policy, quality objectives, internal
audits, data analysis, corrective and preventive actions, and periodic management review.

Pilot Version, September 2009 3-23

4.0 STATISTICAL CONCEPTS AND CALCULATION METHODS
Uncertainty is used to characterize the dispersion of
values that could be reasonably attributed to a measured Section Focus
This section is intended for the user who has
quantity (IPCC, Annex 3, 2001). Uncertainty may be
already generated a GHG emission inventory for a
expressed as a qualitative ranking, such as the letter facility or entity. It links the statistical concepts
introduced in earlier sections to their application
ratings assigned in EPA’s AP-42 publication series (U.S.
for measurement uncertainty and error
EPA, 1995 with Supplements through 2000), or as a propagation. Decision trees are provided to guide
the user through quantifying uncertainty for each
quantified value. For the purpose of quantifying the
part of a GHG inventory and for propagating the
uncertainty of a GHG inventory, this section first uncertainty to the emission inventory total. A
general discussion on reducing uncertainty is also
addresses measurement uncertainty, then discusses
presented. The subsections include:
uncertainty associated with emission factors, and finally
ƒ Measurement uncertainty;
addresses the propagation of uncertainty.
ƒ Uncertainty propagation;
4.1 Measurement Uncertainty ƒ Quantifying emission estimation uncertainty;
At the most basic level, a GHG inventory is comprised of ƒ Quantifying measurement uncertainty;
estimated emissions from individual emission sources. ƒ Aggregating uncertainties; and
For a given emission source, an emission estimate ƒ Assessing data correlations.
generally consists of an emission factor and some
measure of the activity that results in the emission (referred to as the activity factor; see also Sections 2.3
and 2.4). Emissions from multiple sources are then aggregated to produce the inventory. The
quantification of uncertainty should be applied at the emission source level (or grouping of similar
emission sources) and then propagated to the total inventory (as discussed in Section 2.5).

Activity factors are generally a measured quantity, such as a count of equipment or measure of fuel
consumed. Emission factors may be either based on site-specific measurements or based on published
values that were derived from averaging a variety of measurements. Where measurements are used for
either activity factors or emission factors, two components of uncertainty need to be considered: precision
and bias.

4.1.1 Precision and Bias

Precision refers to the variability in the measurement. A method may produce results that are not very
precise (having high variability), but result in a good estimate on average, as shown in Figure 4-1. The
variability or precision associated with such an estimate will decrease as more data points are collected.

Pilot Version, September 2009 4-1

 0.6
0.4
0.2
Error
0
-0.2
-0.4
-0.6
0 10 20 30
Time

Figure 4-1. Measurement Error Over Time of an Unbiased Estimate

Bias, on the other hand, refers to how accurately a method estimates the true value. An example of bias
would be a meter that is not calibrated correctly and consistently overestimates the measurement. Ideally,
the data measurement scheme should be designed in a way to minimize bias. For example, if the
measuring device is well maintained and calibrated to minimize drift, there may be no bias in the
measurement.

Figure 4-1 shows an example of the error of an unbiased measurement over time, where the errors are
centered around zero. When these measurements are aggregated, some of the positive errors are offset by
some of the negative errors, resulting in a lower estimate of uncertainty for the aggregate measurement.

Figure 4-2 shows an example of the error in a biased measurement over time. Similar to the unbiased
case in Figure 4-1, some of the high errors are offset with some of the low errors so the uncertainty in the
precision of the estimate is lower for the aggregate. However, in Figure 4-2, the errors are not centered
on zero as in Figure 4-1; they are centered on five. This means that there is a bias in the estimate. Unlike
precision, the uncertainty due to the bias generally does not decrease as more data points are collected.
Thus, precision and bias should be considered separately, if possible.

Pilot Version, September 2009 4-2

 6

Error 3

0
0 10 20 30
Time

Figure 4-2. Measurement Error Over Time of a Biased Estimate

If bias can be quantified, it should be corrected and thus eliminated from consideration in quantifying
uncertainty. For example, if a fuel stream has two types of measurement devices, data can be collected
from both devices to check for agreement. A bias would be indicated if the measurements differed.
However, quantifying and correcting for bias might not always be practical since it often requires
application of frequent calibration routines and implementation of quality control procedures that would
allow instrument adjustments and/or corrections under prescribed conditions. It will also require proper
quantification routines that account for drift or other causes of bias between calibrations on an ongoing
basis.

The general approach for quantifying bias would depend on prior experience in the laboratory or from
specifically designed field measurement campaigns. Measurement bias can vary from small to very large,
depending on the application, and can even change over time if the measurement instrument is allowed to
drift without calibration. Therefore, in practice, bias will most commonly be determined using expert
judgment and will be based on such parameters as the length of time since the equipment was calibrated
and other factors that would cause the measurement to systematically overestimate or underestimate the
true value.

4.1.2 Confidence Intervals

Uncertainty is commonly expressed in terms of confidence intervals, where the confidence intervals
establish the lower and upper tolerances associated with an estimated number. Expressed as an absolute
value, the confidence interval is computed as:

Pilot Version, September 2009 4-3

 s
±t× (Equation 4-1)
n
where
s = standard deviation;
n = sample size; and
t = t-value for “n-1” degrees of freedom.

and

1 n
s= ∑
n − 1 i =1
( xi − x ) 2 (Equation 4-2)

where
xi = the ith observation in the data set and
x = the mean of the data set.

∑x i
x= i =1
(Equation 4-3)
n
Tables for the Student’s t-distribution can be found in most basic statistics references. Most spreadsheet
software programs have a function that will calculate the necessary t-value. This is the preferred method
since the software generally retains more significant digits for the t-value than a look-up table would
display.

4.2 Overview of Uncertainty Propagation

Uncertainty propagation involves mathematically combining individual sources of uncertainty to establish
an estimate of the overall uncertainty. Specific uncertainty propagation techniques are discussed in
Section 4.2.1.

The following three assumptions are important when applying the uncertainty propagation technique for
overall uncertainty assessment (IPCC, Section A1.4.3.1, 2001):

1. The uncertainties are relatively small, which is defined as the standard deviation divided by the
mean value being less than 0.3;
2. The uncertainties have Gaussian (normal) distributions; and
3. The uncertainty values (i.e., the errors or uncertainties associated with the measured data or
reported values) are mutually independent.
In many cases, the first assumption may be difficult to meet. For example CH4 and N2O emissions often
have very sparse data and large associated uncertainties. Conducting a Monte Carlo simulation
(discussed further in Section 4.6.1) is an option if the standard deviation divided by the mean is greater
than 0.3. However, Monte Carlo simulations require a significant level of detail for the description data

Pilot Version, September 2009 4-4

to characterize the probability distributions. Without such information, the potential error introduced
from incorrectly specifying the distributions for a Monte Carlo simulation could outweigh the potential
error that might be associated with applying an uncertainty propagation technique for sources with large
uncertainties. Therefore, this document suggests that the first assumption can be relaxed for emission
estimates with a small overall contribution to the GHG inventory. Through the propagation of uncertainty
for all emissions in the inventory, the impact of small emission sources with large uncertainties can be
evaluated relative to the entire inventory. This evaluation can be used to identify and prioritize emission
sources that require more data to reduce the overall uncertainty of the inventory.

The second assumption is based on the normality of the distribution of the underlying source data (i.e.,
symmetrical around the mean). According to the Central Limit Theorem, for a large enough sample size
(n>30), we can relax the normality assumption but still assume that the sampling distribution of the
sample means is normally distributed (Casella and Berger, 1990). Hence, if the calculated uncertainty is
based on statistical sampling of the population, one would need to obtain more samples to approach
normality. Alternatively, we might consider data transformation, i.e., mathematically transforming the
data to a different scale and using that transformed ‘normal’ distribution to derive the 95% confidence
interval. For example, in the case where the data distribution is skewed and the uncertainty is > 100% of
the mean (i.e., where the lower limit would be less than zero), the data could be transformed to a
lognormal distribution. This approach, however, requires the confidence interval to be transformed back
to the original scale to express the uncertainty in the original units, which can introduce error. As a result,
there is a trend away from using transformational approaches due to issues in transforming the data back
to their original scale.

The third assumption states that there is no significant covariance between the uncertainties that are to be
combined, which is equivalent to saying that the errors or uncertainties are independent or that there is no
correlation between the uncertainty terms. The uncertainties in two quantities would be considered
independent if they were estimated by entirely separate processes and there was no common source of
uncertainty. The uncertainties in two quantities would be dependent if they had a common source of
uncertainty (Williamson, 1996). Covariance between two uncertainty terms can be addressed through an
additional term in the uncertainty propagation equations (discussed further in Section 4.2.2). However,
the IPCC Good Practices document suggests avoiding the need for the covariance term in the equation by
“…stratifying the data or combining the categories where the covariance occurs” (IPCC, 2001).

Pilot Version, September 2009 4-5

4.2.1 Propagation Equations
There are four general equations for propagating uncertainty that are used in this document and the API
Compendium for compiling the uncertainty associated with a GHG inventory. A general introduction to
the equations is presented here, with example applications provided in Sections 4.3, 4.4, and 5.

Consider two quantities that can be measured: X and Y. The uncertainty for these values can be
expressed on an absolute basis as ±Ux and ±Uy, respectively, where U is calculated through statistical
analysis (as represented by Equation 4-1), determined through the Monte Carlo technique, or assigned by
expert judgment. Uncertainty may also be expressed on a relative basis, generally reported as percentage:

⎛U ⎞ ⎛U ⎞
± 100⎜ X ⎟ % or ± 100⎜ Y ⎟ %, respectively.
⎝ X ⎠ ⎝ Y ⎠

Depending on the uncertainty propagation equation, the absolute or relative uncertainty value may be
required. In addition, selection of the propagation equation also depends on whether the uncertainties
associated with the individual parameters are independent or correlated.

a. Uncertainty Propagation for a Sum (or Difference)

Two potential equations are used for computing the total uncertainty from the addition or
subtraction of two or more measured quantities. The selection between the two equations
depends on whether the uncertainties associated with the measured quantities, X and Y, are
correlated.

For uncertainties that are mutually independent, or uncorrelated (i.e., the uncertainty terms are not
related to each other), the aggregated uncertainty is calculated as the “square root of the sum of
the squares” using the absolute uncertainties, as shown in Equation 4-4.

U (abs ) X +Y +...+ N = U X2 + U Y2 + ... + U N2 (Equation 4-4)

where
U(abs) = the absolute uncertainty.

The absolute uncertainty values are used in the equations, and the resulting aggregated
uncertainty (UX+Y+…+N) is also on an absolute basis. Note that where a constant is also included in
the emission estimation calculation, the absolute uncertainty should include the constant. This is
demonstrated in the example provided in Section 4.4.1.

For two uncertainty parameters that are related to each other, the equation becomes:

U (abs )Correlated X +Y = U X2 + U Y2 + 2r (U X × U Y ) (Equation 4-5)

Pilot Version, September 2009 4-6

 where
r = the correlation coefficient between UX, UY, (discussed further in Sections 4.2.2 and 4.6).

However, the IPCC Good Practices guidance states, “Once the summation exceeds two terms and
the covariance occurs, the use of the Monte Carlo approach is preferable where data resources are
available” (IPCC, 2001).

b. Uncertainty Propagation for a Product (or Quotient)

The equation for propagating uncertainties from the product or quotient of two or more measured
and independent quantities is similar to Equation 4-4. However, in this case the relative
uncertainties are used, as shown in Equation 4-6. When multiplied by 100, the resulting
combined uncertainty (U(Rel)XxYxN) is expressed as a percentage.

2 2 2
⎛U ⎞ ⎛U ⎞ ⎛U ⎞
U (rel ) X ×Y ×...× N = U (rel ) X ÷Y ÷...÷ N = ⎜ X ⎟ + ⎜ Y ⎟ + ... + ⎜ N ⎟ (Equation 4-6)
⎝ X ⎠ ⎝ Y ⎠ ⎝ N ⎠

Equation 4-7 is used to estimate the uncertainty of a product or quotient of two parameters (X and
Y) where the uncertainties are correlated and positive values. Here also, relative uncertainty
values are used in the equation and the resulting combined uncertainty is on a relative basis.

2 2
⎛U ⎞ ⎛U ⎞ ⎛U U ⎞
U (rel )Correlated X ×Y = ⎜ X ⎟ + ⎜ Y ⎟ + 2r ⎜ X × Y ⎟ (Equation 4-7)
⎝ X ⎠ ⎝ Y ⎠ ⎝ X Y ⎠

c. Combining Uncertainties

It may be necessary to combine multiple uncertainty parameters associated with a single

measured value, such as combining uncertainties for precision and bias. For uncertainty
parameters that are independent, the combined uncertainty is calculated using the absolute
uncertainties as shown in Equation 4-4. Similarly, for uncertainty parameters that are related to
each other, Equation 4-5 applies.

4.2.2 Correlation Coefficient

The correlation coefficient, r, used in Equations 4-5 and 4-7, is a number between -1 and 1 that measures
the linear relationship between the errors or uncertainties of two measured parameters. The value of r is
zero when the parameters are independent. As stated previously, once the uncertainty propagation
exceeds two terms and covariance occurs, the use of the Monte Carlo approach (described further in
Section 4.6.1) is preferable (IPCC, 2001). Additional details on calculating the correlation coefficient are
provided in Section 4.6. A simplified explanation follows.

Pilot Version, September 2009 4-7

For two terms that might be correlated, the errors or uncertainties are plotted against each other. For the
purpose of this discussion, Ux represents the uncertainties of one variable plotted along the x-axis, and Uy
represents the uncertainties of the second variable plotted on the y-axis. The correlation coefficient, r, is
determined by a linear regression of the Ux and Uy values.

If one suspects that the uncertainty parameters are correlated, but data are not available to plot or calculate
the correlation coefficient, the following rule-of-thumb values could be applied using expert judgment
(Franzblau, 1958):

o r = 0: no correlation, the data are independent

o r = ±0.2: weak correlation
o r = ±0.5: medium correlation
o r = ±0.8: strong correlation
o r = ±1: perfect correlation, the data fall on a straight line.

4.3 Quantifying Emission Estimation Uncertainty

For the purpose of these guidelines, the general steps for quantifying uncertainty are:

1. Determine the uncertainty for emission factor data;

2. Determine the uncertainty for measured data; and
3. Aggregate uncertainties.
The following provides two simple examples of uncertainty calculations for emissions estimation
methods common in the oil and gas industry.

4.3.1 Simple Emission Estimation: EF × AF

In constructing a facility or entity-wide GHG inventory, many of the emission estimates are based on a
simple multiplication of the emission factor (EF) by a measure of the activity (AF, or activity factor). The
following example demonstrates how uncertainty is determined for this type of emission estimate.

Pilot Version, September 2009 4-8

 EXHIBIT 4-1: Uncertainty Example for a Simple Emission Estimation
Input Data: A gas production facility operated 45 high-bleed pneumatic devices during the previous year.
The average production gas composition is 80% CH4 and 5% CO2.
Emission Factor: The API Compendium provides a default CH4 EF for high-bleed pneumatic devices of
4.941 tonne CH4/device-yr ± 33.1% (Table 5-15 of the 2009 API Compendium. Uncertainty is expressed at
the 95% confidence interval.)
Emission Estimate: Emissions for this source are estimated based on the following calculations.

4.941 tonne CH 4 80 mole % CH 4

CH 4 : (45 pneumatic devices) × × = 226 tonnes CH 4 /yr
device - yr 78.8 mole % CH 4
226 tonne CH 4 tonne mole CH 4 tonne mole gas
CO 2 : × ×
yr 16 tonne CH 4 0.80 tonne mole CH 4
0.05 tonne mole CO 2 44 tonne CO 2
× × = 38.8 tonnes CO 2 /yr
tonne mole gas tonne mole CO 2

Uncertainty Assessment:
Measurement Uncertainty:
o For this example, the activity data are based on a single point measurement. All of the devices were
accounted for, so there is no bias and the uncertainty of the activity value (i.e., the number of
pneumatic devices) is 0.
o The composition measurements are based on multiple measurements representing a sampling of the
composition. Equations 4-1, 4-2, and 4-3 are applied to calculate the standard deviation of the
composition samples collected throughout the year. The standard deviation accounts for the
measurement error and natural variability of the sampled values.
o For this example, the precision uncertainty of the composition data is assumed to be ± 1%. (A more
detailed demonstration of quantifying uncertainty for measured composition data is provided in a
separate example.) On an absolute basis, this equates to 0.8 for the CH4 composition, and 0.05 for
the CO2 composition.
o Bias associated with the sampling and analysis is assumed to be small due to equipment
maintenance and calibration practices. A value of 3 % is assigned for this assessment. On an
absolute basis, this equates to 2.40 for the CH4 composition, and 0.150 for the CO2 composition.

U Composition Data = U Bias

2
+ U Precision
2

U CH4 = 0.802 + 2.402 = 2.53

U CO2 = 0.052 + 0.1502 = 0.158

Emission Factor Uncertainty:
As noted above, an uncertainty of ± 33.1% was specified for the default emission factor. Any bias in the
default emission factor is accounted for in the associated uncertainty value.

Pilot Version, September 2009 4-9

 EXHIBIT 4-1: Uncertainty Example for a Simple Emission Estimation, continued
Uncertainty aggregation – The following calculations step through the aggregation of uncertainty for this
emission source. First, the uncertainty of the emission estimates for CH4 and CO2 are calculated by
applying the relative uncertainties to Equation 4-6.

U (rel ) Emission Estimate = U Number

2
of Devices + U Composition Data + U Emission Factor
2 2

2
⎛ 2.53 ⎞
U (rel )CH 4 = 0 +⎜2
⎟ + 0.331 = 0.333 = 33.3%
2

⎝ 80 ⎠
2
⎛ 0.158 ⎞
U (rel )CO2 = 0 +⎜
2
⎟ + 0.331 = 0.333 = 33.3%
2

⎝ 5 ⎠

Next, the CH4 emissions are converted to a CO2 equivalent basis, since there are no additional emission
sources to sum for this example.
CO 2 e = ( 226 tonnes CH 4 /yr × 21) + 38.8 tonnes CO2 /yr
= 4, 740 + 38.8 = 4,780 tonnes CO 2 e/yr

Note that the global warming potential (GWP) of CH4 is treated as a constant.
The absolute uncertainty for the aggregated CO2 equivalent emissions are based on Equation 4-5 since the
uncertainties for both CH4 and CO2 are perfectly correlated (r = 1).

U (abs )Correlated X +Y = U X2 + U Y2 + 2r (U X × U Y )

( 4, 740 × 0.333) + ( 38.8 × 0.333) + 2r ( ( 4, 740 × 0.333) × ( 38.8 × 0.333) )

2 2
U (abs )Correlated X +Y =

U (abs )Correlated X +Y = 1,5802 + 12.92 + 2 × 1 × (1,580 × 12.9 ) = 1,590

On a relative basis, this equates to 1,590/4,780 = 33.3%

The final emission estimate for this example is 4,780 tonnes CO2e/yr ± 33.3%

4.3.2 Fugitive Emission Estimation

This next example demonstrates the uncertainty calculation for estimating methane emissions resulting
from fugitive sources associated with an electric reciprocating compressor located in a conventional crude
production operation. The example is taken from a Canadian study on GHG emission uncertainty (CAPP,
2004).

Pilot Version, September 2009 4-10

 EXHIBIT 4-2: Uncertainty Example for Fugitive Emissions Estimation
Input Data: Table 4-1 contains components counts and their uncertainties for an electric reciprocating
compressor. The uncertainties shown for the component counts are the 95% confidence limits on the
average number of components and were determined from an analysis of field survey data. It is assumed
that the sampling plan was designed to be representative of the component population, and therefore bias is
eliminated.
Table 4-1. Component Counts and Uncertainties
Equipment schedules for facilities in the upstream O&G industry
Lower Upper
Equipment Component Uncertainty Uncertainty
Type Type Service Count (%) (%)
Electric Connectors Gas/Vapor 275 18 18
reciprocating Connectors Light Liquid 2 77 77
compressor Controllers Fuel Gas 5 31 326
Compressor Gas/Vapor 2 75 75
Seals
Open-ended Gas/Vapor 4 58 58
Lines
Valves Gas/Vapor 20 16 16
Valves Light Liquid 1 77 77
CAPP, Volume 5, Table 4.1, 2004

Emissions Factor: Table 4-2 provides emission factors and their uncertainties for each of the components in
Table 4-1. Here also, bias is assumed to be eliminated due to a well designed sampling plan.
Table 4-2. Emission Factors and Uncertainties

Summary of average emission factors for uncontrolled fugitive THC emissions

(kg/hr/source) at UOG facilities.
Equipment Emissions Lower Upper
Factor Sweet/ Component Factors (kg Uncertainty Uncertainty
Group Sour Service Type THC/source-hr) (%) (%)
Gas All Gas/Vapor Connectors 7.06E-04 31 31
Gas All Light Liquid Connectors 5.51E-04 90 111
Gas All Fuel Gas Controllers 2.38E-01 27 27
Gas All Gas/Vapor Compressor 7.13E-01 36 36
Seals
Gas All Gas/Vapor Open-ended 4.27E-01 62 161
Lines
Gas All Gas/Vapor Valves 2.46E-03 15 15
Gas All Light Liquid Valves 3.52E-03 19 19
CAPP, Volume 3, Table 19, 2004

Weight Fractions of Emissions: Table 4-3 contains the CH4 weight fractions for gas production and light
crude. The weight fractions are from Table C-7 of the 2009 API Compendium. Expert judgment was used
to estimate the uncertainty. Since the uncertainty of the components and the emission factor give lower and
upper uncertainty bounds, it is important to have lower and upper uncertainty levels for the weight fraction
as well.

Pilot Version, September 2009 4-11

 EXHIBIT 4-2: Uncertainty Example for Fugitive Emissions Estimation, continued

Table 4-3. Methane Weight Fractions for Production Operations by Service

Lower Upper
Uncertainty Uncertainty
Operations Wt. Fraction (%) (%)
Conventional Oil: 0.523 5% 5%
Gas Service
Conventional Oil: 0.0363 5% 5%
Light Liquid Service
Source: Picard, D. J., B. D. Ross, and D. W. H. Koon. A Detailed Inventory of CH4
and VOC Emissions from Upstream Oil and Gas Operations in Alberta, Volume II
Development of the Inventory, Canadian Petroleum Association, March 1992,
Tables 12 through 15.
Methane Emissions

The methane emissions are estimated by the following equation.

Methane Emissions = Component Count × Emissions Factor × Weight Fraction

Uncertainty Estimate:
o To estimate the uncertainty of the emissions for the individual components, use Equation 4-6, the
equation for the uncertainty of a product. This applies the relative uncertainties for uncertainties
that are independent. For this example, the equation is written as:

U (rel )component = U count

2
+ U emission
2
factor + U weight fraction
2

For example, the total methane emissions for compressor seals are:
0.713 kg 8760hr tonnes
Emissions = 2 seals × × 0.523 wt. fraction × × = 6.53 tonnes/yr
hr-source year 1000kg

The lower uncertainty of this estimate is calculated as follows:

U (rel )component = U count

2
+ U EF
2
+ U wt2 fraction = 0.752 + 0.362 + 0.052 = 83.3%

o Note that in the emissions formula, there are two constants. Multiplication by a constant does not
change the relative uncertainty of an estimate. The total methane emissions are the sum of the
methane emissions for all of the components. The total uncertainty for methane emissions resulting
from fugitive sources associated with the electric reciprocating compressor are calculated using
Equation 4-4, which applies the absolute uncertainties for the uncertainty of a sum.

Pilot Version, September 2009 4-12

 EXHIBIT 4-2: Uncertainty Example for Fugitive Emissions Estimation, continued

o Table 4-4 gives the methane emissions for the individual components and their uncertainties along
with the total methane emissions and its uncertainty. The 95% confidence interval bounds for the
total methane emissions is shown as (7.41, 30.6) tonnes/year.
Table 4-4. Estimated Fugitive CH4 Emission
Equipment Type Electric reciprocating compressor
Relative Absolute
Lower Upper
Component CH4 Uncertainty Uncertainty Lower Upper
Type Service Emissions (%) (%) Uncertainty Uncertainty
Connectors Gas/Vapor 0.889 36.2 36.2 0.322 0.322
Connectors Light Liquid 0.000350 100a 135 0.000350 0.000474
Controllers Fuel Gas 0.545 41.4 327 0.226 1.78
Compressor Gas/Vapor 6.53 83.3 83.3 5.44 5.44
Seals
Open-ended Gas/Vapor 7.83 85.0 171 6.66 13.4
Lines
Valves Gas/Vapor 0.225 22.5 22.5 0.0507 0.0507
Valves Light Liquid 0.00112 79.5 79.5 0.000889 0.000889
TOTAL 16.0 53.7 91.0 8.61 14.6
a
This value was truncated (set to -100%) since the emissions cannot be less than 0.

4.3.3 Emission Factor Uncertainty

The decision tree presented in Figure 4-3 addresses uncertainty for emission factor data. Uncertainties
may be published along with the literature values for emission factors, in which case, the uncertainty from
the literature should be used. If the uncertainties associated with literature values are unavailable or data
are not available to calculate uncertainty, one must rely on expert judgment to estimate the uncertainty.

Expert judgment involves a person, or group of people, familiar with the systems assigning an
uncertainty to the estimate based on knowledge of the process. It is used when there is no
information to quantify the uncertainty based on data or manufacturer’s specifications of the
measured parameter. Expert judgment is covered in Sections 2.2, [Link], and Annex 2A.1 of the
2006 IPCC Guidelines for National Greenhouse Gas Inventories (IPCC, 2006). Section 7 of the
ISO-5168 explains how to deal with uncertainty estimates based on expert judgment (ISO,
2005).

Pilot Version, September 2009 4-13

Further guidance for assigning uncertainty values is provided in Section 5 in the context of
quantifying uncertainty for the GHG inventory of a hypothetical oil and gas facility. The
following example demonstrates uncertainty estimation for a default emission factor.

EXHIBIT 4-3: Uncertainty Example for Emission Factors

Input Data:
o The CO2 emissions factor for natural gas in production (non-pipeline quality) is 0.0547 tonnes/106
Btu (HHV) (from Table 4-2 of the 2009 API Compendium). An estimate of uncertainty for this
value is not provided in the original reference, so an uncertainty of 10% at the 95% confidence
interval is assumed based on expert judgment. Bias is assumed to be accounted for in this value.
o Similarly, the natural gas heating values is 1020 Btu/scf (from Table 3-8 of the 2009 API
Compendium). An estimate of uncertainty for this value is not provided in the original reference so
an uncertainty of 10% at the 95% confidence interval is assumed based on expert judgment. Bias is
assumed to be accounted for in this value.
Emission Factor Estimate:
o The emission factor is quantified as the product of the carbon content and heating value, as shown
by the following formula:
Emission Factor Gas Carbon Content Heating Value 106 scf/MMscf
× × × 6
(tonnes CO2 /MMscf) (tonnes CO2 /MMBtu) (Btu/scf) 10 Btu/MMBtu
Uncertainty Estimate:
o For the purpose of this example, the uncertainties associated with the carbon content and heating
value are assumed to be independent because the values are cited from different literature
references and are not based on measured data. Therefore, Equation 4-6 is applied to estimate the
uncertainty for the emission factor. This applies the relative uncertainties for the carbon content
and heating value.
2 2
⎛U ⎞ ⎛U ⎞
U (Rel) X ×Y = ⎜ X ⎟ +⎜ Y ⎟
⎝ X ⎠ ⎝ Y ⎠

U (Rel) Emission Factor = 0.102 + 0.102 = 0.141

The resulting emission factor is then:

0.0547 tonnes CO2 1, 020 Btu 106 scf/MMscf
Emission Factor = × × 6
MMBtu scf 10 Btu/MMBtu
Emission Factor = 55.8 ( ±14.1% ) tonnes CO 2 /MMscf

Pilot Version, September 2009 4-14

 Emission Factor Uncertainty

Are you applying a default Site specific emission data

emission factor or are you using Refer to Measurement Decision
site-specific data? Tree

Default emission factor Do the default emission Yes Apply reported uncertainty and
factors have uncertainty progress to Uncertainty
specified or quantified? Aggregation

No
Assign uncertainty based on expert
judgment and document reasons
supporting the assignment.
Progress to Uncertainty Aggregation.

Figure 4-3. Decision Diagram for Emission Factor Uncertainty

Pilot Version, September 2009 4-15

4.4 Quantifying Measurement Uncertainty
Ideally, one would develop emission estimates and associated uncertainty from measured facility data.
This data could be obtained either from periodic sampling or by continuous monitoring. The decision tree
provided in Figure 4-4 directs the user to the appropriate equations for quantifying measurement
uncertainty. Following the decision tree provided, we will first examine the equations for aggregating
uncertainty for a single measurement point.

Determine what parameters

contribute to uncertainty. Note that Apply “Square root of the sum of the
Yes
Are the data based on a single bias may be one of the parameters. squares” to aggregate uncertainty
Single point
point measurement or multiple Are the uncertainties in the parameters that are independent
measurements? measurements independent? (See text (Equation 4-4)
for guidance)
No Apply Equation 4-5 to aggregate the
uncertainty for parameters that are
correlated.

Although there is only a single measurement value in this application, such as the cumulative flow rate
through a totalizer meter, there may be more than one parameter that contributes to the uncertainty of the
measurement. For example, Section 3.2 indicates that the following items need to be considered when
estimating measurement uncertainty:

• Uncertainty of the measurement instrument;

• Additional uncertainty of “context specific” factors (discussed previously in Section 3.2); and
• Uncertainty of measurement corrections, e.g., pressure and temperature corrections for gas meters
measurements.

The aggregated uncertainty from these parameters is calculated by either applying Equation 4-4 for
independent parameters or Equation 4-5 for correlated parameters.

Example –Single Flow Measurement

The following example demonstrates the uncertainty calculation for estimating CO2 emissions resulting
from the combustion of produced natural gas. The scenario presented is based on a single measurement of
the flow.

Pilot Version, September 2009 4-16

 Measurement Uncertainty
Determine what parameters contribute
to uncertainty. Note that bias may be Apply “Square root of the sum of the
Yes squares” to aggregate uncertainty
Are the data based on a single Single point one of the parameters. Are the errors in
point measurement or multiple the measurements independent? (See parameters that are independent (Equation
measurements? text for guidance) 4-4).

No Apply Equation 4-5 to aggregate the

uncertainty for parameters that are
correlated.

Do the measurements Apply Equations 4-1 through 4-3 to calculate the mean and standard
Multiple points represent a sampling of the Yes deviation, respectively.
measured parameter? Apply Equation 4-8 to estimate the uncertainty based on the measured
data, OR apply Equation 4-9 to estimate uncertainty for a single
observed estimate where other data are used to quantify the
No uncertainty.

Determine what (additional) parameters Apply Equation 4-4 to aggregate uncertainty for
contribute to uncertainty. Note that bias Yes addition/ subtraction of data.
may be one of the parameters. Apply Equation 4-6 to aggregate uncertainty for
Are the errors in the measurements multiplication/ division of data.
independent? (See text for guidance)

No Apply Equation 4-5 to aggregate uncertainty for

addition/ subtraction of data.
Apply Equation 4-7 to aggregate uncertainty for
multiplication/ division of data.

Aggregate the uncertainty parameters

by applying the “Square root of the
sum of the squares.” (Equation 4-4)

Figure 4-4. Decision Diagram for Measurement Uncertainty

Pilot Version, September 2009 4-17

 EXHIBIT 4-4: Uncertainty Example for Single Flow Measurement

Input Data
Flow for a gaseous fuel is measured using a totalizer meter prior to being routed to a combustion device.
The total annual flow rate for the fuel is 18,361 MMscf/yr. Because the meter records a cumulative flow,
the uncertainty associated with this measurement is not reduced by taking daily or monthly readings of the
flow. The uncertainty of this measurement is determined by following the decision tree from Figure 4-4
through the measurement uncertainty. The uncertainty of the measurement value is then aggregated.
Uncertainty Assessment:
Measurement uncertainty
o From Section 3.2 Table 3-4, the random error that is expected for a properly installed and operated
orifice meter is 1.5% (95% confidence interval) when the meter is operating at 30-100% of the
measurement range. For this example, this equates to an absolute uncertainty value of:
MMscf
18,361 × 0.015 = 275
yr
o “Context-specific” factors and the uncertainty of pressure and temperature corrections for gas
meters can be based on expert data as summarized in industry standards referenced in Section 3.0.
In the absence of quantitative information about this uncertainty, this example provides a sensitivity
analysis of the impact of assuming a range of context-specific uncertainty values: 10%, 25%, and
100% (on a relative uncertainty basis). These values were selected to test the sensitivity of the
calculations and assess the effect over a wide range of variability.
o For this example, it is assumed that it is not possible to quantify the bias. This meter was installed
according to the manufacture’s requirements 17 years ago and has an expected life span of 25 years.
The meter was last calibrated 7 years ago. Due to the installation and calibration of the equipment,
expert judgment was used to estimate a 5% bias in the measurement. This equates to an absolute
uncertainty value of:
MMscf
18,361 × 0.05 = 918
yr
o As shown in the measurement uncertainty decision tree (Figure 4-4), the selection of an equation
depends on whether the uncertainties are related (correlated). For this example, the uncertainty
components (bias, meter error, and context-specific factors) are independent. To calculate the total
uncertainty of a single measurement for this example, Equation 4-4 is applied to account for the
uncertainties associated with the measurement, the other context-specific factors, and bias:

U (abs ) Aggregated = U 2 measurement + U 2 context specific + U 2bias

For example, if the context specific uncertainty is 10% on a relative basis (1,840 on an absolute basis), the
aggregated uncertainty is:

U (abs ) Aggregated = 2752 + 1836.12 + 9182 = 2, 070

(or 11.3% on a relative uncertainty basis)

Pilot Version, September 2009 4-18

 EXHIBIT 4-4: Uncertainty Example for Single Flow Measurement, continued

The following table shows the calculated aggregate uncertainty for the different values of the context-specific
factors due to pressure and temperature corrections.
Table 4-5. Uncertainty in a Single Measurement

Single
Meter Measurement Assigned Context Specific Bias, Single Measurement Measurement
Uncertainty, 106 scf Context Specific Uncertainty, 106 scf Total Uncertainty, Total
(1.5%) Uncertainty, % 106 scf (absolute) (5 %) 106 scf (absolute) Uncertainty, %
275 10 1,840 918 2,070 11.3%
275 25 4,590 918 4,690 25.5%
275 100 18,400 918 18,400 100%

Example – Multiple Flow Measurements

The next example demonstrates the uncertainty calculation for estimating CO2 emissions resulting from the
combustion of produced natural gas. Two methods are compared. The first calculates an annual average
CO2 emission factor based on monthly natural gas composition measurements and an annual summation of
daily flow rates. The second determines a monthly CO2 emission factor based on a monthly natural gas
composition sample and applies a monthly summation of daily flow rates.

EXHIBIT 4-5: Uncertainty Example for Multiple Flow Measurements

Table 4-5 shows flow measurement data for produced natural gas that is routed to a combustion device. The fuel
flow is measured continuously (scf/sec) with an orifice meter. A data acquisition system records the daily flow rate
measurements. The daily and monthly totals are shown in Table 4-6.
Uncertainty Assessment:
Measurement uncertainty
o The total volume of gas combusted for this example results in an annual flow rate of 1,187×106 scf.
Because the flow is measured continuously (with the total recorded daily), this is not considered a sampling
of data so Equations 4-1 and 4-3 do not apply.
o From Section 3.2 Table 3-4, the random error that is expected for a properly installed and operated orifice
meter is 1.5% (95% confidence interval) when the meter is operating at 30-100% of the measurement range.
o The context-specific factors discussed previously also apply for this example. Again, a sensitivity analysis
is applied for the context-specific uncertainty values of 10%, 25%, and 100%.
o As with the Single Flow Measurement example, expert judgment was used to assign a 5% bias in the
measurement.

Pilot Version, September 2009 4-19

 EXHIBIT 4-5: Uncertainty Example for Multiple Flow Measurements, continued

Table 4-6. Flow Measurements (in Mscf/day)

Day Jan. Feb. Mar. April May June July Aug. Sept. Oct. Nov. Dec.
1 3,374 2,344 3,102 3,483 3,321 3,530 2,974 3,484 3,559 3,269 3,239 3,344
2 3,403 2,373 3,329 3,471 3,331 3,518 3,033 3,400 3,494 3,365 3,210 3,231
3 3,381 2,235 3,406 3,530 3,342 3,480 3,254 3,372 3,503 3,236 3,207 3,275
4 3,319 2,276 3,419 3,606 3,335 3,460 3,417 3,406 3,679 3,162 3,132 3,402
5 3,299 2,319 3,555 3,597 3,342 3,503 3,329 3,385 3,636 3,240 3,107 3,570
6 3,173 2,215 3,737 3,501 3,342 3,487 3,034 3,393 3,604 3,233 3,047 3,364
7 3,182 2,459 3,711 3,486 3,378 3,463 2,915 3,449 3,601 3,239 3,060 3,209
8 3,257 2,288 3,771 3,441 3,253 3,323 3,007 3,370 3,541 3,370 3,101 3,344
9 2,985 2,572 3,806 3,400 3,320 3,253 3,158 3,506 3,544 3,347 2,969 3,286
10 3,022 2,433 3,507 3,348 3,301 3,551 3,330 3,412 3,507 3,433 2,901 3,158
11 2,830 2,769 3,340 3,305 3,332 3,478 3,198 3,464 3,557 3,458 3,154 3,186
12 2,914 2,640 3,328 3,370 3,395 3,510 3,262 3,449 3,459 3,405 3,425 3,037
13 2,864 2,431 3,377 3,452 3,348 3,471 3,263 3,341 3,518 3,424 3,488 2,834
14 2,641 2,767 3,432 3,467 3,246 3,302 3,339 3,435 3,484 3,389 3,323 2,849
15 2,647 2,687 3,390 3,465 3,229 3,534 3,158 3,489 3,418 3,397 3,234 2,819
16 2,519 2,888 3,424 3,579 3,322 3,590 3,188 3,468 3,404 3,404 3,434 2,901
17 2,670 2,912 3,432 3,605 3,344 3,475 3,239 3,369 3,425 3,404 3,381 2,807
18 2,482 3,006 3,424 3,543 3,387 3,445 3,114 3,409 3,417 3,407 3,376 2,892
19 2,569 3,089 3,522 3,477 3,356 3,520 3,225 3,362 3,468 3,216 3,382 2,942
20 2,320 3,299 3,444 3,335 3,136 3,561 3,220 3,356 3,474 3,233 3,389 2,915
21 2,405 3,401 3,551 3,545 3,291 3,449 3,122 3,488 3,483 3,295 3,420 2,968
22 2,368 2,945 3,436 3,486 3,463 3,389 2,830 3,374 3,466 3,200 3,464 2,982
23 2,287 3,147 3,495 3,460 3,430 3,474 2,962 3,296 3,447 3,257 3,463 2,942
24 2,277 3,398 3,446 3,461 3,405 3,351 3,546 3,330 3,355 3,282 3,410 3,028
25 2,214 3,521 3,563 3,530 3,430 3,352 3,333 3,422 3,371 3,241 3,346 3,089
26 2,168 3,414 3,511 3,424 3,472 3,316 3,373 3,308 3,383 3,188 3,434 2,963
27 2,216 3,241 3,392 3,482 3,555 3,269 3,516 3,302 3,384 3,177 3,480 2,891
28 2,306 3,206 3,246 3,295 3,550 3,262 3,564 3,483 3,369 3,222 3,307 2,954
29 2,091 3,394 3,336 3,543 3,297 3,547 3,553 3,376 3,300 3,244 3,113
30 2,108 3,489 3,345 3,555 3,262 3,564 3,593 3,304 3,240 3,305 3,047
31 2,209 3,470 3,537 3,611 3,557 3,241 2,972
Total,
103 scf/ 83,500 78,275 107,449 103,825 104,590 102,874 100,624 106,027 104,232 102,277 98,432 95,313
month

o Since the uncertainties associated with the meter, the context-specific factors, and the bias are independent (there
is no autocorrelation), we can apply Equation 4-4 to estimate the uncertainty. The absolute uncertainties are used
in this equation. The results are shown in Table 4-7 below.

∑ (U )
December
U (abs )Year = 2
Measurementi + U context
2
specifici + U Biasi
2

i = January

Pilot Version, September 2009 4-20

 EXHIBIT 4-5: Uncertainty Example for Multiple Flow Measurements, continued

Table 4-7. Uncertainty in Summation of Flow Measurements - Annual

Annual Measurement Annual Measurement
Assigned Context Total Uncertainty Flow Rate
Specific Uncertainty, % (absolute), 103 scf Uncertainty, %
10 133,947 11.28%
25 303,257 25.54%
100 1,189,036 100.14%

As with the annual volume, the uncertainties associated with the meter (1.5%), the context-specific factors
(assigned 10%, 25%, and 100% to examine the sensitivity of this uncertainty), and bias (5%) apply to the
daily readings and monthly totals. Summing the daily flow rates for the monthly totals results in the same
relative uncertainties as shown in Table 4-6. Table 4-8 summarizes the absolute uncertainties for the monthly
flow rates.
Table 4-8. Uncertainty in Summation of Flow Measurements – Monthly
Monthly Measurement Flow Rate Uncertainty
Relative Uncertainty, % 11.28% 25.54% 100.14%
10% 25% 100%
Month MMscf (absolute uncertainties, 103 scf)
January 83.50 9,419 21,325 83,614
February 78.28 8,830 19,991 78,382
March 107.45 12,121 27,442 107,595
April 103.83 11,712 26,516 103,966
May 104.59 11,798 26,712 104,733
June 102.87 11,605 26,273 103,014
July 100.62 11,351 25,699 100,761
August 106.03 11,960 27,078 106,171
September 104.23 11,758 26,620 104,374
October 102.28 11,537 26,121 102,417
November 98.43 11,104 25,139 98,566
December 95.31 10,752 24,342 95,443

4.4.1 Measurement Uncertainty – Multiple Measurements

Continuing with the decision tree provided in Figure 4-4, where multiple measurement points are available,
the corresponding uncertainty can be derived on the basis of a statistical sample. See Cochran, 1977 for a
comprehensive guide to sampling.

Pilot Version, September 2009 4-21

 Are the data based on a single
point measurement or multiple
measurements?

Do the measurements Apply Equations 4-1 through 4-3 to calculate the mean and
Multiple points Yes standard deviation, respectively.
represent a sampling of the
measured parameter? Apply Equation 4-8 to estimate the uncertainty based on the
measured data, OR apply Equation 4-9 to estimate uncertainty for a
single observed estimate where other data is used for to quantify
the uncertainty.

The sample standard deviation “includes contributions to the precision both from the measurement system
and from the material composition variation from sample to sample” (Coleman and Steele, 1989). In other
words, if the measured data for an emission source are derived from a statistical sample, the use of the
standard deviation as a measure of the spread of the data accounts for uncertainty of the measurement
instrument and the differences among the samples. For example, if three samples are taken every month
and the data for the 36 samples are used to calculate a yearly mean, the uncertainty calculated from these
samples accounts for both the uncertainty in the measurement instrument and the variability among
observations. Thus, this uncertainty will be larger than the uncertainty due to measurement error alone. As
the sample size increases, the uncertainty that combines the instrument error and context-specific factors
will decrease.

Next, we would calculate the standard deviation of the sample using Equation 4-1. The IPCC Good
Practices document and others recommend applying the Equation 4-8 to quantify the uncertainty of the
data set (IPCC, 2001), which is also referred to as the relative expanded uncertainty and can be thought of
as half the bandwidth of a 95% confidence interval (ISO, 2005).

t × s ( x) / n
U (rel )( x) = × 100% (Equation 4-8)
x
where
U(Rel)(x) = the relative expanded uncertainty in the data set (%);
t = a value based on Student’s t-distribution with n-1 degrees of freedom which gives a 95%
confidence interval;
x = is the mean for the set of data calculated in Equation 4-3;
s( x ) = the standard deviation of the data set calculated in Equation 4-2; and
n = the sample size for the set of data.

If one uses a single observation as an estimate and uses other data to calculate the uncertainty,
Section A1.2.3 of the IPCC Good Practices document explains that Equation 4-9 should be used instead of
Equation 4-8 (IPCC, 2006). IPCC’s example application of this approach is for the use of single emission
estimate for a particular year that has been calculated on more than one occasion. The recalculations have

Pilot Version, September 2009 4-22

occurred as a result of changes in the methodology, corrections, or as a result of new data. In this case, it is
the standard deviation of the sample set that is appropriate and not the standard deviation of the mean.

t × s( x)
U ( rel )( x ) = × 100% (Equation 4-9)
x
where
U(Rel)(x) = the relative expanded uncertainty in the data set (%);
t = a value based on Student’s t-distribution with n-1 degrees of freedom which gives a 95%
confidence interval;
x = is the mean for the set of data calculated in Equation 4-3; and
s( x ) = the standard deviation of the data set calculated in Equation 4-2.

Continuing with the decision tree from Figure 4-4, uncertainty from other parameters may also require
consideration, despite the fact that the standard deviation of a statistical sample of observations accounts
for the measurement uncertainty and variability.

Are the data based on a single

point measurement or multiple
measurements?

Do the measurements Apply Equations 4-1 through 4-3 to calculate the mean and
Multiple points Yes standard deviation, respectively.
represent a sampling of
the measured parameter? Apply Equation 4-8 to estimate the uncertainty based on the
measured data, OR apply Equation 4-9 to estimate uncertainty
for a single observed estimate where other data is used for to
No quantify the uncertainty.

Determine what (additional) parameters Apply Equation 4-4 to aggregate uncertainty

contribute to uncertainty. Note that bias may be
Yes for addition/ subtraction of data.
one of the parameters. Apply Equation 4-6 to aggregate uncertainty
Are the errors in the measurements independent? for multiplication/ division of data.
(See text for guidance)
Apply Equation 4-5 to aggregate uncertainty
No for addition/ subtraction of data.
Apply Equation 4-7 to aggregate uncertainty
for multiplication/ division of data.

Aggregate the uncertainty

parameters by applying the “Square
root of the sum of the squares.”
(Equation 4-4)

Where multiple uncertainty parameters apply, selection of the equation to aggregate uncertainty parameters
depends on whether the uncertainties are independent. In addition, the activity value may be based on the
product or sum of multiple data or variables. The decision diagram references the appropriate equation
based on these criteria.

Example – Statistical Sampling Uncertainty

For this example, natural gas samples are collected to calculate the CO2 emission factor associated with
combusting the gas. (Note: this example would also apply to composition data for a flared gas stream.)

Pilot Version, September 2009 4-23

Two methods are compared. The first calculates an annual average CO2 emission factor based on monthly
natural gas composition measurements and an annual summation of daily flow rates. The second
determines monthly CO2 emission factors based on monthly natural gas composition samples and applies a
monthly summation of daily flow rates.

Although this example is examining uncertainty associated with an emission factor, the decision tree
provided in Figure 4-3 references the figure above (derived from Figure 4-4) for quantifying emission
factor uncertainty where the emission factor is based on multiple measurements from a statistical sample.

EXHIBIT 4-6: Uncertainty Example for Statistical Sampling

Annual Composition Data
o The facility collected 12 natural gas composition samples throughout the year in order to quantify the
CO2 content of the gas (expressed as tonnes CO2/MMscf), as shown in Table 4-9.
o Since the data are based on a statistical sample, the uncertainty of the average composition is
estimated using the sample standard deviation for each compound in the natural gas. The standard
deviation will account for the uncertainty due to measurement error and the natural variability of the
sampled values. Sampling procedures, maintenance activities, and equipment calibration are assumed
to eliminate bias. Equation 4-7 was used to calculate the uncertainty. The calculation is shown for
CH4.

tα =0.05,df =12−1 × s ( x) / n 2.201×1.529369 / 12

U (rel )( x) = 100% × = 100% × = 1.04%
x 93.11333

The molecular weight of the average composition is calculated by applying the following equation:

1 # compounds
MWMixture = × ∑ (Mole%i × MWi )
100 i =1

resulting in 17.25045 lb/lbmole, as shown in Table 4-10.

Pilot Version, September 2009 4-24

 Table 4-9. Measured Composition Data

Mole% dry January February March April May June July August September October November December
Methane 90.73 91.3 91.37 91.21 94.86 93.97 93.91 94.3 94.45 94.33 94.08 92.85
Ethane 3.83 3.74 4.1 4.16 2.27 2.7 2.81 2.59 2.27 2.3 2.3 2.88
CO2 3.69 3.24 2.93 2.96 1.32 1.23 0.79 1.13 1.38 1.38 1.4 1.49
Propane 0.9 0.8 0.82 0.88 0.61 0.79 0.66 0.65 0.65 0.65 0.66 0.76
i-Butane 0.11 0.1 0.1 0.11 0.09 0.12 0.09 0.1 0.1 0.1 0.1 0.11
n-Butane 0.14 0.13 0.13 0.13 0.12 0.16 0.15 0.14 0.13 0.13 0.13 0.14
i-Pentane 0.03 0.04 0.03 0.04 0.04 0.05 0.04 0.05 0.04 0.04 0.04 0.05
n-Pentane 0.02 0.03 0.02 0.03 0.03 0.04 0.03 0.04 0.03 0.03 0.03 0.04
C6+ 0.03 0.05 0.04 0.05 0.05 0.07 0.06 0.24 0.05 0.05 0.04 0.08
Total 99.48 99.43 99.54 99.57 99.39 99.13 98.54 99.24 99.1 99.01 98.78 98.4

Table 4-10. Average Composition Calculations

Mean, MW wt% C
mole% U(abs), Calculation, U(abs), calculation, U(abs),
dry Std dev t*s/sqrt(n) U(rel) mole % lb/lbmole lb/lbmole wt% wt%
Methane 93.11333 1.529369 0.971714736 1.04% 0.971715 14.93538 0.1559 64.7728 1.4797
Ethane 2.995833 0.747498 0.474937795 15.85% 0.474938 0.900847 0.1428 4.168007 0.6662
CO2 1.911667 0.988652 0.62815969 32.86% 0.62816 0.841325 0.2765 1.32982 0.4378
Propane 0.735833 0.100856 0.064080962 8.71% 0.064081 0.324503 0.0283 1.535612 0.1373
i-Butane 0.1025 0.00866 0.005502463 5.37% 0.005502 0.059573 0.0032 0.28521 0.0164
n-Butane 0.135833 0.010836 0.006885023 5.07% 0.006885 0.078946 0.0040 0.377961 0.0206
i-Pentane 0.040833 0.006686 0.004247814 10.40% 0.004248 0.029461 0.0031 0.142025 0.0151
n-Pentane 0.030833 0.006686 0.004247814 13.78% 0.004248 0.022246 0.0031 0.107244 0.0149
C6+ 0.0675 0.055942 0.035544067 52.66% 0.035544 0.058172 0.0306 0.281732 0.1485
Total 99.13417 17.25045 73.00041
Sum uncertainty (abs) 0.350567 1.693261
Sum uncertainty (rel) 2.03% 2.32%
For sample size of n=12, the Student’s t distribution value is 2.201.
Note, the resulting estimates are not rounded to show the comparison.

Pilot Version, September 2009 4-25

 .
EXHIBIT 4-6: Uncertainty Example for Statistical Sampling, continued

o The uncertainty associated with this value is calculated by applying Equation 4-4, using the absolute
uncertainties for each molar compound.

U (abs )
∑ ( mole %×MW )
= ∑U (abs) 2
mole %× MW

(0.0104 × 14.93538) 2 + (0.1585 × 0.900847) 2 + (0.3286 × 0.841325) 2

= + (0.0871× 0.324503) 2 + (0.0537 × 0.059573) 2 + (0.0507 × 0.078946) 2
+ (0.1040 × 0.029461) 2 + (0.1378 × 0.022246) 2 + (0.5266 × 0.058172) 2
(0.1559) 2 + (0.1428) 2 + (0.2765) 2 + (0.0283) 2 + (0.0032) 2 + (0.0040) 2
= = 0.350567
+ (0.0031) 2 + (0.0031) 2 + (0.0306) 2
0.350567
U (rel ) = ×100% = 2.03%
∑ ( mole %×MW ) 17.25045

The weight percent carbon of the average composition is calculated by applying the following equation:
lbmolei x lbmole C 12 lb C 1
Wt% CTotal = ∑ × × ×
100 lbmole total lbmolei lbmole C MWTotal

This equates to 73.00041% C, as shown in Table 4-9. The uncertainty associated with this value is calculated
by applying Equation 4-1, using the absolute uncertainties for the weight percent carbon of each molar
compound.
o The uncertainty associated with this value is calculated by applying Equation 4-1, using the absolute
uncertainties for each molar compound:

U (abs )
∑ ( wt %)
= ∑U (abs) 2
wt %

(0.0228 × 64.7728) 2 + (0.1598 × 4.168007) 2 + (0.3292 ×1.32982) 2

= + (0.0894 × 1.535612) 2 + (0.0574 × 0.28521) 2 + (0.0546 × 0.377961) 2 = 1.693261
+ (0.1060 × 0.142025) 2 + (0.1393 × 0.107244) 2 + (0.5270 × 0.281732) 2
1.693261
U (rel ) = × 100% = 2.32%
∑ ( wt %)
73.00041
o The average annual CO2 emissions factor is then calculated by applying the following equation:
lb CTotal 44lb CO2 /lbmole CO2 lbTotal lbmoleTotal 106 scf tonne CO2
× × × × ×
lbTotal 12lb C/lbmole C lbmoleTotal 379.3 scf MMscf 2204.62 lb CO2
=55.2180 tonnes CO2/MMscf.
o The uncertainty for this quantity is calculated by applying Equation 4-6 for the relative uncertainties
of MWTotal and wt% CTotal
2 2
⎛U ⎞ ⎛U ⎞
U (rel ) = ⎜ X ⎟ + ⎜ Y ⎟ = 2.032 + 2.322 = 3.08%
⎝ X ⎠ ⎝ Y ⎠

Pilot Version, September 2009 4-26

 EXHIBIT 4-6: Uncertainty Example for Statistical Sampling, continued

Monthly Composition Data

o The facility collected one natural gas composition sample each month. For comparison with the annual average
gas composition example above, the same composition values shown in Table 4-8 will be used.
o For this example, only one data point is available each month. Table C-2 (Appendix C) provides reproducibility
uncertainty associated with natural gas samples. These values can be applied to account for the measurement
error in each monthly sample. Note that the reproducibility uncertainty varies based on the mole % of each gas
compound. The values are shown for January in Table 4-10.
o An additional 5% uncertainty is assigned by expert judgment to account for potential variability and bias in the
gas composition during the month. The combined uncertainty is calculated by applying Equation 4-4, using the
absolute uncertainties. The calculation is demonstrated for CH4. Results for January are shown in Table 4-11.

U (abs )Composition = U (abs ) 2Re producibility + U (abs )Variability

2

U (abs )CH 4 = 0.152 + 4.53652 = 4.539

4.539
U (rel ) = × 100% = 5.00%
90.73

Table 4-11. Measured Monthly Composition Data

Combined wt%
Reproducibility Variability mole% MW Carbon
Uncertainty Uncertainty Uncertainty Calculation, Calculation, wt% C
mole% dry January (abs), mole% (Uabs=mole%x5%) (rel) lb/lbmole % U(rel)
Methane 90.73 0.15 4.5365 5.00% 14.5531 60.71 6.49%
Ethane 3.83 0.1 0.1915 5.64% 1.1517 5.13 6.99%
CO2 3.69 0.1 0.1845 5.69% 1.6240 2.47 7.03%
Propane 0.9 0.07 0.045 9.25% 0.3969 1.81 10.13%
i-Butane 0.11 0.07 0.0055 63.83% 0.0639 0.29 63.97%
n-Butane 0.14 0.07 0.007 50.25% 0.0814 0.37 50.42%
i-Pentane 0.03 0.02 0.0015 66.85% 0.0216 0.10 66.98%
n-Pentane 0.02 0.02 0.001 100.12% 0.0144 0.07 100.21%
C6+ 0.03 0.02 0.0015 66.85% 0.0259 0.12 66.98%
Total 99.48 17.9329 71.0717
Sum uncertainty (abs) 0.7404 3.9734
Sum uncertainty (rel) 4.13% 5.59%

o As with the annual composition example, the molecular weight of the monthly composition is calculated by
applying the following equation:

1 # compounds
MWMixture = × ∑ ( mole%i × MWi )
100 i =1

resulting in 17.9329 lb/lbmole, as shown in Table 4-11.

Pilot Version, September 2009 4-27

 EXHIBIT 4-6: Uncertainty Example for Statistical Sampling, continued

o For each individual gas compound, the relative uncertainty of the mole%i × MWi is equivalent to the
combined reproducibility and variability uncertainties since the molecular weight of each gas
compound is a constant. The aggregated uncertainty associated with the mixture’s MW is calculated
by applying Equation 4-1, using the absolute uncertainties for each molar compound.

U (abs )
∑ ( MW Total )
= ∑U (abs) 2
mole %× MW

(0.0500 × 14.5531) 2 + (0.0564 × 1.1517) 2 + (0.0569 × 1.6240) 2

= + (0.0925 × 0.3969) 2 + (0.6383 × 0.0639) 2 + (0.5025 × 0.0814) 2
+ (0.6685 × 0.0216) 2 + (1.0012 × 0.0144) 2 + (0.6685 × 0.0259) 2
(0.7281) 2 + (0.0650) 2 + (0.0924) 2 + (0.0367) 2 + (0.0408) 2 + (0.0409) 2
= = 0.7404
+ (0.0145) 2 + (0.0144) 2 + (0.0173) 2
0.7404
U (rel ) = × 100% = 4.13%
∑ ( MW Total ) 17.9329

o The weight percent carbon of the average composition is calculated by applying the following
equation:
lbmolei x lbmole C 12 lb C 1
Wt% CTotal = ∑ × × ×
100 lbmole total lbmolei lbmole C MWTotal

This equates to 71.07% C, as shown in Table 4-11.

o For each compound in the gas mixture, the uncertainty from this calculation is determined by
applying Equation 4-6, using the relative uncertainties of the lbmolei and MWTotal. This is
demonstrated for the January CH4 composition.

U (rel )Wt %C i = U (rel )2mole %i × U (rel ) 2MW Total = 52 + 4.132 =6.49%

o The uncertainty associated with the wt% C of the mixture is calculated by applying Equation 4-4,
using the absolute uncertainties for the weight percent carbon of each molar compound.

U (abs )
∑ ( wt %)
= ∑U (abs) 2
Wt %

(0.0649 × 60.71) 2 + (0.0699 × 5.13)2 + (0.0703 × 2.47) 2

= +(0.1013 ×1.81) 2 + (0.6397 × 0.29) 2 + (0.5042 × 0.37) 2 = 3.9734
+(0.6698 × 0.10) 2 + (1.0021× 0.07) 2 + (0.6698 × 0.12) 2
3.9724
U (rel ) = ×100% = 5.59%
∑ (Wt %) 71.0717

Pilot Version, September 2009 4-28

 EXHIBIT 4-6: Uncertainty Example for Statistical Sampling, continued

o The monthly CO2 emissions factor is then calculated by applying the following equation:

lb CTotal 44lb CO2 /lbmole CO 2 lb Total lbmoleTotal 106 scf tonne CO 2

× × × × ×
lbTotal 12lb C/lbmole C lbmoleTotal 379.3 scf MMscf 2204.62 lb CO 2
=55.8856 tonnes CO2/MMscf for January.

o The uncertainty for this quantity is calculated by applying Equation 4-3 for the relative uncertainties
of MWTotal and wt% CTotal
2 2
⎛U ⎞ ⎛U ⎞
U (rel ) = ⎜ X ⎟ + ⎜ Y ⎟ = 4.132 + 5.592 =6.95%
⎝ X ⎠ ⎝ Y ⎠

o The same methods are applied to each of the monthly samples. Table 4-12 summarizes the resulting
emission factors (tonnes CO2/MMscf) and uncertainties.
Table 4-12. Measured Monthly Emission Factors
Tonnes U(abs), Tonnes
CO2/MMscf CO2/MMscf U(rel), %
January 55.8856 3.8840 6.95
February 55.7698 3.9276 7.04
March 55.9697 3.9414 7.04
April 56.1644 3.9304 7.00
May 54.7437 4.1879 7.65
June 55.2278 4.1441 7.50
July 54.7069 4.1606 7.61
August 55.4646 4.1751 7.53
September 54.6648 4.1705 7.63
October 54.6332 4.1647 7.62
November 54.4964 4.1527 7.62
December 54.8858 4.0798 7.43

4.5 Aggregating Uncertainty

Figure 4-5 provides the decision tree for aggregating uncertainties. This diagram steps through the
combination of activity data and emission factors to result in the final emission estimate. It also addresses
the aggregation of emissions from multiple sources, applying the GWP to convert non-CO2 emissions to a
CO2 equivalent basis, and the final summation of CO2 equivalent emissions.

An important point to note is that, although the GWP values reported by IPCC have corresponding
uncertainties, the GWP values are treated as constants in compiling a facility- or entity-wide GHG
inventory. Multiplying by a constant does not change the relative uncertainty.

Pilot Version, September 2009 4-29

Table 4-13 shows the uncertainty in the annual estimate of emissions for the different estimates of the
emission factor uncertainty and the activity rate uncertainty presented in the previous examples that address
emissions from the combustion of natural gas. The table shows the results of using literature values and
sampling data to determine the emission factors.

Total CO2 emissions are calculated based on the product of the activity data (fuel consumption) and the
emission factor (tonnes CO2/volume fuel). Table 4-13 compares the emission estimate results from three
approaches:

− Annual emissions based on the measured gas consumption and a default emission factor;

− Annual emissions based on the measured gas consumption and an emission factor derived from an
annual average gas composition; and

− Annual emissions based on an aggregate of monthly gas composition and flow rate measurements.

Pilot Version, September 2009 4-30

 Uncertainty Aggregation

Are the errors for the Independent errors: Apply Equation 4-4
following aggregations If two emission sources have the same to aggregate uncertainty for addition/
correlated? emission factor (for example, boilers subtraction of data.
Refer to the text for and heaters) combine the activity data Correlated errors: Apply Equation 4-5
guidance on testing for and aggregate the uncertainty before to aggregate uncertainty for addition/
autocorrelation and for applying the emission factor. subtraction of data.
methods to make the terms
independent.
Independent errors: Apply Equation 4-6
to aggregate uncertainty for
Aggregate emissions for multiplication/ division of data.
the source as AF x EF. Correlated errors: Apply Equation 4-7
to aggregate uncertainty for
multiplication/ division of data.

Independent errors: Apply Equation 4-4

to aggregate uncertainty for addition/
Sum the emissions for subtraction of data.
multiple sources. Correlated errors: Apply Equation 4-5
to aggregate uncertainty for addition/
subtraction of data.

Apply the GWP values to

convert total emissions to GWP values are considered constants, so
CO2e basis and apply there is no aggregation of uncertainty for
GWP uncertainties. this step.

Independent errors: Apply Equation 4-4

to aggregate uncertainty for addition/
Sum the CO2e emissions subtraction of data.
and aggregate total Correlated errors: Apply Equation 4-5 to
uncertainty. aggregate uncertainty for addition/
subtraction of data.

Figure 4-5. Step C – Decision Diagram for Uncertainty Aggregation

Pilot Version, September 2009 4-31

 Table 4-13. Comparison of Annual Emission Estimates

Uncertainty (rel)
Context specific factor sensitivity
Tonnes Emissions, analysis
Month CO2/MMscf U(rel) MMscf tonnes CO2 10% 25% 100%
Emissions based on January 55.8856 6.95% 83.50 4,666.44 13.25% 26.47% 100.38%
monthly flow and February 55.7698 7.04% 78.28 4,365.38 13.30% 26.49% 100.38%
carbon content March 55.9697 7.04% 107.45 6,013.89 13.30% 26.49% 100.38%
April 56.1644 7.00% 103.83 5,831.27 13.27% 26.48% 100.38%
May 54.7437 7.65% 104.59 5,725.67 13.63% 26.66% 100.43%
June 55.2278 7.50% 102.87 5,681.48 13.55% 26.62% 100.42%
July 54.7069 7.61% 100.62 5,504.85 13.60% 26.65% 100.42%
August 55.4646 7.53% 106.03 5,880.73 13.56% 26.63% 100.42%
September 54.6648 7.63% 104.23 5,697.83 13.62% 26.65% 100.43%
October 54.6332 7.62% 102.28 5,587.75 13.61% 26.65% 100.43%
November 54.4964 7.62% 98.43 5,364.20 13.61% 26.65% 100.43%
Annual Total 1,187.42 65,550.85 3.91% 7.71% 29.10%
Emissions based on annual flow and annual
average carbon content 55.2180 3.08% 1,187.42 65,566.90 11.69% 25.72% 100.18%
Emissions based on annual flow and default
emission factor 55.7940 14.14% 1,187.42 66,250.86 18.09% 29.19% 101.13%
Note, the resulting emission estimates are not rounded to show the comparison.

Pilot Version, September 2009 4-32

4.5.1 Rounding-off of Statistical Estimate

Inappropriate rounding of the data can lead to errors in the final estimate. Using computer software, such
as spreadsheets, helps the user avoid rounding during intermediate steps. The estimate of emissions should
be rounded to smallest unit of measure (API, [Link], 1985). The uncertainty should be rounded to the
same number of digits as the estimate.

4.6 Assessing Data Correlations

As shown in Equations 4-5 and 4-7, the uncertainty propagation equation can be extended to account for
uncertainty terms that are correlated or not independent. A simple way to assess if the uncertainties are
correlated is to examine a graph of the uncertainties. If there is no pattern, the uncertainties are most likely
independent. If there is a pattern to the uncertainties, they are not independent.

The correlation between the uncertainties of two measured parameters can be calculated by applying the
following equation:

rU X ,UY =
∑ (U Xi ) (
− U X × U Yi − U Y ) (Equation 4-9)
(n − 1) × s (U X ) × s (U Y )

where
rUxUy = the correlation coefficient of the uncertainties for X and Y;
n= the sample size;
Uxi = the uncertainties associated with sample points from source X;
UX = the mean of the uncertainties from source X;
UY =the mean of the uncertainties from source y;
Uyi = the uncertainties associated with sample points from source Y;
s(UX) = the standard deviation for the uncertainties of source X; and
s(UY) = the standard deviation for the uncertainties of source Y.

Before combining the data it is important to eliminate the correlation of uncertainties, if possible. ISO
5168:2005(E) Annex F discusses methods for making measurement uncertainties independent, such as
calibrating instruments against different references and “redefining mathematical relationships to eliminate
correlations” (ISO, 2005).

Section A1.4.5 of the IPCC Good Practices document lists four sources for correlation:

• Use of common activity data for several emissions estimates;

• Mutual constraints on a group of emission estimates (such as a specified total fuel usage which
provides input to a number of processes);

Pilot Version, September 2009 4-33

 • The evolution of activity and emission factors associated with new processes, technology etc.,
which decouples the uncertainty from one time period to the next; and

• External drivers that affect a suite of emissions (economic, climatic, resource based) (IPCC, 2001).

To eliminate uncertainty correlations due to common activity data, IPCC suggests summing the emission
factor estimates and then multiplying the activity factor by the sum to obtain total emissions. For mutual
constraint, the IPCC Good Practices document suggests “to leave one of the proportions unspecified, and
to determine it by the difference between the other proportions and the total fraction.”

In addition, IPCC’s Good Practices document recommends that if there are more than two correlated
variables, a Monte Carlo simulation should be applied instead of uncertainty propagation.

4.6.1 Monte Carlo Simulation

Monte Carlo simulation is a more complex, model-based method for iteratively evaluating uncertainty
associated with individual parameters. It may be the preferred approach where more complex equations are
assessed. It is one of many methods for analyzing uncertainty propagation where the goal is to determine
how random variation, lack of knowledge, or error affects the sensitivity, performance, or reliability of the
resulting emission inventory.

Monte Carlo simulation is a “repeated sampling and calculation method” because the inputs are randomly
generated from probability distributions of the respective variables to simulate the process of sampling
from an actual population. Inputs are defined as probabilistic parameters rather than simple estimates. The
uncertainty model relies on repeated random sampling of all inputs and simultaneous recalculation of
emissions (outputs) to measure variation over the course of numerous model iterations. The data generated
from the Monte Carlo simulation can be represented as probability distributions (or histograms) or
converted to error bars and confidence intervals.

Monte Carlo simulation has an advantage over uncertainty propagation in that one can specify multivariate
distributions to account for correlations between different sources of uncertainty. The IPCC Good
Practices document recommends choosing one of the following distributions: normal, lognormal, uniform,
or triangular.1 The major difficulty in using the Monte Carlo simulation is the need to determine the
distribution of the data. If there are not enough data to assume normality by the Central Limit Theorem
(more than 30 data points), there are most likely not enough data to determine the underlying distribution of
the data. Consequently, such analyses are often forced to rely on subject matter expert opinion to
determine distributions rather than on observed data.

1 The IPCC Good Practice document discusses how to perform Monte Carlo simulation in Section 6.2 (IPCC, 2006).
It discusses choice of distribution in Section A1.2.5.

Pilot Version, September 2009 4-34

Additionally, Monte Carlo simulation can also be computationally intensive, with 10,000 simulations being
the norm. The bigger difficulty/barrier is the need to build a simulation model in Excel for uncertainty
analysis that reflects the complexity of the emissions estimation model and allows generation of the
variables required for the Monte Carlo simulation software.

To perform a Monte Carlo simulation, one must first determine the distribution of the data for each of the
uncertain variables used in the emissions model estimate. Ideally, each of these distributions should be
derived from data and knowledge of the underlying process. It is helpful in many instances to first graph
the data, and using the shape of the graph to determine the underlying distributions. The parameters that
define the distribution could be derived from the data. For example, a normal distribution is defined by its
mean and variance. If data are limited, one may have to rely on expert judgment to determine the
underlying distribution.

It is then necessary to statistically test the hypothesis that the data follow a certain distribution. The test
will vary based on the hypothesized distribution. For example, the Shapiro-Wilks test is often used to test
if the data are normal or lognormal. Options to test for other distributions include Empirical Distribution
Functions (Singh and Singh, 2006).

Once distributions are determined for all of the data sources, the Monte Carlo simulation will proceed by
randomly sampling each of the distributions that describe the data used for estimating emissions. As many
as 10,000 replicate samples are typically taken, with the total emissions being estimated for each replicate.
These repeated determinations of emission are used to generate a distribution of the total emissions with its
mean being the estimate of total emissions, and its uncertainty determined by its variance.

4.6.2 Comparison of Uncertainty Propagation and Monte Carlo

Section 6.3.1 of the IPCC Good Practices document compares the uncertainty propagation method and the
Monte Carlo simulations (IPCC, 2006). It notes that the uncertainty propagation method’s assumption of
normality leads to symmetric 95% confidence intervals whereas the Monte Carlo method can take into
account the fact that emissions are bounded below by zero to fit an asymmetric (and thus narrower)
confidence interval. If the data are skewed and one transforms the data (discussed earlier), one could
achieve the asymmetric confidence intervals using uncertainty propagation, as well.

Since the Monte Carlo simulations can assume a truncated distribution, the lower confidence limits tend to
be closer to the mean than the upper confidence limits. The IPCC Good Practices document goes on to
state that the two methods produce results that are fairly comparable. It recommends that countries report
the results of the uncertainty propagation method and those countries with “sufficient resources and
expertise” report Monte Carlo results as well.

Pilot Version, September 2009 4-35

These guidelines concentrate solely on the uncertainty propagation method due to the potential to introduce
further errors in assigning the probability distributions for the Monte Carlo simulations. As stated
previously, applying the uncertainty propagation methods, even where the assumptions are not met, is
advised in these guidelines, particularly for emission sources with a small contribution to the overall GHG
inventory. As data collection methods improve for GHG inventories, the ability to quantify uncertainties
will also improve.

Pilot Version, September 2009 4-36

5.0 UNCERTAINTY CALCULATION EXAMPLES
5.1 Introduction Section Focus
The section demonstrates the concepts provided in This section combines the guidelines,
procedures, and equations for calculating
Section 4.0 through examples taken directly from the uncertainty that were outlined in Section 4 and
2009 API Compendium of Greenhouse Gas Emissions applies them to a hypothetical facility to
calculate total uncertainty for the facility-level
Estimation Methodologies for the Oil and Gas Industry GHG inventory. The statistical approaches are
(API Compendium, 2009). Each of the examples also applied to two select refinery operations
to examine and compare uncertainty estimates
represents a hypothetical situation rather than any for different emission estimation
particular actual facility or operation. The facility methodologies.
example is based on a GHG inventory from an “Onshore This section will not describe how the
inventory was generated or calculated. For
Field with High CO2 Content,” as described in Section that background, the reader is referred to the
8.1.1 of the API Compendium. The uncertainties API Compendium (API, 2009). Much of the
emission calculation information is simply
associated with vented emissions from refinery catalytic reprinted directly from the API Compendium
cracking units and hydrogen plants are based on without elaboration in this report.
methodologies and examples presented in Sections 5.2.1 This section will describe how uncertainties
are assigned to each value in the inventory and
and 5.2.2 of the API Compendium, respectively. Details then propagated to a single overall uncertainty
on the source-by-source uncertainty calculations are associated with the total summed inventory for
the facility.
provided in Appendix F. Highlighted text in this section
is used to designate uncertainty calculations.

5.2 Example 1: Onshore Oil Field with High CO2 Content

5.2.1 Background

Tables 5-1 and 5-2 summarize the emission sources and uncertainty values associated with this facility.
This example field is described in detail in the API Compendium, but is summarized here for context. A
discussion of the uncertainty values associated with the example facility parameters follows.

Pilot Version, September 2009 5-1

 Table 5-1. Onshore Oil Field (High CO2 Content) Emission Sources
(Reprinted from the API Compendium)

Average Annual Activity

No. of Uncertainty Unit Capacity Uncertainty Operation (per Uncertainty Factor (all units Uncertainty
Source Fuel/Refrigerant Units (±%)a (per unit) (±%)a unit per year) (±%)a combined) (±%)a
Combustion Sources
Boilers Produced Gas 6 0 N/A N/A 40×106 scf/yr 15
6
Heaters/reboilers Produced Gas 3 0 2×10 Btu/hr 5 343 days/yr 2 53.2×106 scf/yr 6.71
Compressor engines
Produced Gas 11 0 N/A N/A 250×106 scf/yr 15
– turbines
6
Emergency flare Produced Gas 1 0 N/A N/A 500×10 scf/yr 15
Emergency 6
Diesel 1 0 1800 hp 5 200 hr/yr 10 2,912 ×10 Btu/yr 12.3
generator IC engine
Fire water pump IC 24 hr/yr; 10;
Diesel 1 0 460 hp 5 77.7 ×106 Btu/yr 13.2
engine 87% load 20
Fleet vehicles
Gasoline 5 0 N/A 40,000 mi/yr ea. 15.0 N/A
(trucks)
Vented Sources
Dehydration vents
30×106
(also has Kimray Produced Gas 1 0 5 343 days/yr 2 10,290×106 scf/yr 5.39
scf/day
pump emissions)
Central tank battery Crude Oil 1 0 6,100 bbl/day 5 343 days/yr 2 2,092,300 bbl/yr 5.39
Storage tanks Chemical 1 0 N/A N/A N/A
Naphtha 1 0 N/A N/A N/A
Glycol 1 0 N/A N/A N/A
Water blowdown N/A 1 0 N/A N/A N/A
tank
Slop oil tank Slop Oil 1 0 N/A N/A N/A
Amine unit for CO2 removal
Inlet Produced Gas N/A 30×106 5 343 days/yr 2 10,290×106 scf/yr 5.39
scf/day
Outlet Outlet Gas N/A N/A N/A 8,997×106 scf/yr 5
(0.5 mole% CO2)
Pneumatic devices Produced Gas 64 5 N/A N/A 64 pneumatic 5
devices

Pilot Version, September 2009 5-2

 Table 5-1. Onshore Oil Field (High CO2 Content) Emission Sources (continued)
(Reprinted from the API Compendium)

Unit Average Annual Activity

No. of Uncertainty Capacity Uncertainty Operation (per Uncertainty Factor (all units Uncertainty
Source Fuel/Refrigerant Units (±%)a (per unit) (±%)a unit per year) (±%)a combined) (±%)a
Vented Sources, continued
Chemical Produced Gas 67 5 N/A N/A 67 CIPs 5
injection pumps
(CIPs)
Vessel Produced Gas 112 0 N/A N/A 112 vessels 0
blowdowns
(non-routine)
Compressor Produced Gas 11 0 N/A N/A 11 compressors 0
starts (non-
routine)
Compressor Produced Gas 11 0 N/A N/A 11 compressors 0
blowdowns
(non-routine)
Well workovers Produced Gas 24 0 N/A N/A 24 well workovers 0
(non-routine)
Pressure relief Produced Gas 482 1 N/A N/A 482 PRVs 1
valves
Fugitive
Sources
Equipment leaks Produced Gas N/A N/A 8,760 hr/yr 0 See Table 5-2
Fleet vehicle R-134a 5 0 N/A N/A Unknown 5
refrigeration trucks
Indirect Sources
Electricity N/A N/A N/A N/A 917 MW-hr/yr 2
consumed
Footnote:
Note: the values shown above are for example only. They do not reflect actual operations.
a
Uncertainty is based on engineering judgment at a 95% confidence interval.

Pilot Version, September 2009 5-3

 Table 5-2. Onshore Oil Field (High CO2 Content) Fugitive Emission Sources
(Reprinted from the API Compendium)

Average Component Uncertainty (±%)a

Component Service Count
Valves Liquid and Gas 2,740 75
Pump seals Liquid and Gas 185 75
Connectors Gas 110 75
Flanges Liquid and Gas 10,000 75
Open-ended lines Gas 6 75
Others Liquid and 710 75
Footnotes:
Note, the values shown above are for example only. They do not reflect actual operations.
a
Uncertainty is based on engineering judgment at a 95% confidence interval.

Facility Description: An onshore oil field in Texas consists of 320 producing oil wells.

Throughput: The average daily oil and gas production rates are 6,100 bbl/day and 30×106 scf/day,
respectively.

Operations: The facility operates approximately 343 days per year. The facility imports 917 MW-hr
annually from the eGRID subregion “ERCOT all”. The facility gas composition is presented in Table 5-3
and results in a heating value of 928 Btu/scf with an uncertainty of ± 4%, based on engineering judgment.

Table 5-3. Gas Composition for Onshore Oil Field (High CO2 Content)
(Reprinted from the API Compendium)

Gas Compound Produced Gas Mole % Uncertainty a

(±%)
CO2 12 4
N2 2.1 4
CH4 80 4
C2H6 4.2 4
C3H8 1.3 4
C4H10 0.4 4
Footnote:
a
Uncertainty is based on engineering judgment at a 95% confidence interval.
Note: the values shown above are for example only. They do not reflect average operations.

Most of the GHG emissions from this facility are from combustion sources. A small additional amount
comes from vented sources since the oil field gas contains a relatively high CO2 content (12 mole %).

In this example, emissions from some emission sources are calculated by use of emission factors, using the
following simple equation:

Pilot Version, September 2009 5-4

Emission Factor (EF) × Activity Factor (AF) = Total emissions from that source

For these emission sources, the first task in the calculation of total inventory uncertainty is assigning
uncertainties to each of the activity factors and emission factors that were used in the inventory. Some
emission calculation equations have multiple terms, rather than only two simple terms. For example:

CO 2 from flares = CO 2 in the gas + CO 2 formed by combustion

= ( flared gas rate × CO 2 gas composition ) + ( gas rate × hydrocarbon composition × combustion efficiency )

The same approaches discussed below for assigning uncertainties to simple activity factors will apply to
each term in a more complicated equation. Further detail is also provided in Section 4.0.

5.2.2 Assigning Uncertainties to Activity Factors

Sites may have a combination of directly measured emissions and emissions calculated using activity factors
and emission factors. For direct measurements of emissions, the equations and decision trees provided in
Section 4.0 can be used to quantify uncertainties.

This section discusses assignment of uncertainties to the activity factors. In some cases, measured data will
allow assignment of calculated confidence bounds, as covered in detail in the Section 4.0 examples.
However, there will be many cases where confidence bounds on activity factor data will be assigned by
expert judgment. This section covers how expert judgment assigned some of those bounds in this
hypothetical case.

Some assignment of expert judgment may require that the expert look at the activity value, say a total count
of 67 chemical injection pumps, and decide how many pumps a person might miss (undercount), or how
many devices might be non-operating, or incorrectly identified as chemical injection pumps (over counted).
If the expert, familiar with the count, decided that at most three pumps were under or over counted, the
uncertainty limit might be assigned simply to be 3/67 = 4%. In this case, the company expert decided that a
blanket 5% uncertainty was more appropriate. Some general categories of uncertainty assignments follows.

Perfectly-known Activity Factors

In this hypothetical example, some of the activity factors were simply counts of large equipment types at the
facility. These counts, such as the number of boilers at the facility, are known with absolute certainty since
they are discreet values and were produced by company personnel intimately familiar with the specific
facility. Therefore, the 95% confidence bounds (shown as “% uncertainty”) of the activity factor value is
zero (0), since the value is perfectly known. Table 5-1 shows uncertainties assumed for each activity factor.
The perfectly known activity factors are in the second column under “No. of Units” and the uncertainties

Pilot Version, September 2009 5-5

associated with them are assigned by expert judgment to be 0% in the third column. The reader should note
that if these counts of equipment had been produced in a different way, for example if they were taken from
an industry average of equipment counts at this general type of facility, uncertainty then would have to be a
real value, not zero. In that case, confidence bounds might be provided with the industry average, or may
have to be assigned again by expert judgment.

Well-known Activity Factors

Counts of other devices on site, even if produced by company personnel, may not be perfectly known. It is
possible to know the exact number of amine units with perfect certainty, but not to know some less
significant equipment components. In this hypothetical case, counts were taken from the exact facility, such
as counts of pressure relief valves (PRVs), pneumatic devices, and chemical injection pumps. However,
some small uncertainty was assigned to these counts by expert judgment since the counts are not perfectly
known. In this hypothetical case, uncertainties between 1% and 5% were assigned to the counts of PRVs,
pneumatic devices, and chemical injection pumps. For unit capacities, the fifth column in Table 5-1, these
capacities are well known, and expert judgment assigned them to be ±5%. In this hypothetical example,
“days of operation” are supplied by company experts. In this example, these estimates were not based on
measured operating hours but were estimated based on operator logs. An uncertainty of 2% was assigned
by expert judgment.

Approximated Activity Factors

For other sources, such as counts of valves, seals, flanges for fugitive emissions, there may be larger
uncertainties associated with the activity factor counts. This is true even if the count were produced for the
exact facility. The uncertainty is determined using expert judgment based on the quality of the component
inventory and the length of time since the last inventory of fugitive emission sources, since changes to the
facility and therefore component counts are possible. In this hypothetical example, the counts were not
taken for this exact facility, but were applied from average component count information from industry-
wide counts of fugitive components from this general type of facility. Therefore expert judgment assigned a
much wider uncertainty to each count of ±75%. In cases where exact component counts were taken for this
facility, the uncertainty percentage would be much lower. Table 5-2 shows the activity factor counts for
fugitive components (reprinted directly from the API Compendium), and the assigned uncertainties.

Measured Activity Factors

In this hypothetical example, some of the activity factors were flow rate information measured by on-site
totalizing meters that record data. These flow rate measurements were assigned a ±15% uncertainty by

Pilot Version, September 2009 5-6

expert judgment. The assignment is based upon company knowledge that the meters are not regularly
calibrated. While many meters can have much better accuracy, without a quality control (QC) program
including calibrations, for this example the expert determined that the meters of this type have an accuracy
of within ±15%. If a QC program were in place, or if independent multiple repeat measurements had
instead been used, the user could have calculated the exact uncertainty and the confidence bounds, using the
equations from Section 4.0 based on a data set of repeat measurements. No expert judgment would be
required if that were the case. It is likely that the confidence bounds would be much tighter (lower) for
repeat measurements.

5.2.3 Propagating Uncertainty in Calculated Activity Factors

Assignment of uncertainty to the most detailed level of activity data used in the inventory is the preferred
approach. Then statistical methods are used to propagate the uncertainty values.

For example, in Table 5-1, the sum of the Annual Activity Factors is calculated as:

Annual Activity Factor = Number of Units × Unit Capacity × Average Operations Time

Each term that is multiplied to result in the annual activity factor has its own uncertainties. For sources
such as the six boilers, there is no (0%) uncertainty associated with the count of equipment since it is
know that there are exactly six boilers at this facility. The uncertainty in the unit capacity is the
uncertainty in the unit’s average operational capacity.

Uncertainties for these combustion sources are calculated by applying Equation 4-6 and using the relative
uncertainty values.

U (rel ) Activity Factor = U (rel ) 2Number Of Units + U (rel )Unity

2
Capacity + U ( rel ) AverageOperations
2

For example, the uncertainty for the activity factor for the heaters/reboilers is calculated using the values
shown in Table 5-1:

U (rel ) Activity Factor = U (rel ) 2Number Of Units + U (rel )Unity

2
Capacity + U ( rel ) AverageOperations
2

U (rel ) Activity Factor = 02 + 52 + 22 = 5.39%

Appendix F provides details on the uncertainty estimates for each emission source included in this example.
The resulting uncertainty estimates for this example facility are provided in Table 5-4 and Appendix F.

Pilot Version, September 2009 5-7

 Table 5-4. Onshore Oil Field (High CO2 Content) Emissions

Source CO2 CH4 N2O and Other GHG Total Emissions, CO2 Eq
Uncertainty Uncertainty Uncertainty Uncertainty
Emissions % Emissions % Emissions % Emissions %
Source Type (tonnes/yr) Lower Upper (tonnes/yr) Lower Upper (tonnes/yr) Lower Upper (tonnes/yr) Lower Upper
Combustion Boiler/heaters 5,200 8.78 8.78 0.0865 26.1 26.1 0.0242 100 150 5,210 8.77 8.77
Sources Natural gas engines 13,900 15.7 15.7 0.904 29.4 29.4 0.325 100 151 14,100 15.6 15.6
Emergency generator IC
0.0108 27.8 27.8 0.00175 100 151
engine 219 15.6 15.6 220 15.5 15.5
Fire water pump IC engine 0.00112 100 106 0.0000467 100 151
Flares 27,400 23.4 23.4 153 25.3 25.3 0.223 100 200 30,700 21.1 21.1
Fleet vehicles 127 19.4 19.4 0.00643 100 151 0.00871 100 151 129 19.1 19.2
Combustion Total 46,800 14.5 14.5 154 25.2 25.2 0.582 68.9 114 49,900 13.7 13.7
Vented sources Dehydration and Kimray
105 77.5 77.5 254 77.5 77.5 NA NA NA 5,440 76.0 76.0
pump vents
Tanks – flashing losses 775 90.4 90.4 1,880 90.4 90.4 NA NA NA 40,300 88.7 88.7
Amine unit 62,600 6.97 6.97 193 100 119 NA NA NA 66,700 8.94 9.77
Pneumatic devices 64.6 50.2 50.2 157 50.2 50.2 NA NA NA 3,360 49.2 49.2
Chemical injection pumps 48.6 100 108 118 100 108 NA NA NA 2,530 98.1 106
Vessel blowdowns 0.0702 100 326 0.171 100 326 NA NA NA 3.65 98.1 319
Compressor starts 0.745 100 190 1.81 100 190 NA NA NA 38.7 98.1 187
Compressor blowdowns 0.333 100 179 0.808 100 179 NA NA NA 17.3 98.1 175
Well workovers 0.0181 100 300 0.0439 100 300 NA NA NA 0.939 98.1 294
Other non-routine (PRVs) 0.131 100 310 0.318 100 310 NA NA NA 6.81 98.1 319
Vented Total 63,600 6.95 6.95 2,610 66.3 66.5 NA NA NA 118,000 30.9 31.0

Pilot Version, September 2009 5-8

 Table 5-4. Onshore Oil Field (High CO2 Content) Emissions, continued

Source CO2 CH4 N2O and Other GHG Total Emissions, CO2 Eq
Uncertainty Uncertainty Uncertainty Uncertainty
Emissions % Emissions % Emissions % Emissions %
Source Type (tonnes/yr) Lower Upper (tonnes/yr) Lower Upper (tonnes/yr) Lower Upper (tonnes/yr) Lower Upper
Fugitive Sources Fugitive components NA NA NA 52.6 66.2 83.3 NA NA NA 1,100 66.2 83.3
Fleet vehicle
NA NA NA NA NA NA 0.00100 100 112 1.30 100 112
refrigeration, R-314a
Fugitive Total NA NA NA 52.6 66.2 83.3 0.00100 100 112 1,100 66.1 83.2
Indirect Sources Electricity consumed 551 10.2 10.2 0.00776 100 100 0.00628 100 100 553 10.2 10.2
Indirect Total 551 10.2 10.2 0.00776 100 100 0.00628 100 100 553 10.2 10.2
TOTAL (tonnes of each gas) 111,000 7.29 7.29 2,820 61.4 61.7 N2O:0.588 67.1 113
R134a:
0.00100 100 112
TOTAL (CO2e) 111,000 7.29 7.29 59,100 61.4 61.7 184 66.6 112 170,300 21.9 21.9

Pilot Version, September 2009 5-9

5.2.4 Propagating Assymetric Uncertainty Distribution

Appendix F demonstrates the use of an asymmetric uncertainty distribution. Where the uncertainty values for
emissions data were greater than 100%, the lower bound estimate was truncated at -100% and tracked
separately from the upper bound estimate. The upper and lower bound uncertainties were then propagated
through to the total CO2 equivalent emissions for the facility. For this example, the resulting difference
between the upper and lower uncertainty values was very small.

5.3 Example 2: Refinery

5.3.1 Background

Example 1 provides a very detailed, step-by-step demonstration of the uncertainty calculations for a crude oil
production facility. Because many of the source types are similar across all industry sectors, this example
focuses on a few distinct refinery process units. A comparison in the uncertainty calculations is provided for
methodologies presented in the API Compendium for fluid catalytic cracking unit (FCCU) and hydrogen plant
emissions. As with the previous example, much of the information is reprinted directly from the API
Compendium without change.

5.3.2 Uncertainty Comparison for FCCU Emission Estimation Methods

The following example applies uncertainty calculations to Exhibit 5.6 from the API Compendium for the
FCCU GHG emission estimation methods presented in the API Compendium. The same operating
parameters specified in the API Compendium are applied here, with the following assignment of uncertainties
added.

• The catalytic cracking unit has a coke burn rate of 119,750 tonnes per year ± 15% and a blower air
capacity of 2,150 m3/min ± 15% (assigned by expert judgment). The air blower is assumed to
operate continuously for the year (a ± 2% uncertainty is applied to this assumption).
• The carbon fraction of the coke is 0.93 ± 5.5% based on site-specific data (determined from
measured compositions).
• The flue gas concentrations are 11% for CO2 and 9% for CO exiting the regenerator. Table D-3 of
this uncertainty document provides reproducibility values for the precision of Reformed Gas
Samples based on ASTM 1946-90. For molar compositions between 5 and 25 percent, a
reproducibility factor of 0.5 applies. An additional 5% uncertainty is assigned by expert judgment
to account for potential variability in the composition.
• It is assumed that no CH4 is formed during the regeneration process.
• A CO boiler is used for control of the flue gas stream. Supplemental firing with natural gas is
employed at a rate of 100×106 ± 5% Btu/hr on a higher heating value basis.

The API Compendium presents three equations for estimating CO2 emissions from FCCUs. The following
demonstrates the uncertainty quantification for each of the three methods.

Pilot Version, September 2009 5-10

Uncertainty for Regenerator CO2 Emissions – Coke Burn Rate Approach
(Compendium Equation 5-4)
Applying API Compendium Equation 5-4, the estimated CO2 emissions from the regenerator would be:

tonnes Coke Burned 0.93 tonnes C 44 tonnes CO 2

E CO2 = 119,750 × × = 408,348 tonnes CO 2 /year
year tonnes Coke 12 tonnes C

The uncertainty is calculated by applying Equation 4-6 and using the relative uncertainty values.

U (rel )CO2 = U (rel )Coke

2
burned + U ( rel ) C content
2

U (rel )CO2 = 152 + 5.52 = 15.98%

Uncertainty for Regenerator CO2 Emissions – “K1, K2, K3” Approach

(API Compendium Equation 5-5)

Applying the air rate to API Compendium Equation 5-5, the CO2 emission estimate is:

⎡ 0.2982 kg - min 2,150 dscm ⎤ 44 tonne 8760 hr

E CO2 = ⎢ × × (11% + 9% ) ⎥ × × × = 411,862 tonnes CO 2 /yr
⎣ hr - dscm % min ⎦ 12 1,000 kg yr

Note, the K1 term (0.2982 kg-min/hr-dscm%) is a constant.

For this calculation, the uncertainty associated with the sum of the CO2 and CO is determined first. This
uncertainty is calculated by applying Equation 4-4, using the absolute uncertainties.

U (abs ) concentrations = U (abs ) 2Reproducibility + U (abs ) 2Variability

0.743
U (abs ) CO2 = 0.5 2 + (0.05 × 11) 2 = 0.743 U (rel ) CO2 = 100% × = 6.76%
11
0.673
U (abs ) CO = 0.5 2 + (0.05 × 9) 2 = 0.673 U (rel ) CO2 = 100% × = 7.47%
9

Equation 4-4, using the absolute uncertainties, is also applied to combine these compositions.

U (abs ) concentrations = U (abs ) CO

2
2
+ U (abs ) CO
2

U (abs ) concentrations = 0.743 2 + 0.673 2 = 1.002

1.002
U (rel ) concentrations = 100% × = 5.01%
11 + 9

The CO2 uncertainty is then calculated by applying Equation 4-6 and using the relative uncertainty values.

Pilot Version, September 2009 5-11

U (rel )CO2 = U (rel ) 2Air rate + U (rel )CO
2
and CO2 + U ( rel ) Annual hours
2

U (rel )CO2 = 152 + 5.012 + 2 2 = 15.94%

Uncertainty for Regenerator CO2 Emissions – Air Blower Rate Approach

(API Compendium Equation 5-6)

Using the air rate in API Compendium Equation 5-6 yields:

2150 m3 ⎛ 0.11 m3 CO 2 0.09 m3 CO m3 CO 2 ⎞ 44 kg CO 2 /kgmole CO 2

E CO2 = ×⎜ 3
+ 3
× 3 ⎟×
min ⎝ m gas m gas m CO ⎠ 23.685 m3 CO2 /kgmole CO2
525,600 min tonnes
× ×
year 1000 kg

E CO2 = 419,859 tonnes CO 2 /year

From an uncertainty perspective, this calculation is the same as shown for the “K1, K2, K3” Approach (API
Compendium Equation 5-5). Both equations apply the summed composition of CO2 and CO, the air rate,
and an annual operational time. The uncertainty of 15.94%, calculated above, would apply for this
approach as well.

Uncertainty for Supplemental Natural Gas Firing

The emissions from the supplemental firing are in addition to the CO2 emissions from the FCCU
regenerator. Emissions from the supplemental firing of natural gas are estimated using the combustion
emission approaches presented in API Compendium Section 4. For CO2, the emission factor is taken from
API Compendium Table 4-3 for pipeline natural gas.

100 × 106 Btu 0.0531 tonne CO 2 8760 hr

E CO2 = × ×
hr 106 Btu yr

E CO2 = 46,516 tonnes CO 2 / yr

The CO2 uncertainty is calculated by applying Equation 4-6 and using relative uncertainty values. API
Compendium Table 4-3 does not provide an indication of the uncertainty associated with this emission
factor, so a value of 5% is assigned by expert judgment.

U (rel )CO2 = U (rel ) 2Heat rate + U (rel ) Emission

2
Factor

U (rel )CO2 = 52 + 52 = 7.07%

The CH4 emission factor is taken from API Compendium Table 4-7 for natural gas fired boilers.

Pilot Version, September 2009 5-12

 100 × 106 Btu 1.0 ×10-6 tonne CH 4 8760 hr
E CH4 = × ×
hr 106 Btu yr

E CH4 = 0.88 tonnes CH 4 / yr

The uncertainty is calculated by applying Equation 4-6 using the relative uncertainty values. The CH4
emission factor has a “B” quality rating assigned to it. Based on this, an uncertainty of 10% is applied to
the emission factor.

U (rel )CH 4 = U (rel ) 2Heat rate + U (rel ) 2Emission Factor

U (rel )CH 4 = 52 + 102 = 11.2%

The N2O emission factor is also taken from API Compendium Table 4-7 for controlled natural gas fired
boilers.

100 × 106 Btu 9.8 × 10-7 tonne N 2 O 8760 hr

E N2O = × ×
hr 106 Btu yr
E N2O = 0.86 tonnes N 2 O / yr

The uncertainty is calculated by applying Equation 4-6 using the relative uncertainty values. The N2O
emission factor has an “E” quality rating assigned to it. Based on this, an uncertainty of 50% is applied to
the emission factor.

U (rel ) N2O = U (rel ) 2Heat rate + U (rel ) 2Emission Factor

U (rel ) N2O = 52 + 502 = 50.2%

Table 5-5 summarizes the emission and uncertainty estimates for the different FCCU methodologies.
Equation 4-4, using the absolute uncertainty values, is applied where the emission estimates are summed.

Based on the assumptions applied for the FCCU methodologies, the uncertainty associated with each of the
methods provided in the API Compendium is comparable. In all three equations, the aggregated uncertainty is
influenced primarily by the ± 15% uncertainty values assigned to the coke burn rate (used in the first and third
methods) and the blower air capacity (used in the second method).

Pilot Version, September 2009 5-13

 Table 5-5. Uncertainty Comparison for FCCU Estimation Methods
Coke burn rate Contribution CO2 CH4 N2O
approach Coke Burn 408,348 ± 16.0%
(Equation 5-4) CO Boiler 46,516 ± 7.07% 0.88 ± 11.2% 0.86 ± 50.2%
Total 454,864 ± 14.4% 0.88 ± 11.2% 0.86 ± 50.2%
Total, CO2 Eq. 454,149 ± 14.4%
“K1, K2, K3” Contribution CO2 CH4 N2O
approach Coke Burn 411,862 ± 15.9%
(Equation 5-5), CO Boiler 46,516 ± 7.07% 0.88 ± 11.2% 0.86 ± 50.2%
Total 458,378± 14.3% 0.88 ± 11.2% 0.86 ± 50.2%
Total, CO2 Eq. 458,663 ± 14.3%
Air blower rate Contribution CO2 CH4 N2O
approach Coke Burn 419,859 ± 15.9%
(Equation 5-6)
CO Boiler 46,516 ± 7.07% 0.88 ± 11.2% 0.86 ± 50.2%
Total 466,375± 14.4% 0.88 ± 11.2% 0.86 ± 50.2%
Total, CO2 Eq. 466,660 ± 14.4%
Rounding is limited to the uncertainty values to facilitate comparison.
5.3.3 Uncertainty Comparison for Hydrogen Plant Emission Estimation Methods

The API Compendium provides two rigorous calculation approaches for estimating the CO2 generation rate
from the hydrogen plant, both using a specific feed gas composition. The approaches are based on either the
volume of feedstock used or the hydrogen production rate. This section compares the uncertainty estimates
associated with these methods. For this comparison, the following operating parameters and uncertainties are
assigned:

• A hydrogen plant has a feedstock rate of 3×109 ± 15% standard cubic feet per year and produces
13×109 ± 15% standard cubic feet of hydrogen per year.
• The feed gas composition (molar basis) is CH4 = 85%, C2H6 = 8%, C4H10= 3%; the balance is inerts
(assume N2 for the inerts). Table D-2 of this Uncertainty document provides reproducibility
uncertainty associated with natural gas samples. These values can be applied to account for the
measurement error in the composition sample. An additional 5% uncertainty is assigned by expert
judgment to account for potential variability and bias in the gas composition during the month.
• It is assumed that no CH4 is entrained in the hydrogen product.

For both methods, the uncertainty associated with the feedstock composition is needed. The combined
uncertainty of the reproducibility and variability is calculated by applying Equation 4-4, using the absolute
uncertainties. The calculation is demonstrated for ethane below; results for all of the compounds are shown in
Table 5-6.

Pilot Version, September 2009 5-14

U (abs )Composition = U (abs ) 2Re producibility + U (abs )Variability
2

U (abs ) Ethane = 0.122 + 0.42 = 0.4176

0.4176
U (rel ) = × 100% = 5.22%
8

Table 5-6. Composition Data

Combined
Uncertainty wt%
Reproducibility Variability (rel) MW Carbon wt%
mole% Uncertainty Uncertainty (applies to Calculation, Calculation, C
dry (abs), mole% (Uabs=mole%×5%) MW) lb/lbmole % U(rel)
Methane 85 0.15 4.25 5.00% 13.6340 53.96 6.24%
Ethane 8 0.12 0.4 5.22% 2.4056 10.16 6.41%
Butane 3 0.1 0.15 6.01% 1.7436 7.62 7.07%
N2 4 0.1 0.2 5.59% 1.1200 0.00 6.72%
Total 100 18.9032 71.7339
Sum uncertainty
(abs) 0.7042 3.4704
Sum uncertainty
(rel) 3.73% 4.84%

The molecular weight of the gas is calculated by applying the following equation:

1 # compounds
MWMixture = × ∑ (Mole% i × MWi )
100 i =1

This results in 18.9 lb/lbmole, as shown in Table 5-6.

For each individual gas compound, the relative uncertainty of the mole%i × MWi is equivalent to the
combined reproducibility and variability uncertainties, since the molecular weight of each gas compound is a
constant. The aggregated uncertainty associated with the mixture’s MW is calculated by applying
Equation 4-4, using the absolute uncertainties for each molar compound.

U (abs )
∑ ( MW Total )
= ∑U (abs) 2
mole %× MW

U (abs ) = (0.0500 ×13.6340) 2 + (0.0522 × 2.4056) 2 + (0.0601×1.7436) 2 + (0.0559 ×1.12) 2

∑ ( MW Total )
U (abs ) = (0.6821) 2 + (0.1256) 2 + (0.1048) 2 + (0.0626) 2 = 0.7042
∑ ( MW Total )
0.7042
U (rel ) = × 100% = 3.73%
∑ ( MW Total )
18.9032

The weight percent carbon of the average composition is calculated by applying the following equation:

Pilot Version, September 2009 5-15

 lbmolei x lbmole C 12 lb C 1
Wt% CTotal = ∑ × × ×
100 lbmole total lbmolei lbmole C MWTotal

This equates to 71.7% C, as shown in Table 5-6.

For each compound in the gas mixture, the uncertainty from this calculation is determined by applying
Equation 4-6, using the relative uncertainties of the lbmolei and MWTotal. This is demonstrated for ethane.

U ( R e l )Wt %Ci = U (rel ) 2mole %i × U (rel ) 2MW Total = 5.222 + 3.732 = 6.41%

The uncertainty associated with the wt% C of the mixture is calculated by applying Equation 4-4, using the
absolute uncertainties for the weight percent carbon of each molar compound.

U (abs )
∑ ( wt %)
= ∑U (abs) 2
Wt %

U (abs ) = (0.0624 × 53.96) 2 + (0.0641×10.16) 2 + (0.0707 × 7.62) 2 + (0.0672 × 0) 2

∑ ( wt %)
U (abs ) = 3.4704
∑ ( wt %)
3.4704
U (rel ) = × 100% = 4.84%
∑ (Wt %) 71.7339

The uncertainties associated with the composition are used in the following comparison of the emission
estimation methodologies for H2 plants.

Uncertainty for H2 Plant Emissions – Feedstock Rate Approach

(API Compendium Equation 5-8)
The CO2 emissions can be calculated using the feedstock rate and carbon content, applying Equation 5-8 from
the API Compendium:

3 × 10 9 scf lbmole feed 18.9 lb feed 0.7173 lb C 44 lb CO2 tonne

E CO2 = × × × × ×
yr 379.3 scf lbmole feed lb feed 12 lb C 2204.62 lb
ECO2 = 178,336 tonnes CO2 / yr

The CO2 uncertainty is then calculated by applying Equation 4-6 and using the relative uncertainty values.

U (rel )CO2 = U (rel ) 2Feedstock rate + U (rel ) MW

2
feedstock + U ( rel ) C content feedstock
2

U (rel )CO2 = 152 + 3.732 + 4.842 = 16.2%

Uncertainty for H2 Plant Emissions – H2 Production Approach
(API Compendium Equation 5-9)
A second approach for estimating emissions associated with a hydrogen plant is based on the H2 production
rate rather than the feedstock rate. This approach applies the stoichiometric ratio of H2 formed to CO2

Pilot Version, September 2009 5-16

formed, as shown in API Compendium Equation 5-9. For this example, the chemical reaction for each
compound is:

CH4: CH4 + 2H2O = 4H2 + 1CO2

C2H6: C2H6 + 4H2O = 7H2 + 2CO2
C4H10: C4H10 + 8H2O = 13H2 + 4CO2
For this approach, the moles of carbon and hydrogen are determined by multiplying the number of molecules
of each in each compound by the composition of each compound in the feed gas (i.e., CH4, C2H6, and C4H10).
These results are used to determine the ratio of moles of carbon to moles of H2, and are shown in Table 5-7.

Table 5-7. Hydrogen and Carbon Composition Data

Combined
#C # H2 Uncertainty (rel)
Compound Atoms Molecules Concentration (mole%) Moles C Moles H2
CH4 1 4 0.85 5.00% 0.85 3.4
C2H6 2 7 0.08 5.22% 0.16 0.56
C4H10 4 13 0.03 6.01% 0.12 0.39
Total Moles 1.13 4.35
Sum uncertainty
0.0439 0.0439
(abs)
Sum uncertainty
3.89% 3.89%
(rel)

The carbon-to-hydrogen ratio is calculated by dividing the total moles C by the total moles H2 (1.13/4.35 =
0.26). Since each mole of carbon produces 1 mole of CO2, the CO2/H2 ratio is the
same as the C/H2 ratio (0.26).
The uncertainty associated with the total moles of C and H2 is calculated by applying Equation 4-4, using the
absolute uncertainties of each molar compound.

U (abs )
∑ ( mole C)
= ∑U (abs) 2
mole C

U (abs ) = (0.0500 × 0.85) 2 + (0.0522 × 0.08) 2 + (0.0601× 0.03) 2

∑ ( mole C)
U (abs ) = 0.0439
∑ ( mole C)
0.0439
U (rel ) = ×100% = 3.89%
∑ ( mole C)
1.13

U (abs )
∑ ( mole H2 )
= ∑U (abs) 2
mole H 2

U (abs ) = (0.0500 × 3.4) 2 + (0.0522 × 0.53) 2 + (0.0601× 0.39) 2

∑ ( mole H2 )
U (abs ) = 0.1742
∑ ( mole H2 )
0.1742
U (rel ) = × 100% = 4.00%
∑ ( mole H2 ) 4.35

Pilot Version, September 2009 5-17

The CO2 emissions are calculated using Equation 5-9:
106scf H 2 lbmole H 2 lbmole gas 1.13 lbmole CO 2 44 lb CO 2 tonne
E CO2 = 13,000 × × × × ×
year 379.3 scf 4.35 lbmole H 2 lbmole gas lbmole CO2 2204.62 lb

E CO2 = 177,700 tonnes CO 2 /yr

The CO2 uncertainty is then calculated by applying Equation 4-6 and using the relative uncertainty values.

U (rel )CO2 = U (rel ) 2Feedstock rate + U (rel ) 2moles C + U (rel ) 2moles H2

U (rel )CO2 = 152 + 3.892 + 42 = 16.0%

Based on the assumptions applied for the H2 plant emission methodologies, the uncertainty associated with
each of the methods provided in the API Compendium is comparable. In both equations, the aggregated
uncertainty is influenced primarily by the ±15% uncertainty values assigned to the feedstock rate (used in the
first method) and the H2 production rate (used in the second method).

5.4 Strategic Reduction of Uncertainty

This section addresses the potential need to refine the emission inventory to reduce the uncertainty in the
overall emission estimate. There may be several reasons to do this. In some specific locations, there may be
state or national regulations, or guidelines for voluntary programs that suggest or require an emission
inventory to have uncertainties lower than a certain percentage level. A company may also independently
wish to refine its own uncertainty limits, if it considers the uncertainties too large.

The reader should note that none of the strategies mentioned here are aimed at reducing the actual emissions
of GHGs. That is a separate subject, and while emission reductions are achievable in some cases, they are
beyond the scope of this report. This section focuses on reducing the mathematical and statistical uncertainty
associated with an existing emission inventory.

Once a decision has been made to refine and reduce the uncertainty associated with a given inventory, some
strategic analysis of the major sources contributing to the uncertainty is in order. It is important to know
which sources significantly contribute to the overall inventory. It would make little sense to refine a term
with large confidence bounds, but that contributed very little to the overall inventory. It may also be useful to
have a target uncertainty in mind. For example, if the current total uncertainty is ±50.0 % and the company
wishes to reduce it to ±25.0%, then that target is useful in the analysis. As uncertainties of individual values
in the calculations are examined, values that have uncertainties that are already at or below the 25.0% target,

Pilot Version, September 2009 5-18

are less likely to be fruitful targets for reduction. The goal is to find the largest contributors to emissions that
have the largest uncertainties.

The largest contributors to uncertainty can be determined by multiplying the emission estimate by the
maximum error bound for each source. This would result in the upper bound emission estimate for the
particular source. The emission estimates can then be sorted by the largest contributors. This is demonstrated
in Table 5-8 for the example crude oil production facility.

Table 5-8. Emission Uncertainty Ranking for Onshore Oil Production Example

Maximum
Emissions Emissions,
(tonnes Maximum (tonnes
Source Type Source CO2e /yr) Uncertainty, % CO2e /yr) Ranking
Combustion Boiler/heaters 5,210 8.77 5,661 6
Sources Natural gas engines 14,100 15.6 16,244 4
Diesel engines 220 15.5 254
Flares 30,700 21.1 37,123 3
Fleet vehicles 129 19.2 154
Vented Sources Dehydration and Kimray pump vents 5,440 76.0 9,571 5
Tanks – flashing losses 40,300 88.7 76,039 1
Amine unit 66,700 9.77 73,216 2
Pneumatic devices 3,360 49.2 5,013 8
Chemical injection pumps 2,530 106 5,215 7
Vessel blowdowns 3.65 319 15.3
Compressor starts 38.7 187 111
Compressor blowdowns 17.3 175 47.6
Well workovers 0.939 294 3.70
Other non-routine (PRVs) 6.81 319 28.5
Fugitive Fugitive components
1,100 83.3 2,025 9
Sources
Fleet vehicle
1.30 112 2.75
refrigeration, R-314a
Indirect Electricity consumed
553 10.2 609 10
Emissions

Figure 5-1 provides an illustration of the emissions (which are not expressed in CO2e, but simply in tonnes of
each type of gas) for the onshore oil field which is a reprint from the API Compendium.

Pilot Version, September 2009 5-19

 60,000 Carbon Dioxide

GHG Emissions, tonnes/yr

Methane
50,000 Nitrous Oxide
Other GHGs
40,000

30,000

20,000

10,000

0
Combustion Vented Fugitive Indirect
Emission Source Category

Figure 5-1. Onshore Oil Field: Summary of Emissions

Figure 5-2 illustrates emissions for the facility with each emission category and each gas converted to CO2e,
using each gas’ GWP. It should be noted that although the GWP itself has uncertainty associated with it, this
report treats the GWP as a selected constant. Therefore, no uncertainty is associated or propagated from the
GWP values. Figure 5-2 also illustrates the bounds of uncertainty (at 95% confidence) for each emission
source category.

5.4.1 Opportunities to Improve the Uncertainty Estimate for the Onshore Oil Field

Each facility should examine the major categories of emissions and emission uncertainty, and then examine
specific emission sources within the category. Prioritizing the largest sources of uncertainty can be done with
this simple approach.

Among the major types of emissions in Figure 5-2, there are three very significant emission categories:
1) combustion sources of CO2; 2) vented sources of CO2; and 3) vented sources of CH4. These are the most
significant source of greenhouse gas emissions for this facility. Together they comprise almost 95% of all
GHG emissions from the facility.

Pilot Version, September 2009 5-20

 90,000
Carbon Dioxide
80,000 Methane
Nitrous Oxide
CO2 Equilvalents, tonnes/yr 70,000 Other GHGs

60,000

50,000

40,000

30,000

20,000

10,000

0
Combustion Vented Fugitive Indirect
Emission Source Category

Figure 5-2. Onshore Oil Field: Summary of CO2 Equivalent Emissions and Uncertainties

Examination of the uncertainties, as shown in the bars on Figure 5-2, or the absolute uncertainty values in
Table F-15, shows that “vented emissions of CH4” are the most significant source of uncertainty, contributing
36,400 CO2e tonnes of uncertainty (based on 2600 ±66.5% CO2e tonnes) in a total inventory that had only
37,300 CO2e tonnes of uncertainty (based on 170,000 ±21.9% CO2e tonnes). Carbon dioxide combustion
sources are the next largest contributor of uncertainty, contributing 6,780 CO2e tonnes of uncertainty. The
third largest source is CO2 vented, with 4,420 CO2e tonnes of uncertainty (based on 63,600 ±6.95% CO2e
tonnes).

Therefore, should the company wish to reduce uncertainty in the GHG emission inventory from this onshore
oil field facility, these categories would be the primary targets for uncertainty reduction. These appear to be
the sources where uncertainty reduction efforts could be effectively spent to refine the inventory and reduce
the uncertainty.

Reduction of Uncertainty in Vented CH4 Sources

Within vented emissions, there are 10 sources listed in Table 5-8. The highest ranking source, tank flashing
losses, contributes a significant part of the vented emissions (33% of total vented emissions), as well as the

Pilot Version, September 2009 5-21

vast majority of the uncertainty for the entire vented CH4 category. That single category source’s uncertainty
is 1700 tonnes of methane, or 35,700 tonnes of CO2e. Therefore, improvement of this estimate could greatly
reduce uncertainty in the overall inventory.

Examination of the detailed calculation for that category (presented earlier), shows that the largest uncertainty
is in the emission factor, which is a general industry-wide emission factor. The uncertainty in that factor is
±90.4%. This uncertainty can be reduced simply by using an improved estimation method to determine tank
flashing losses.

As elaborated in the API Compendium, other emission estimation methodologies can be used to estimate
emissions with lower uncertainties. For example, if the “EUB Rule of Thumb” approach (API Compendium
Section 5.4.1) were applied instead of the default tank flashing emission factor for the example onshore
production facility, the emissions for this category would be 20,500 ±49.7% tonnes CO2e.2 This one revision
would change the overall inventory emissions to 151,000 ±9.88% tonnes CO2e.

The company may also decide to take direct measurements of methane emissions from the tanks or other
vented sources with large uncertainties, such as amine unit emissions and dehydration unit emissions.
Repeated direct measurements or a site-specific emission factor would likely reduce uncertainty. In some
cases, the company might be able to produce an improved emission factor with lower uncertainty without
direct emission measurements. For example, dehydrator emission factors were produced by computer
simulations of dehydrators using national average input data; the company might elect to produce simulations
specifically for their dehydrators, using their dehydrator operating conditions, and thus produce a company
specific emission factor with less uncertainty.

Reduction of Uncertainty in Combustion CO2 Sources

If the company decided to take the next step of emission reductions, it may target the next largest category of
uncertainty. Within combustion emissions, there are many individual sources, as were shown in Table 5-8
earlier:

• Boilers/heaters;
• Natural gas engines;
• Diesel engines;
• Flares; and
• Fleet vehicles.

2
This emission estimate is based on an assumed separator pressure of 30 ±5% psi, and an assumed uncertainty of
±50% applied to the correlation constant used in API Compendium Equation 5-20.

Pilot Version, September 2009 5-22

As was shown in Table 5-8, emissions from boilers/heaters, natural gas engines and flares are ranked in the
top ten highest emission sources, with one source, emergency flare, contributing about 62% of the total.
Uncertainty associated with the emergency flare was calculated to be 21.1%, or about ±6,500 tonnes CO2e/yr.

By examining the calculations used for flaring, the following general strategies could be selected by the
operating company to reduce uncertainty in flaring:

• Refine the measurement of total gas flared (the activity factor), to reduce the activity factor
uncertainty from the current value of 15% and/or;
• Refine the gas composition measurements to reduce the uncertainty from 4%.
Refining the measurement of total gas flared may result from many approaches, such as improving the meter
quality (even possibly replacing the meter), improving the quality control of the existing meter (such as
number of calibrations and inspections), or improving the number of measurements taken and recorded from
the meter that is used to calculate the total gas flared (this assumes measurements were not already
continuous). These approaches have varying costs, and some may be cost-prohibitive. The company would
have to determine which approach was the most cost effective.

Refining the gas composition data may also come from several methods, such as taking additional routine
samples, installing a continuous gas analyzer, or using a better analysis method. As with the gas flow rate
measurement, these approaches have varying costs, and some may be cost-prohibitive. The company would
have to determine which approach was the most cost effective.

If the company was effective in reducing the uncertainty in flare gas CO2 emissions, it might then elect to
proceed to the next largest CO2 uncertainty source. The end-user will have to recalculate total emissions for
the company or facility each time, and determine if the uncertainty goal or target has been reached.

Reduction of Uncertainty in Vented CO2 Sources

If the company decided to take the next step of emission reductions, it may target the next largest category of
uncertainty, which is vented CO2 sources. Vented CO2 emissions result from seven sources listed in
Table F-15. The amine unit accounts for the over 98% of the emissions in this category.

Amine unit CO2 vented emissions have an uncertainty of 4,360 tonnes of CO2e in a category that only has an
uncertainty of 4,420 tonnes of CO2e. Therefore any uncertainty reduction efforts in this area should be
directed at the amine unit calculations. However, the calculations used for the amine units reveal that the
uncertainty is actually relatively low. The uncertainty in that emission category is less than 7%. It is possible
to reduce this uncertainty by improving the uncertainties for gas composition, and gas throughput. While it is
possible to reduce this uncertainty, given the low percentage level of the uncertainty, any improvement may
be small. Therefore this is likely a case where company efforts are best spent on the previous categories.

Pilot Version, September 2009 5-23

6.0 REFERENCES
AGA, 2008, Greenhouse Gas Emission Estimation Guidelines for the Natural Gas Distribution Sector;
American Gas Association, Washington DC, April 2008
API 1985, Manual of Petroleum Measurement Standards, Chapter 13, Statistical Aspects of Measuring and
Sampling, Part 1, “Statistical Concepts and Procedures in Measurement”, Reaffirmed February 2006, API
Publications, Washington DC
API, 2004 and 2009, Compendium of GHG Emissions Methodologies for the O&G Industry, American
Petroleum Institute, Washington DC, February 2004 (Addendum 2005). Updated August 2009.
API, 2009, revised and updated version of Reference 2.
API/IPIECA, 2007, Greenhouse Gas Emissions Estimation & Inventories: An International Workshop
Addressing Uncertainty & Accuracy, Workshop Report, London, UK
[Link]
API/IPIECA, 2007, Oil & Natural Gas Industry Guidelines for Greenhouse Gas Reduction Projects,
IPIECA, London, March 2007
API/IPIECA/OGP, 2003, Petroleum Industry Guidelines for Reporting GHG Emissions, IPIECA, London,
December 2003
API Manual of Petroleum Measurement Systems, 2006, Chapter 14, “Natural Gas Fluids Measurement”,
Sections 1 to 10, reaffirmed 2006, API Publications, Washington DC
API Manual of Petroleum Measurement Systems, Chapter 14.1, “Collecting and Handling of Natural Gas
Samples for Custody Transfer”, Sixth Edition, February 2006, API Publications, Washington DC
API Manual of Petroleum Measurement Systems, Chapter 14.10, “Measurement of Flow to Flares”, First
Edition, July 2007, API Publications, Washington DC
ASTM 1945-03, “Analysis of Natural Gas by Gas Chromatography”, current edition approved May 10,
2003. Published July 2003. Originally approved in 1962. Last previous edition approved in 2001 as
D1945–96(2001).
ASTM D 1946 – 90 (Reapproved 2006), Standard Practice for Analysis of Reformed Gas by Gas
Chromatography, current edition approved June 1, 2006. Published June 2006 (Originally approved in
1962)
ASTM UOP, 1997, UOP539-97, Refinery Gas Analysis by Gas Chromatography
ASTM D 2650, 2004, Standard Test Method for Chemical Composition of Gases by Mass Spectrometry,
Published November 1, 2004.
ASTM D 1826 – 94 (Reapproved 2003), “Standard Test Method for Calorific (Heating) Value of Gases in
Natural Gas Range by Continuous Recording Calorimeter”, Current edition approved May 10, 2003.
Published May 2003. Originally approved in 1961. Last previous edition approved in 1998 as D 1826 – 94
(1998).
ASTM D-3588-98, “Standard Practice for Calculating Heat Value, Compressibility Factor, and Relative
Density of Gaseous Fuels”, Current edition approved May 10, 2003. Published May 2003. Originally
approved in 1998. Last previous edition approved in 1998 as D 3588 – 98.
ASTM D 4891 – 89 (Reapproved 2006), Standard Test Method for Heating Value of Gases in Natural Gas
Range by Stoichiometric Combustion, Current edition approved June 1, 2006. Published June 2006.
Originally approved in 1989. Last previous edition approved in 2001 as D4891–89 (2001)
ASTM D7313-08, “Standard Practice for Determination of the Heating Value of Gaseous Fuels using
Calorimetry and On-line/At-line Sampling”, Current edition approved May 1, 2008. Published May 2008.

Pilot Version, September 2009 6-1

Canadian Association of Petroleum Producers (CAPP). A National Inventory of Greenhouse Gas (GHG),
Criteria Air Contaminant (CAC) and Hydrogen Sulphide (H2S) Emissions by the Upstream Oil and Gas
Industry, Technical Report, Volume 3, Methodology for Greenhouse Gases, September 2004.
Canadian Association of Petroleum Producers (CAPP). A National Inventory of Greenhouse Gas (GHG),
Criteria Air Contaminant (CAC) and Hydrogen Sulphide (H2S) Emissions by the Upstream Oil and Gas
Industry, Technical Report, Volume 5, Compendium of Terminology, Information Sources, Emission
Factors, Equipment Schedules and Uncertainty Data, September 2004.
CARB, 2008, Regulation for the Mandatory Reporting of Greenhouse Gas Emissions, California Air
Resources Board (CARB), Final Review Draft, September 18, 2008, Sacramento, California
Casella, G, and R.L. Berger. Statistical Inference, Duxbury Press, Belmont, CA, 1990.
Cochran, William G, 1977, Sampling Techniques, 3rd Edition, John Wiley & Sons, Inc, New York.
Coleman, Hugh W. and W. Glenn Steele, Jr., 1989, Experimentation and Uncertainty Analysis for
Engineers, John Wiley & Sons, Inc, New York.
EMEP/CORINAIR, 2007, Emission Inventory Guidebook, European Environmental Agency
([Link]
ETSG, 2007, Compendium of Notes, Emission Trading Technical Support Group, 30 July 2007
EU-ETS MRG, 2007, “Establishing guidelines for the monitoring and reporting of greenhouse gas
emissions pursuant to Directive 2003/87/EC of the European Parliament and of the Council”, 2007/589/EC,
Brussels, Belgium, July 18, 2007.
Franzblau, A. A Primer of Statistics for Non-Statisticians, Harcourt, Brace & World, Chapter 7, 1958.
INGAA, 2005, Greenhouse Gas Emission Estimation Guidelines for Natural Gas Transmission and
Storage; Interstate Natural Gas Association of America (INGAA), Washington, DC, September 2005
IPCC Emissions Factors Database, [Link]
IPCC, 2000. “IPCC Good Practices Guidance and Uncertainty Management in National Greenhouse Gas
Inventories”, accepted by the IPCC Plenary at its 16th session held in Montreal, 1-8 May, 2000,
Corrigendum, June 15, 2001.
IPCC, 2006, “2006 IPCC Guidelines for National Greenhouse Gas Inventories” [Link]
[Link]/public/2006gl/pdf/1_Volume1/V1_3_Ch3_Uncertainties.pdf
ISO 6974, 2002, Natural gas – Determination of composition with defined uncertainty by gas
chromatography (in 6 parts); First edition 2002-10-15, Geneva, Switzerland
ISO 5168, 2005, Measurement of fluid flow — Procedures for the evaluation of uncertainties, Second
edition, 2005-06-15, Geneva, Switzerland
ISO 5168-2005-06-15, 2005, “International Standard: Measurement of fluid flow – Procedures for the
evaluation of uncertainties”, Geneva, Switzerland.
ISO/IEC 17025:2005, “General requirements for the competence of testing and calibration laboratories”.
Current version effective as of 2005-05-12. Revises: ISO/IEC 17025:1999
Robinson, J.R. On Uncertainty in the Computation of Global Emissions for Biomass Burning. Climatic
Change, 14, pp. 243-262, 1989.
Shires, T.M., and M.R. Harrison. Methane Emissions from the Natural Gas Industry, Volume 6: Vented and
Combustion Source Summary, Final Report, GRI-94/0257.23 and EPA-600/R-96-080f, Gas Research Institute
and US Environmental Protection Agency, June 1996. [Link]

Pilot Version, September 2009 6-2

The Alberta Energy and Utilities Board (EUB/Board), Directive 017, Measurement Requirements for
Upstream Oil and Gas Operations, Revised edition May 7, 2007
U.S. EPA, Technology Transfer Network, Clearinghouse for Inventories & Emissions Factors, Emissions
Factors & AP-42, United States Environmental Protection Agency (U.S. EPA), Research Triangle Park,
North Carolina ([Link]
Williamson, H.J., M.B. Hall, and M.R. Harrison. Methane Emissions from the Natural Gas Industry, Volume
4: Statistical Methodology, Final Report, GRI-94/0257.21 and EPA-600/R-96-080d, Gas Research Institute
and US Environmental Protection Agency, June 1996. [Link]

Pilot Version, September 2009 6-3

 APPENDIX A

GLOSSARY OF STATISTICAL AND GHG INVENTORY TERMS

 Appendix A
GLOSSARY OF STATISTICAL AND GHG INVENTORY TERMS1
TERM STATISTICAL DEFINITION GHG INVENTORY APPLICATION
ACCURACY The ability to indicate values closely Relative measure of the exactness of an
approximating the true value of the emission or removal estimate.
measured variable. Estimates should neither over nor under
estimate true emissions or removals
systematically.
ACTIVITY DATA Magnitude of human activity resulting
in emissions or removals.
For example, the total amounts of fuel
burned for fuel combustion sources, or
the number of piping components for
fugitive emissions.
ARITHMETIC The sum of the values divided by the
MEAN number of values.
BIAS Bias is any influence on a result that It can be introduced by using improper
produces an incorrect approximation of the measurements or by using
true value of the variable being measured. unrepresentative activity data.
Bias is the result of a predictable systematic For example, using only leakage rates
error. from high/medium pressure pipelines
without the low pressure distribution
system.
CENTRAL LIMIT A mathematical/statistical theorem, which Under some conditions, the central
THEOREM states that the arithmetic mean of n limit theorem can justify the
independently distributed, and random approximation that the total emissions
variables, approximates a normal from a bottom-up inventory have a
distribution as n tends to infinity. normal distribution.
COEFFICIENT OF The ratio of the population standard
VARIATION deviation, and the mean.
Sample coefficient of variation is the
ratio of the sample standard deviation
and the sample mean.
CONFIDENCE Confidence interval or range of uncertainty, In practice a confidence interval is
INTERVAL C, is the range or interval within which the defined by a probability value, say
true value is expected to lie with a stated 95%, and confidence limits on either
degree of confidence. side of the mean value.
CONFIDENCE The degree of confidence that may be Level of trust in a measurement or
LEVEL placed on an estimated range of uncertainty. estimate.
CORRELATION Mutual dependence between two quantities.
CORRELATION A number lying between –1 and +1 that
COEFFICIENT measures the mutual dependence between
two variables that are observed together.
It is defined as the covariance of the two
variables divided by the product of their
standard deviations.

Pilot Version, September 2009 A-1

 Appendix A
GLOSSARY OF STATISTICAL AND GHG INVENTORY TERMS, continued
TERM STATISTICAL DEFINITION GHG INVENTORY APPLICATION
COVARIANCE A measure of the mutual dependence
between two variables.
DEGREES OF The number of independent results used in
FREEDOM estimating the standard deviation.
DIRECT A measurement that produces a final
MEASUREMENT result directly from the scale on an
instrument.
DISTRIBUTION A distribution function, or cumulative
FUNCTION distribution function, F(x), for a random
variable X, specifies the probability Pr(X ≤
x) that X is less than or equal to x.
EMISSION A coefficient that relates the activity
FACTOR data to the amount of chemical
compound that is emitted.
Emission factors are often based on a
sample of measurement data, averaged
to develop a representative rate of
emission for a given activity level
under a given set of operating
conditions.
ERROR The difference between true and observed
values.
ESTIMATION The assessment of the value of a quantity or
its uncertainty through the assignment of
numerical observation values in an
estimation formula.
EXPERT A carefully considered, well-
JUDGEMENT documented qualitative or quantitative
judgment made in the absence of
unequivocal observational evidence by
a person or persons who have expertise
in the given field.
INDEPENDENCE Two random variables are independent if
there is a complete absence of association
between how their sample values vary.
Correlation coefficients are commonly used
to document lack of independence between
two random variables.
INDIRECT A measurement that produces a final
MEASUREMENT result by calculation using results from
one or more direct measurements.
KEY SOURCE A key source category is one whose
CATEGORY estimate has a significant influence on
the total direct GHGs in terms of their
absolute level of emissions, their
emission, or both.

Pilot Version, September 2009 A-2

 Appendix A
GLOSSARY OF STATISTICAL AND GHG INVENTORY TERMS, continued
TERM STATISTICAL DEFINITION GHG INVENTORY APPLICATION
LINEAR Linear regression provides a way of fitting a For example, if emissions observations
REGRESSION straight line to a set of observed data points, are plotted against corresponding
taking into account the effects of activity levels, the slope of the line
observational variability. fitted by a linear regression provides an
estimate of the appropriate emission
factor.
LOGNORMAL This is an asymmetric distribution, which
DISTRIBUTION starts from zero, rises to a maximum and
then tails off more slowly to infinity.
MEAN The average of two or more observed
values.
The sample mean, or arithmetic average, is
an estimator for the mean.
MEASUREMENT Procedure for determining a value for a
physical variable.
MEDIAN The median, or population median, is the
50th population percentile.
For symmetric distributions it equals the
mean.
MONTE CARLO The principle of Monte Carlo analysis is to Uncertainties in emission factors and/or
METHOD perform the inventory calculation many activity data are often large and may
times, where each time the uncertain not have normal distributions. In this
emission factors or model parameters and case the conventional statistical rules
activity data are chosen randomly (by the for combining uncertainties become
computer) from within the distribution of very approximate.
uncertainties specified initially by the user. Monte Carlo analysis can deal with this
situation by generating an uncertainty
distribution for the inventory estimate.
NORMAL The normal (or Gaussian) distribution has a
DISTRIBUTION probability density function that can be
defined by two parameters (the mean and
the standard deviation).
OUTLIER A result that differs considerably from
the main body of results in a set.
POPULATION The population is the totality of items under
consideration. In the case of a random
variable, the probability distribution is
considered to define the population of that
variable.
PRECISION The degree to which data within a set Precision is the inverse of uncertainty
cluster together. in the sense that the more precise
something is, the less uncertain it is.

Pilot Version, September 2009 A-3

 Appendix A
GLOSSARY OF STATISTICAL AND GHG INVENTORY TERMS, continued
TERM STATISTICAL DEFINITION GHG INVENTORY APPLICATION
PROBABILITY A probability is a real number in the scale 0
to 1 attached to a random event.
The probability that a random event E
occurs is often denoted as Pr(E).
PROBABILITY A function giving the probability that a
DISTRIBUTION random variable takes any given value or
belongs to a given set of values.
The probability on the whole set of values
of the random variable equals 1.
PROBABILITY This is a mathematical function that In practical situations, the PDF used is
DENSITY characterizes the probability behavior of a chosen from a relatively small number
FUNCTION (PDF) given population. of standard PDFs and the main
It is a function that specifies the relative statistical task is to estimate its
likelihood of a continuous random variable parameters.
of taking any specific value. A knowledge of which PDF has been
Most PDFs require one or more parameters used is a necessary item in the
to specify them fully. documentation of an uncertainty
assessment.
PROPAGATION OF A set of rules that specify how to See Section 4.0
UNCERTAINTIES algebraically combine the quantitative
measures of uncertainty associated with the
input values to the mathematical formulae
used in inventory compilation, so as to
obtain corresponding measures of
uncertainty for the output values.
QUALITY A planned system of review procedures
ASSURANCE (QA) conducted by personnel not directly
involved in the inventory
compilation/development process to
verify that data quality objectives were
met and ensure that the inventory
represents the best possible estimate of
emissions and sinks given the current
state of scientific knowledge and data
available.
QUALITY A system of routine technical activities,
CONTROL (QC) to measure and control the quality of
the inventory as it is being developed.
These may include data accuracy
checks, and the use of applicable
procedures for emission calculations
and reporting.

Pilot Version, September 2009 A-4

 Appendix A
GLOSSARY OF STATISTICAL AND GHG INVENTORY TERMS, continued
TERM STATISTICAL DEFINITION GHG INVENTORY APPLICATION
RANDOM ERROR The difference between an individual An error that varies in an unpredictable
measurement and the limiting value of the manner when a large number of
sample mean. measurements of the same variable are
made under effectively identical
conditions.
RANDOM SAMPLE A sample of n items chosen from a
population such that every possible sample
has the same probability of being chosen.
RANDOM A variable that may take any of the values A ‘discrete’ random variable is one that
VARIABLE of a specified set of values that are may take only isolated values.
represented by a probability distribution. A ‘continuous’ random variable is one
that may take any value within a finite
or infinite interval.
RANGE Region between the limits within which a
quantity is measured.
REPEATABILITY A measure of the agreement between the
results of successive measurements of the
same variable carried out by the same
method, with the same instrument, at the
same location, and within a short period of
time.
REPRODUCIBILITY A measure of the agreement between the
results of measurements of the same
variable where individual measurements are
carried out by the same methods, with the
same type of instruments, but by different
observers at different locations, and after a
long period of time.
RESIDUAL The difference between the observed value It is the component of an observation
and the value predicted by a statistical that cannot be explained by the model.
model, e.g. by linear regression.
SAMPLE A sample is a finite set of observations
drawn from a population.
SENSITIVITY This is a study to determine how sensitive This analysis can include observing the
ANALYSIS (or stable) is a calculation method to range of output values that correspond
variations of its input data or underlying to input variables; and calculating finite
assumptions. difference approximations for studying
error propagation within a system.
STANDARD The population standard deviation is the
DEVIATION positive square root of the variance. It is
estimated by the sample standard deviation
that is the positive square root of the sample
variance.
STANDARD ERROR A term often used to signify the sample
OF THE MEAN standard deviation of the mean.

Pilot Version, September 2009 A-5

 Appendix A
GLOSSARY OF STATISTICAL AND GHG INVENTORY TERMS, continued
TERM STATISTICAL DEFINITION GHG INVENTORY APPLICATION
SYSTEMATIC The difference between the true, but usually It is an error that in the course of a
ERROR unknown, value of a quantity being number of measurements made under
measured, and the mean observed value as the same conditions, on material
would be estimated by the sample mean of having the same true value of a
an infinite set of observations. variable, either remains constant in
Systematic errors result in a bias. absolute value and sign or varies in a
predictable manner.
TRUE VALUE The correct value of a variable.
TIME SERIES A series of values that are affected by
random processes and are observed at
successive time points.
UNCERTAINTY A parameter associated with the result of A general term that refers to the lack of
measurement that characterizes the certainty (in inventory components)
dispersion of the values that could be resulting from causal factors such as
reasonably attributed to the measured unidentified sources and sinks, lack of
quantity transparency etc.
UNCERTAINTY A quantitative measure of the uncertainty of
ANALYSIS output values caused by uncertainties in its
input values, and to examine the relative
importance of these factors.
VALIDATION Involves checking to ensure that the
inventory has been compiled correctly
in line with reporting instructions and
guidelines.
VARIABILITY The observed differences attributable to Variability derives from processes that
true heterogeneity or diversity in a are either inherently random or whose
population. nature and effects are influential but
It can be characterized by quantities such as unknown.
the sample variance. Variability is not usually reducible by
further measurement or study.
VARIANCE The measure of the dispersion or scatter of
the values of the random variable about the
mean.
VARIANCE OF The mean of a sample taken from a
SAMPLE MEAN population is itself a random variable with
its own characteristic behavior and its own
variance.
VERIFICATION The collection of activities and
procedures to be followed during the
planning and development, or after
completion of an inventory, to help
establish its reliability for intended use.
1
Extracted from: Chapter 13.1 of API MPMS: Statistical Concepts and Procedures in Measurement; and Annex 3 of the IPCC Good Practice Guidance
and Uncertainty Management in National Greenhouse Gas Inventories, 2000.

Pilot Version, September 2009 A-6

 APPENDIX B

LIST OF INDUSTRY MEASUREMENT STANDARDS

 Appendix B

LIST OF INDUSTRY MEASUREMENT STANDARDS

This Appendix provides a list of over 300 hydrocarbon measurement standards used to measure quantities of
petroleum products and natural gas liquids. Entire series of standards from particular standards developers are
listed as many of the standards are inter-related and have to be used in conjunction with one another for
measurement of quantities. This list is provided for users’ convenience to highlight the fact that companies may
use different sets of measurement standards that fit their operating environment and available instrumentation.

Each of the standards cited is designed to meet a variety of data needs and a defined level of accuracy, precision
and uncertainty ranges. The compilation provided here includes standards developed by the American
Petroleum institute (API); the American Society of Testing and Materials (ASTM)’ the American Gas
Association (AGA); the Gas Processors Association (GPA), and the international organization for
standardization (ISO). The list does not comprise an exhaustive reference to all standards; it does not cite
applicable standards from the EU, Japan, or other standards setting organizations.

B.1 API Manual of Petroleum Measurement Standards (MPMS)

Chapter 1 Vocabulary
Chapter 2.2A Measurement and Calibration of Upright Cylindrical Tanks by the Manual Strapping
Method
[Chapter 2.2A should be used in conjunction with Chapter 2.2B. These two standards
combined supersede the previous API Standard 2550, Measurement and Calibration of
Upright Cylindrical Tanks]
Chapter 2.2B Calibration of Upright Cylindrical Tanks Using the Optical Reference Line Method
Chapter 2.2C Calibration of Upright Cylindrical Tanks Using the Optical-Triangulation Method
Chapter 2.2D Calibration of Upright Cylindrical Tanks Using the Internal Electro-optical Distance
Ranging Method
Chapter 2.2E Petroleum and Liquid Petroleum Products—Calibration of Horizontal Cylindrical Tanks—
Part 1: Manual Methods
Chapter 2.2F Petroleum and Liquid Petroleum Products—Calibration of Horizontal Cylindrical Tanks—
Part 2: Internal Electro-optical Distance-ranging Method
Std 2552 Measurement and Calibration of Spheres and Spheroids
Std 2554 Measurement and Calibration of Tank Cars
Std 2555 Liquid Calibration of Tanks
RP 2556 Correcting Gauge Tables for Incrustation
Chapter 2.7 Calibration of Barge Tanks
Chapter 2.8A Calibration of Tanks on Ships and Oceangoing Barges

Pilot Version, September 2009 B-1

Chapter 2.8B Establishment of the Location of the Reference Gauge Point and the Gauge Height of
Tanks on Marine Tank Vessels
Chapter 3.1A Standard Practice for the Manual Gauging of Petroleum and Petroleum Products
Chapter 3.1B Standard Practice for Level Measurement of Liquid Hydrocarbons in Stationary Tanks by
Automatic Tank Gauging
Chapter 3.2 Standard Practice for Gauging Petroleum and Petroleum Products in Tank Cars
Chapter 3.3 Standard Practice for Level Measurement of Liquid Hydrocarbons in Stationary
Pressurized Storage Tanks by Automatic Tank Gauging
Chapter 3.4 Standard Practice for Level Measurement of Liquid Hydrocarbons on Marine Vessels by
Automatic Tank Gauging
Chapter 3.5 Standard Practice for Level Measurement of Light Hydrocarbon Liquids Onboard Marine
Vessels by Automatic Tank Gauging
Chapter 3.6 Measurement of Liquid Hydrocarbons by Hybrid Tank Measurement Systems
Chapter 4.1 Proving Systems – Introduction
Chapter 4.2 Displacement Provers
Chapter 4.4 Tank Provers
Chapter 4.5 Master-Meter Provers
Chapter 4.6 Pulse Interpolation
Chapter 4.7 Field-Standard Test Measures
Chapter 4.8 Operation of Proving Systems
Chapter 4.9.1 Methods of Calibration for Displacement and Volumetric Tank Provers, Part 1—
Introduction to the Determination of the Volume of Displacement and Tank Provers
Chapter 4.9.2 Methods of Calibration for Displacement and Volumetric Tank Provers, Part 2—
Determination of the Volume of Displacement and Tank
Chapter 5.1 General Consideration for Measurement by Meters
Chapter 5.2 Measurement of Liquid Hydrocarbons by Displacement Meters
Chapter 5.3 Measurement of Liquid Hydrocarbons by Turbine Meters
Chapter 5.4 Accessory Equipment for Liquid Meters
Chapter 5.5 Fidelity and Security of Flow Measurement Pulsed-Data Transmission Systems
Chapter 5.6 Measurement of Liquid Hydrocarbons by Coriolis Meters
Chapter 5.8 Measurement of Liquid Hydrocarbons by Ultrasonic Flowmeters Using Transit Time
Technology
Draft Standard Vortex Shedding Flowmeter for Measurement of Hydrocarbon Fluids
Chapter 6.1 Lease Automatic Custody Transfer (LACT) Systems
Chapter 6.2 Loading Rack Metering Systems
Chapter 6.4 Metering Systems for Aviation Fueling Facilities
Chapter 6.5 Metering Systems for Loading and Unloading Marine Bulk Carriers
Chapter 6.6 Pipeline Metering Systems
Chapter 6.7 Metering Viscous Hydrocarbons

Pilot Version, September 2009 B-2

Chapter 7 Temperature Determination
Chapter 8.1 Manual Sampling of Petroleum and Petroleum Products
Chapter 8.2 Automatic Sampling of Petroleum and Petroleum Products
Chapter 8.3 Mixing and Handling of Liquid Samples of Petroleum and Petroleum Products
Chapter 8.4 Standard Practice for Sampling and Handling of Fuels for Volatility Measurement
Chapter 9.1 Standard Test Method for Density, Relative Density (Specific Gravity), or API Gravity of
Crude Petroleum and Liquid Petroleum Products by Hydrometer Method
Chapter 9.2 Standard Test Method for Density or Relative Density of Light Hydrocarbons by Pressure
Hydrometer
Chapter 9.3 Standard Test Method for Density, Relative Density, and API Gravity of Crude Petroleum
and Liquid Petroleum Products by Thermohydrometer Method
Chapter 10.1 Standard Test Method for Sediment in Crude Oils and Fuel Oils by the Extraction Method
Chapter 10.2 Determination of Water in Crude Oil by Distillation
Chapter 10.3 Standard Test Method for Water and Sediment in Crude Oil by the Centrifuge Method
(Laboratory Procedure)
Chapter 10.4 Determination of Sediment and Water in Crude Oil by the Centrifuge Method (Field
Procedure)
Chapter 10.5 Standard Test Method for Water in Petroleum Products and Bituminous Materials by
Distillation
Chapter 10.6 Standard Test Method for Water and Sediment in Fuel Oils by the Centrifuge Method
(Laboratory Procedure)
Chapter 10.7 Standard Test Method for Water in Crude Oils by Potentiometric Karl Fischer Titration
Chapter 10.8 Standard Test Method for Sediment in Crude Oil by Membrane Filtration
Chapter 10.9 Standard Test Method for Water in Crude Oils by Coulometric Karl Fischer Titration
Chapter 11.1 Temperature and Pressure Volume Correction Factors for Generalized Crude Oils, Refined
Products, and Lubricating Oils
Chapter 11.2.2 Compressibility Factors for Hydrocarbons: 0.350 – 0.637 Relative Density (60 °F/60 °F)
and –50 °F to 140 °F Metering Temperature
Chapter 11.2.2M Compressibility Factors for Hydrocarbons: 350 – 637 Kilograms per Cubic Meter Density
(15 °C) and –46 °C to 60 °C Metering Temperature
Chapter 11.2.4 Temperature Correction for the Volume of NGL and LPG Tables 23E, 24E, 53E, 54E,
59E, 60E
Chapter 11.2.5 A Simplified Vapor Pressure Correlation for Commercial NGLs
Chapter [Link] Ethylene Density
Chapter [Link] Propylene Compressibility
Chapter 11.4.1 Properties of Reference Materials, Part 1—Density of Water and Water Volume
Correction Factors for Calibration of Volumetric Provers
Chapter 11.5.1 Density/Weight/Volume Intraconversion, Part 1—Conversions of API Gravity at 60 °F
Chapter 11.5.2 Density/Weight/Volume Intraconversion, Part 2—Conversions for Relative Density
(60/60 °F)

Pilot Version, September 2009 B-3

Chapter 11.5.3 Density/Weight/Volume Intraconversion, Part 3—Conversions for Absolute Density at
15 °C
Chapter 12.1.1 Calculation of Static Petroleum Quantities, Part 1—Upright Cylindrical Tanks and Marine
Vessels
Chapter 12.1.2 Calculation of Static Petroleum Quantities, Part 2—Calculation Procedures for Tank Cars
Chapter 12.2 Calculation of Liquid Petroleum Quantities Measured by Turbine or Displacement Meters
Chapter 12.2.1 Calculation of Petroleum Quantities Using Dynamic Measurement Methods and Volume
Correction Factors, Part 1—Introduction
Chapter 12.2.2 Calculation of Petroleum Quantities Using Dynamic Measurement Methods and
Volumetric Correction Factors, Part 2—Measurement Tickets
Chapter 12.2.3 Calculation of Petroleum Quantities Using Dynamic Measurement Methods and
Volumetric Correction Factors, Part 3—Proving Reports
Chapter 12.2.4 Calculation of Petroleum Quantities Using Dynamic Measurement Methods and
Volumetric Correction Factors, Part 4—Calculation of Base Prover Volumes by
Waterdraw Method
Chapter 12.2.5 Calculation of Petroleum Quantities Using Dynamic Measurement Methods and
Volumetric Correction Factors, Part 5—Base Prover Volume Using Master Meter Method
Chapter 12.3 Calculation of Volumetric Shrinkage from Blending Light Hydrocarbons with Crude Oil
Chapter 13.1 Statistical Concepts and Procedures in Measurement
Chapter 13.2 Statistical Methods of Evaluating Meter Proving Data
Chapter 14.1 Collecting and Handling of Natural Gas Samples for Custody Transfer
Chapter 14.2 Compressibility Factors of Natural Gas and Other Related Hydrocarbon Gases
Chapter 14.3.1 Concentric, Square-edged Orifice Meters, Part 1—General Equations and Uncertainty
Guidelines
Chapter 14.3.2 Concentric, Square-Edged Orifice Meters, Part 2—Specification and Installation
Requirements
Chapter 14.3.3 Concentric, Square-Edged Orifice Meters, Part 3—Natural Gas Applications
Chapter 14.3.4 Concentric, Square-Edged Orifice Meters, Part 4—Background, Development,
Implementation Procedures and Subroutine Documentation
Chapter 14.4 Converting Mass of Natural Gas Liquids and Vapors to Equivalent Liquid Volumes
Chapter 14.5 Calculation of Gross Heating Value, Specific Gravity, and Compressibility of Natural Gas
Mixtures from Compositional Analysis
Chapter 14.6 Continuous Density Measurement
Chapter 14.7 Mass Measurement of Natural Gas Liquids
Chapter 14.8 Liquefied Petroleum Gas Measurement
Chapter 14.9 Measurement of Natural Gas by Coriolis Meter
Chapter 14.10 Measurement of Flow to Flares
Chapter 15 Guidelines for Use of the International System of Units (SI) in the Petroleum and Allied
Industries
Chapter 16.2 Mass Measurement of Liquid Hydrocarbons in Vertical Cylindrical Storage Tanks by
Hydrostatic Tank Gauging

Pilot Version, September 2009 B-4

Chapter 17.1 Guidelines for Marine Cargo Inspection
Chapter 17.2 Measurement of Cargoes on Board Tank Vessels
Chapter 17.3 Guidelines for Identification of the Source of Free Waters Associated With Marine
Petroleum Cargo Movements
Chapter 17.4 Method for Quantification of Small Volumes on Marine Vessels (OBQ/ROB)
Chapter 17.5 Guidelines for Cargo Analysis and Reconciliation
Chapter 17.6 Guidelines for Determining Fullness of Pipelines Between Vessels and Shore Tanks
Chapter 17.7 Recommended Practices for Developing Barge Control Factors (Volume Ratio)
Chapter 17.8 Guidelines for Pre-Loading Inspection of Marine Vessel Cargo Tanks
Chapter 17.9 Vessel Experience Factor (VEF)
Chapter 17.10.2 Measurement of Refrigerated and/or Pressurized Cargoes on Board Marine Gas Carriers,
Part 2—Liquefied Petroleum and Chemical Gases
Chapter 17.11 Measurement and Sampling of Cargoes on Board Tank Vessels Using Closed and
Restricted Equipment
Chapter 17.12 Procedure for Bulk Liquid Chemical Cargo Inspection by Cargo Inspectors
Chapter 18.1 Measurement Procedures for Crude Oil Gathered From Small Tanks by Truck
Publ 2514A Atmospheric Hydrocarbon Emissions from Marine Vessel Transfer Operations
Publ 2524 Impact Assessment of New Data on the Validity of American Petroleum Institute Marine
Transfer Operation Emission Factors
Publ 2558 Wind Tunnel Testing of External Floating-Roof Storage Tanks
TR 2567 Evaporative Loss from Storage Tank Floating Roof Landings
TR 2568 Evaporative Loss from the Cleaning of Storage Tanks
TR 2569 Evaporative Loss from Closed-vent Internal Floating-roof Storage Tanks
Chapter 19.1 Evaporative Loss from Fixed-roof Tanks
Chapter 19.1A Evaporation Loss from Low-pressure Tanks
Chapter 19.1D Documentation File for API Manual of Petroleum Measurement Standards Chapter 19.1
“Evaporative Loss from Fixed-roof Tanks”
Chapter 19.2 Evaporative Loss from Floating-roof Tanks
Chapter 19.3, Part A Wind Tunnel Test Method for the Measurement of Deck-fitting Loss Factors for External
Floating-roof Tanks
Chapter 19.3, Part B Air Concentration Test Method—Rim-seal Loss Factors for Floating-roof Tanks
Chapter 19.3, Part C Weight Loss Test Method for the Measurement of Rim-seal Loss Factors for Internal
Floating-roof Tanks
Chapter 19.3, Part D Fugitive Emission Test Method for the Measurement of Deck-seam Loss Factors for
Internal Floating-roof Tanks
Chapter 19.3, Part E Weight Loss Test Method for the Measurement of Deck-fitting Loss Factors for Internal
Floating-roof Tanks
Chapter 19.3, Part F Evaporative Loss Factor for Storage Tanks Certification Program
Chapter 19.3, Part G Certified Loss Factor Testing Laboratory Registration
Chapter 19.3, Part H Tank Seals and Fittings Certification—Administration

Pilot Version, September 2009 B-5

Chapter 19.4 Recommended Practice for Speciation of Evaporative Losses
Chapter 20.1 Allocation Measurement
RP 85 Use of Subsea Wet-gas Flowmeters in Allocation Measurement Systems
RP 86 Recommended Practice for Measurement of Multiphase Flow
RP 87 Recommended Practice for Field Analysis of Crude Oil Samples Containing from Two to
Fifty Percent Water by Volume
Chapter 21.1 Electronic Gas Measurement
Chapter 21.2 Flow Measurement—Electronic Liquid Measurement
Chapter 21.2-A1 Addendum 1 to Flow Measurement Using Electronic Metering Systems
Chapter 22.1 Testing Protocols—General Guidelines for Developing Testing Protocols for Devices
Used in the Measurement of Hydrocarbon Fluids
Chapter 22.2 Testing Protocols—Differential Pressure Flow Measurement Devices
Std 2560 Reconciliation of Liquid Pipeline Quantities
Publ 2566 State of the Art Multiphase Flow Metering

B.2 International Organization for Standardization (ISO)

ISO 91-1:1992 Petroleum measurement tables — Part 1: Tables based on reference temperatures of 15
degrees C and 60 degrees F
ISO 91-2:1991 Petroleum measurement tables — Part 2: Tables based on a reference temperature of 20
degrees C
ISO 2714:1980 Liquid hydrocarbons — Volumetric measurement by displacement meter systems other
than dispensing pumps
ISO 2715:1981 Liquid hydrocarbons — Volumetric measurement by turbine meter systems
ISO 3170:2004 Petroleum liquids — Manual sampling
ISO 3171:1988 Petroleum liquids — Automatic pipeline sampling
ISO 3675:1998 Crude petroleum and liquid petroleum products — Laboratory determination of density —
Hydrometer method
ISO 3733:1999 Petroleum products and bituminous materials — Determination of water — Distillation
method
ISO 3734:1997 Petroleum products — Determination of water and sediment in residual fuel oils —
Centrifuge method
ISO 3735:1999 Crude petroleum and fuel oils — Determination of sediment — Extraction method
ISO 3838:2004 Crude petroleum and liquid or solid petroleum products — Determination of density or
relative density — Capillary-stoppered pyknometer and graduated bicapillary pyknometer
methods
ISO 3993:1984 Liquefied petroleum gas and light hydrocarbons — Determination of density or relative
density — Pressure hydrometer method
ISO 4124:1994 Liquid hydrocarbons — Dynamic measurement — Statistical control of volumetric
metering systems
ISO 4257:2001 Liquefied petroleum gases — Method of sampling

Pilot Version, September 2009 B-6

ISO 4266-1:2002 Petroleum and liquid petroleum products — Measurement of level and temperature in
storage tanks by automatic methods — Part 1: Measurement of level in atmospheric tanks
ISO 4266-2:2002 Petroleum and liquid petroleum products — Measurement of level and temperature in
storage tanks by automatic methods — Part 2: Measurement of level in marine vessels
ISO 4266-3:2002 Petroleum and liquid petroleum products — Measurement of level and temperature in
storage tanks by automatic methods — Part 3: Measurement of level in pressurized storage
tanks (non-refrigerated)
ISO 4266-5:2002 Petroleum and liquid petroleum products — Measurement of level and temperature in
storage tanks by automatic methods — Part 5: Measurement of temperature in marine
vessels
ISO 4266-6:2002 Petroleum and liquid petroleum products — Measurement of level and temperature in
storage tanks by automatic methods — Part 6: Measurement of temperature in pressurized
storage tanks (non-refrigerated)
ISO 4267-2:1988 Petroleum and liquid petroleum products — Calculation of oil quantities — Part 2:
Dynamic measurement
ISO 4268:2000 Petroleum and liquid petroleum products — Temperature measurements — Manual
methods
ISO 4269:2001 Petroleum and liquid petroleum products — Tank calibration by liquid measurement —
Incremental method using volumetric meters
ISO 4512:2000 Petroleum and liquid petroleum products — Equipment for measurement of liquid levels in
storage tanks — Manual methods
ISO 5024:1999 Petroleum liquids and liquefied petroleum gases — Measurement — Standard reference
conditions
ISO 6296:2000 Petroleum products — Determination of water — Potentiometric Karl Fischer titration
method
ISO 6551:1982 Petroleum liquids and gases — Fidelity and security of dynamic measurement — Cabled
transmission of electric and/or electronic pulsed data
ISO 7278-1:1987 Liquid hydrocarbons — Dynamic measurement — Proving systems for volumetric meters
— Part 1: General principles
ISO 7278-2:1988 Liquid hydrocarbons — Dynamic measurement — Proving systems for volumetric meters
— Part 2: Pipe provers
ISO 7278-3:1998 Liquid hydrocarbons — Dynamic measurement — Proving systems for volumetric meters
— Part 3: Pulse interpolation techniques
ISO 7278-4:1999 Liquid hydrocarbons — Dynamic measurement — Proving systems for volumetric meters
— Part 4: Guide for operators of pipe provers
ISO 7507-1:2003 Petroleum and liquid petroleum products — Calibration of vertical cylindrical tanks —
Part 1: Strapping method
ISO 7507-2:2005 Petroleum and liquid petroleum products — Calibration of vertical cylindrical tanks —
Part 2: Optical-reference-line method
ISO 7507-3:2006 Petroleum and liquid petroleum products — Calibration of vertical cylindrical tanks —
Part 3: Optical-triangulation method
ISO 7507-4:1995 Petroleum and liquid petroleum products — Calibration of vertical cylindrical tanks —
Part 4: Internal electro-optical distance-ranging method
ISO 7507-5:2000 Petroleum and liquid petroleum products — Calibration of vertical cylindrical tanks —
Part 5: External electro-optical distance-ranging method

Pilot Version, September 2009 B-7

ISO 8222:2002 Petroleum measurement systems — Calibration — Temperature corrections for use when
calibrating volumetric proving tanks
ISO 8697:1999 Crude petroleum and petroleum products — Transfer accountability — Assessment of on
board quantity (OBQ) and quantity remaining on board (ROB)
ISO 9029:1990 Crude petroleum — Determination of water — Distillation method
ISO 9030:1990 Crude petroleum — Determination of water and sediment — Centrifuge method
ISO 9114:1997 Crude petroleum — Determination of water content by hydride reaction — Field method
ISO 9200:1993 Crude petroleum and liquid petroleum products — Volumetric metering of viscous
hydrocarbons
ISO 9403:2000 Crude petroleum — Transfer accountability — Guidelines for cargo inspection
ISO 9770:1989 Crude petroleum and petroleum products — Compressibility factors for hydrocarbons in
the range 638 kg/m3 to 1 074 kg/m3
ISO 10336:1997 Crude petroleum — Determination of water — Potentiometric Karl Fischer titration
method
ISO 10337:1997 Crude petroleum — Determination of water — Coulometric Karl Fischer titration method
ISO 11223:2004 Petroleum and liquid petroleum products — Direct static measurements — Measurement
of content of vertical storage tanks by hydrostatic tank gauging
ISO 11563:2003 Crude petroleum and petroleum products — Bulk cargo transfer — Guidelines for
achieving the fullness of pipelines
ISO 12185:1996 Crude petroleum and petroleum products — Determination of density — Oscillating U-
tube method
ISO 12917-1:2002 Petroleum and liquid petroleum products — Calibration of horizontal cylindrical tanks —
Part 1: Manual methods
ISO 12917-2:2002 Petroleum and liquid petroleum products — Calibration of horizontal cylindrical tanks —
Part 2: Internal electro-optical distance-ranging method
ISO 12937:2000 Petroleum products — Determination of water — Coulometric Karl Fischer titration
method
ISO/TR 13739:1998 Petroleum products — Methods for specifying practical procedures for the transfer of
bunker fuels to ships
ISO 13740:1998 Crude petroleum and petroleum products — Transfer accountability — Assessment of
vessel experience factor on loading (VEFL) and vessel experience factor on discharging
(VEFD) of ocean-going tanker vessels
ISO 15169:2003 Petroleum and liquid petroleum products — Determination of volume, density and mass of
the hydrocarbon content of vertical cylindrical tanks by hybrid tank measurement systems
ISO 6578:1991 Refrigerated hydrocarbon liquids — Static measurement — Calculation procedure
ISO 8310:1991 Refrigerated light hydrocarbon fluids — Measurement of temperature in tanks containing
liquefied gases — Resistance thermometers and thermocouples
ISO 8311:1989 Refrigerated light hydrocarbon fluids — Calibration of membrane tanks and independent
prismatic tanks in ships — Physical measurement
ISO 8943:2007 Refrigerated light hydrocarbon fluids — Sampling of liquefied natural gas — Continuous
and intermittent methods
ISO 9091-1:1991 Refrigerated light-hydrocarbon fluids — Calibration of spherical tanks in ships — Part 1:
Stereo-photogrammetry

Pilot Version, September 2009 B-8

ISO 9091-2:1992 Refrigerated light hydrocarbon fluids — Calibration of spherical tanks in ships — Part 2:
Triangulation measurement
ISO 13398:1997 Refrigerated light hydrocarbon fluids — Liquefied natural gas — Procedure for custody
transfer on board ship
ISO 18132-1:2006 Refrigerated light hydrocarbon fluids — General requirements for automatic level gauges
— Part 1: Gauges onboard ships carrying liquefied gases
ISO 18132-2:2008 Refrigerated light hydrocarbon fluids — General requirements for automatic level gauges
— Part 2: Gauges in refrigerated-type shore tanks
ISO 4006:1991 Measurement of fluid flow in closed conduits — Vocabulary and symbols
ISO 4185:1980 Measurement of liquid flow in closed conduits — Weighing method
ISO 5168:2005 Measurement of fluid flow — Procedures for the evaluation of uncertainties
ISO/TR 7066-1:1997 Assessment of uncertainty in calibration and use of flow measurement devices — Part 1:
Linear calibration relationships
ISO 7066-2:1988 Assessment of uncertainty in the calibration and use of flow measurement devices — Part
2: Non-linear calibration relationships
ISO 8316:1987 Measurement of liquid flow in closed conduits — Method by collection of the liquid in a
volumetric tank
ISO 9368-1:1990 Measurement of liquid flow in closed conduits by the weighing method — Procedures for
checking installations — Part 1: Static weighing systems
ISO 9951:1993 Measurement of gas flow in closed conduits — Turbine meters
ISO 11631:1998 Measurement of fluid flow — Methods of specifying flowmeter performance
ISO 2186:2007 Fluid flow in closed conduits — Connections for pressure signal transmissions between
primary and secondary elements
ISO/TR 3313:1998 Measurement of fluid flow in closed conduits — Guidelines on the effects of flow
pulsations on flow-measurement instruments
ISO 5167-1:2003 Measurement of fluid flow by means of pressure differential devices inserted in circular
cross-section conduits running full — Part 1: General principles and requirements
ISO 5167-2:2003 Measurement of fluid flow by means of pressure differential devices inserted in circular
cross-section conduits running full — Part 2: Orifice plates
ISO 5167-3:2003 Measurement of fluid flow by means of pressure differential devices inserted in circular
cross-section conduits running full — Part 3: Nozzles and Venturi nozzles
ISO 5167-4:2003 Measurement of fluid flow by means of pressure differential devices inserted in circular
cross-section conduits running full — Part 4: Venturi tubes
ISO 9300:2005 Measurement of gas flow by means of critical flow Venturi nozzles
ISO/TR 9464:2008 Guidelines for the use of ISO 5167:2003
ISO/TR 12767:2007 Measurement of fluid flow by means of pressure differential devices — Guidelines on the
effect of departure from the specifications and operating conditions given in ISO 5167
ISO/TR 15377:2007 Measurement of fluid flow by means of pressure-differential devices — Guidelines for the
specification of orifice plates, nozzles and Venturi tubes beyond the scope of ISO 5167
ISO 3966:2008 Measurement of fluid flow in closed conduits — Velocity area method using Pitot static
tubes
ISO 6817:1992 Measurement of conductive liquid flow in closed conduits — Method using
electromagnetic flowmeters

Pilot Version, September 2009 B-9

ISO 7194:2008 Measurement of fluid flow in closed conduits — Velocity-area methods of flow
measurement in swirling or asymmetric flow conditions in circular ducts by means of
current-meters or Pitot static tubes
ISO 9104:1991 Measurement of fluid flow in closed conduits — Methods of evaluating the performance
of electromagnetic flow-meters for liquids
ISO 10790:1999 Measurement of fluid flow in closed conduits — Guidance to the selection, installation and
use of Coriolis meters (mass flow, density and volume flow measurements)
ISO/TR 12764:1997 Measurement of fluid flow in closed conduits — Flowrate measurement by means of
vortex shedding flowmeters inserted in circular cross-section conduits running full
ISO 13359:1998 Measurement of conductive liquid flow in closed conduits — Flanged electromagnetic
flowmeters — Overall length
ISO 14511:2001 Measurement of fluid flow in closed conduits — Thermal mass flowmeters
ISO 13443:1996 Natural gas — Standard reference conditions
ISO 14532:2001 Natural gas — Vocabulary
ISO 15112:2007 Natural gas — Energy determination
ISO 15970:2008 Natural gas — Measurement of properties — Volumetric properties: density, pressure,
temperature and compression factor
ISO 15971:2008 Natural gas — Measurement of properties — Calorific value and Wobbe index
ISO 6976:1995 Natural gas — Calculation of calorific values, density, relative density and Wobbe index
from composition
ISO 10101-1:1993 Natural gas — Determination of water by the Karl Fischer method — Part 1: Introduction
ISO 10101-2:1993 Natural gas — Determination of water by the Karl Fischer method — Part 2: Titration
procedure
ISO 10101-3:1993 Natural gas — Determination of water by the Karl Fischer method — Part 3: Coulometric
procedure
ISO 10715:1997 Natural gas — Sampling guidelines
ISO 11541:1997 Natural gas — Determination of water content at high pressure
ISO 12213-1:2006 Natural gas — Calculation of compression factor — Part 1: Introduction and guidelines
ISO 12213-2:2006 Natural gas — Calculation of compression factor — Part 2: Calculation using molar-
composition analysis
ISO 12213-3:2006 Natural gas — Calculation of compression factor — Part 3: Calculation using physical
properties
ISO 20765-1:2005 Natural gas — Calculation of thermodynamic properties — Part 1: Gas phase properties
for transmission and distribution applications
ISO/TR 26762:2008 Natural gas — Upstream area — Allocation of gas and condensate

B.3 ASTM International

ASTM D4057 Standard Practice for Manual Sampling of Petroleum and Petroleum Products
ASTM D4177 Standard Practice for Automatic Sampling of Petroleum and Petroleum Products
ASTM D1298 Standard Test Method for Density, Relative Density (Specific Gravity), or API Gravity of
Crude Petroleum and Liquid Petroleum Products by Hydrometer Method

Pilot Version, September 2009 B-10

ASTM D1657 Standard Test Method for Density or Relative Density of Light Hydrocarbons by Pressure
Hydrometer
ASTM D4052 Standard Test Method for Density and Relative Density of Liquids by Digital Density
Meter
ASTM D4002 Standard Test Method for Density and Relative Density of Crude Oils by Digital Density
Analyzer
ASTM D473 Standard Test Method for Sediment in Crude Oils and Fuel Oils by the Extraction Method
ASTM D4006 Standard Test Method for Water in Crude Oil by Distillation
ASTM D4007 Standard Test Method for Water and Sediment in Crude Oil by the Centrifuge Method
(Laboratory Procedure)
ASTM D95 Standard Test Method for Water in Petroleum Products and Bituminous Materials by
Distillation
ASTM D1796 Standard Test Method for Water and Sediment in Fuel Oils by the Centrifuge Method
(Laboratory Procedure)
ASTM D4377 Standard Test Method for Water in Crude Oils by Potentiometric Karl Fischer Titration
ASTM D4807 Standard Test Method for Sediment in Crude Oil by Membrane Filtration
ASTM D4928 Standard Test Methods for Water in Crude Oils by Coulometric Karl Fischer Titration
ASTM D6304 Standard Test Method for Determination of Water in Petroleum Products, Lubricating
Oils, and Additives by Coulometric Karl Fischer Titration
ASTM D71 Standard Test Method for Relative Density of Solid Pitch and Asphalt (Displacement
Method)
ASTM D287 Standard Test Method for API Gravity of Crude Petroleum and Petroleum Products
(Hydrometer Method)
ASTM D1070 Standard Test Methods for Relative Density of Gaseous Fuels
ASTM D1217 Standard Test Method for Density and Relative Density (Specific Gravity) of Liquids by
Bingham Pycnometer
ASTM D1480 Standard Test Method for Density and Relative Density (Specific Gravity) of Viscous
Materials by Bingham Pycnometer
ASTM D1481 Standard Test Method for Density and Relative Density (Specific Gravity) of Viscous
Materials by Lipkin Bicapillary Pycnometer
ASTM D1550 Standard ASTM Butadiene Measurement Tables
ASTM D1555 Standard Test Method for Calculation of Volume and Weight of Industrial Aromatic
Hydrocarbons and Cyclohexane
ASTM D2320 Standard Test Method for Density (Relative Density) of Solid Pitch (Pycnometer Method)
ASTM D2638 Standard Test Method for Real Density of Calcined Petroleum Coke by Helium
Pycnometer
ASTM D2962 Standard Test Method for Calculating Volume-Temperature Correction for Coal-Tar
Pitches
ASTM D3142 Standard Test Method for Specific Gravity, API Gravity, or Density of Cutback Asphalts
by Hydrometer Method
ASTM D3505 Standard Test Method for Density or Relative Density of Pure Liquid Chemicals
ASTM D4311 Standard Practice for Determining Asphalt Volume Correction to a Base Temperature

Pilot Version, September 2009 B-11

ASTM D4292 Standard Test Method for Determination of Vibrated Bulk Density of Calcined Petroleum
Coke
ASTM D4892 Standard Test Method for Density of Solid Pitch (Helium Pycnometer Method)
ASTM D5002 Standard Test Method for Density and Relative Density of Crude Oils by Digital Density
Analyzer
ASTM D5004 Standard Test Method for Real Density of Calcined Petroleum Coke by Xylene
Displacement
ASTM D7042 Standard Test Method for Dynamic Viscosity and Density of Liquids by Stabinger
Viscometer (and the Calculation of Kinematic Viscosity)
ASTM D7454 Standard Test Method for Determination of Vibrated Bulk Density of Calcined Petroleum
Coke using a Semi-Automated Apparatus

B.4 Gas Processors Association (GPA)

GPA Technical Standards Manual
GPA RB 101-43 Compression and Charcoal Tests for Determining the Natural Gasoline Content of Natural
Gas
GPA RB 181-86 Heating Value-Basis for Custody Transfer
GPA RB 194-94 Tentative NGL Loading Practices
GPA 1167 GPA Glossary—Definition of Words and Terms Used in the Gas Processing Industry
GPA 2103 Tentative Method for the Analysis of Natural Gas Condensate Mixtures Containing
Nitrogen and Carbon Dioxide by Gas Chromatography
GPA 2145 Table of Physical Properties for Hydrocarbons and Other Compounds of Interest to the
Natural Gas Industry
GPA 2166 Obtaining Natural Gas Samples for Analysis by Gas Chromatography
GPA 2172 Calculation of Gross Heating Value, Relative Density, Compressibility and Theoretical
Hydrocarbon Liquid Content for Natural Gas Mixtures for Custody Transfer
GPA 2174 Obtaining Liquid Hydrocarbon Samples for Analysis by Gas Chromatography
GPA 2177 Analysis of Natural Gas Liquid Mixtures Containing Nitrogen and Carbon Dioxide by Gas
Chromatography
GPA 2186 Method for the Extended Analysis of Hydrocarbon Liquid Mixtures Containing Nitrogen
and Carbon Dioxide by Temperature Programmed Gas Chromatography
GPA 2377 Test for Hydrogen Sulfide & Carbon Dioxide in Natural Gas Using Stain Tubes
GPA 8173 Methods for Converting Mass Natural Gas Liquids and Vapor to Equivalent Liquid
Volumes
GPA 8182 Standard for Mass Measurement of Natural Gas Liquids
GPA 8186 Measurement of Liquid Hydrocarbon by Truck Scales
GPA 8195 Tentative Standard for Converting Net Vapor Space Volumes to Equivalent Liquid
Volumes
GPA TP-27 Temperature Correction for the Volume of NGL and LPG Tables 23E, 24E, 53E, 54E, 59E
& 60E

Pilot Version, September 2009 B-12

B.4 American Gas Association (AGA)
AGA Report No. 3 Orifice Metering of Natural Gas, Part 1: General Equations & Uncertainty Guidelines
AGA Report No. 3 Orifice Metering of Natural Gas, Part 2: Specification and Installation Requirements
AGA Report No. 3 Orifice Metering of Natural Gas, Part 3: Natural Gas Applications
AGA Report No. 3 Orifice Metering of Natural Gas, Part 4: Background, Development Implementation
Procedure
AGA Report No. 4A Natural Gas Contract Measurement and Quality Clauses
AGA Report No. 5 Fuel Gas Energy Metering
AGA Report No. 7 Measurement of Natural Gas by Turbine Meter
AGA Report No. 8 Compressibility Factor of Natural Gas and Related Hydrocarbon Gases
AGA Report No. 9 Measurement of Gas by Multipath Ultrasonic Meters
AGA Report No. 10 Speed of Sound in Natural Gas and Other Related Hydrocarbon Gases
AGA Report No. 11 Measurement of Natural Gas by Coriolis Meter

Pilot Version, September 2009 B-13

 APPENDIX C

OPERATING CONDITIONS, INSPECTION, CALIBRATION

AND EXPECTED UNCERTAINTIES FOR COMMON FLOW METERS
 Appendix C
OPERATING CONDITIONS, INSPECTION, CALIBRATION
AND EXPECTED UNCERTAINTIES FOR COMMON FLOW METERS a
METER TYPE MEDIUM OPERATING CONDITIONS INSPECTION & CALIBRATION RANDOM ERROR b
Rotary meter Gas − Application filter for polluted − Cleaning: once per 10 years, − 0-20% of the measurement
gas streams recalibration and if necessary range: 3%
(Expected life span: 25 adjusting − 20-100% of the
years) − Annual inspection: oil level of measurement range: 1.5%
the carter
Turbine flow meter Gas − Application filter for polluted − Cleaning: once per 5 year, − 0-20% of the measurement
gas streams recalibration and adjusting, if range: 3%
(Expected life span: 25 − No pulsating gas stream necessary − 20-100% of the
years) − No overload of longer than 30 − Visual inspection: annual measurement range: 1.5%
minutes › 120% of maximum − Lubrication of bearings: once
measurement range per three months (not for
permanent lubricated bearings)
Bellows meter Gas − Cleaning: once per 10 year, − 0-20% of the measurement
recalibration and adjusting, if range: 6%
(Expected life span: 25 necessary − 20-100% of the range: 4%
years) − Annual maintenance: according
to general instructions for
measurement principle
Orifice meter Gas and − No corrosive gases and liquids − Annual calibration: pressure − 30-100% of the
Liquid − Orifice placement guidelines: transmitter measurement range: 1.5%
(Expected life span: 30 o Minimum of 4D free input − Calibration of the orifice meter:
years) flow length before the orifice once in 5 years
o Minimum of 2D after the − Annual inspection: abrasion
orifice orifice and fouling
− Smooth surface of inner wall − Annual maintenance according
to general instructions for
measurement principle

Pilot Version, September 2009 C-1

 Appendix C
OPERATING CONDITIONS, INSPECTION, CALIBRATION
AND EXPECTED UNCERTAINTIES FOR COMMON FLOW METERS (continued) a
METER TYPE MEDIUM OPERATING CONDITIONS INSPECTION & CALIBRATION UNCERTAINTY b
Venturi meter Gas and − No corrosive gases and liquids − Annual calibration: pressure − 20-100% of the
Liquid transmitter measurement range: 1.5%
(Expected life span: 30 − Calibration: entire measurement
years) instrument – once per 5 year
− Annual: visual inspection
− Annual maintenance according to
general instructions for
measurement principle
Ultrasonic meter Gas and − Transducer assembly to be − Cleaning: once per 5 years, − 1-100% of the measurement
Liquid replaced according to the recalibration and adjusting, if range: 0.5%
(Expected life span: 15 manufacturer’s specifications necessary
years) − No disturbances in frequencies − Annual inspection
− Composition of flowing o Contact between transducer
medium should be known and tube wall
− Ultrasonic meters guidelines: o Wall corrosion
o Minimum of 10D free input o Transducers
flow length before the − Annual maintenance according to
meter, and instructions of manufacturer/
o Minimum of 5D after the general instructions measurement
meter principles
Vortex meter Gas − Set-up is free of vibration − Cleaning: once per 5 years, − 10-100% of the
− Avoid compressive shocks recalibration and adjusting, if measurement range: 2%
(Expected life span: 10 − Vortex meters guidelines: needed
years) o Minimum of 15D free input − Annual inspection of: sensors,
flow length before the bluff body, and wall corrosion,
meter, and − Annual maintenance according to
o Minimum of 5D after the general instructions for
meter measurement principle
a
Based on material presented in the ETSG, July 2007 document and the sources cited within
b
Random errors are based on experience of operating the respective flowmeters under designated installation, operation, calibration, inspection and maintenance procedures.

Pilot Version, September 2009 C-2

 APPENDIX D

SELECT MEASUREMENT METHODS SUMMARIES

 Appendix D

SELECT MEASUREMENT METHODS SUMMARIES

D.1 Carbon Content Measurement Methods

a. ASTM Test Method 1945-03, Analysis of Natural Gas by Gas Chromatography (July 2003)

This test method covers the determination of the chemical composition of natural gases and similar gaseous
mixtures within the range of composition shown in Table D-1. This test method may be abbreviated for the
analysis of lean natural gases containing negligible amounts of hexanes and higher hydrocarbons, or for the
determination of one or more components, as required.

Table D-1. Natural Gas Components and Range of Composition Covered

COMPONENT MOL %
Helium 0.01 to 10
Hydrogen 0.01 to 10
Oxygen 0.01 to 20
Nitrogen 0.01 to 100
Carbon dioxide 0.01 to 20
Methane 0.01 to 100
Ethane 0.01 to 100
Hydrogen sulfide 0.3 to 30
Propane 0.01 to 100
Isobutane 0.01 to 10
n-Butane 0.01 to 10
Neopentane 0.01 to 2
Isopentane 0.01 to 2
n-Pentane 0.01 to 2
Hexane isomers 0.01 to 2
Heptanes+ 0.01 to 1

Components in a representative sample are physically separated by gas chromatography (GC) and compared
to calibration data obtained under identical operating conditions from a reference standard mixture of known
composition. The numerous heavy-end components of a sample can be grouped into irregular peaks by
reversing the direction of the carrier gas through the column at such time as to group the heavy ends either
as C5 and heavier, C6 and heavier, or C7 and heavier. The composition of the sample is calculated by
comparing either the peak heights, or the peak areas, or both, with the corresponding values obtained with
the reference standard. This test method is of significance for providing data for calculating physical
properties of the sample, such as heating value and relative density, or for monitoring the concentrations of
one or more of the components in a mixture.

Pilot Version, September 2009 D-1

 For natural gas samples that meet the US pipeline quality specifications (1000 Btu/scf, or 38 MJ/m3), the
precision of this test method has been determined by the statistical examination of the interlaboratory test
results, as documented in Table D-2.

Table D-2. ASTM D1945-03 Precision for Natural Gas Samples1

COMPONENT
(MOLE %) REPEATABILITY REPRODUCIBILITY
0 to 0.09 0.01 0.02
0.1 to 0.9 0.04 0.07
1.0 to 4.9 0.07 0.10
5.0 to 10 0.08 0.12
Over 10 0.10 0.15
1
ASTM 1945-03, July 2003

b. ASTM D1946-90, Analysis of Reformed Gas by Gas Chromatography, (Reapproved 2006)

This practice covers the determination of the chemical composition of reformed gases and similar gaseous
mixtures containing the following components: hydrogen, oxygen, nitrogen, carbon monoxide, carbon
dioxide, methane, ethane, and ethylene.

Components in a sample of reformed gas are physically separated by gas chromatography and compared to
corresponding components of a reference standard separated under identical operating conditions, using a
reference standard mixture of known composition. The composition of the reformed gas is calculated by
comparison of either the peak height or area response of each component with the corresponding value of
that component in the reference standard.

The chemical composition data can be used to calculate physical properties of the gas and its
interchangeability with other fuel gases. The quality of data obtained by this method depends on the
preparation of moisture-free and homogenous mixtures of known composition for comparison with the test
sample. The fraction of a component in the reference standard should not be < 0.5 mole%, nor differ by
more than 10 mole%, from the fraction of the corresponding component in the tested sample, and its
composition should be known to within 0.01 mole% for any component.

Method precision in terms of reproducibility and repeatability are provided in Table D-3 below.

Table D-3. ASTM D1946-90 Precision for Reformed Gas Samples

COMPONENT
(MOLE %) REPEATABILITY REPRODUCIBILITY
0 to 1 0.05 0.1
1 to 5 0.1 0.2
5 to 25 0.3 0.5
Over 25 0.5 1.0

Pilot Version, September 2009 D-2

c. ASTM UOP539-97, Refinery Gas Analysis by Gas Chromatography

This method is for determining the composition of refinery gas samples or expanded liquefied petroleum gas
(LPG) samples that are obtained from refining processes or natural sources. It provides individual results
for non-condensable gases, hydrogen sulfide, C1 through C4 hydrocarbons and C5 paraffins, while C5
olefins and C6+ hydrocarbons are provided as a composite. The method yields quantitative results from 0.1
to 99.9 mole% for a single component or composite, except for hydrogen sulfide that yields quantitative
results between 0.1 and 25 mole%.

d. ISO 6974, Natural Gas - Determination of composition with defined uncertainty by gas chromatography
(October 2002)

ISO 6974 consists of the following six parts:

Part 1: Guidelines for tailored analysis.

Part 2: Measuring-system characteristics and statistics for processing of data.

Part 3: Determination of hydrogen, helium, oxygen, nitrogen, carbon dioxide and hydrocarbons up to C8
using two packed columns.

Part 4: Determination of nitrogen, carbon dioxide and C1 to C5 and C6+ hydrocarbons for a laboratory and
on-line measuring system using two columns.

Part 5: Determination of nitrogen, carbon dioxide and C1 to C5 and C6+ hydrocarbons for a laboratory and
on-line process application using three columns.

Part 6: Determination of hydrogen, helium, oxygen, nitrogen, carbon dioxide, and C1 to C8 hydrocarbons
using three capillary columns.

ISO 6974 describes a gas chromatographic methods for the quantitative determination of the content of
hydrogen, helium, oxygen, nitrogen, carbon dioxide, and C1 to C8 hydrocarbons in natural gas samples
using two or three packed or capillary columns combinations. They are applicable to the analysis of gases
containing constituents within the mole fraction ranges given in the respective parts of the standard, and is
commonly used for laboratory applications. These ranges do not represent the limits of detection, but the
limits within which the stated precision of the method applies. Although one or more components in a
sample may not be present at detectable levels, the method can still be applicable. This method can also be
applicable to the analysis of natural gas substitutes.

For example, the set-up for the determination of hydrogen, helium, oxygen, nitrogen, carbon dioxide, and
hydrocarbons from C1 to C8 by gas chromatography using three capillary columns (part 6 of the standard) is
as follows:

− A PLOT precolumn is used for the separation of carbon dioxide (CO2) and ethane (C2H6);

Pilot Version, September 2009 D-3

 − A molecular sieve PLOT column is used for the separation of the permanent gases helium (He),
hydrogen (H2), oxygen (O2), nitrogen (N2), and methane (CH4).

− A thick film WCOT2) column coated with an apolar phase is used for the separation of the C3 to C8
(and heavier) hydrocarbons.

The permanent gases helium (He), hydrogen (H2), oxygen (O2), nitrogen (N2), and methane (CH4) are
detected with a thermal conductivity detector (TCD). The C2 to C8 hydrocarbons are detected with a flame
ionization detector (FID).

For the analysis of natural gas substitutes, carbon monoxide (CO) and carbon dioxide (CO2) are detected
using an FID after reduction of the components to CH4 by a methanizer. Use of a methanizer makes it
possible to detect CO and CO2 at mole fractions greater than 0,001%. If the samples do not include CO or
CO2, or if the CO and/or the CO2 mole fraction exceeds 0,02%, a methanizer is not required. CO and CO2
may then alternatively be detected with the TCD.

When analyzing natural gas substitutes, the PLOT column described in 3.1 can also be used for the
separation of ethyne (C2H2) and ethene (C2H4) and the molecular sieve PLOT column can also be used for
the analysis of carbon monoxide (CO).

Typical precision values for this method are provided in Table A.1 of the standard. These values have been
obtained from practical experience and give an indication of the performance of the method. As such, they
cannot be compared with precision values mentioned in informative annexes of other parts of ISO 6974, as
they are very much dependent on the quality of the calibration gases used and the laboratory skills.

For specifies concentrations < 1.0 mole%, the relative repeatability and reproducibility is expected to be 2
and 4%, respectively. For higher concentrations, ranging from 1-50 mole%, the relative repeatability and
reproducibility are around 0.8 and 1.6% respectively.

e. ASTM D2650 – 04, Standard Test Method for Chemical Composition of Gases By Mass Spectrometry
(November 2004)

This test method is applicable for the quantitative analysis of gases containing specific combinations of the
following components: hydrogen; hydrocarbons with up to six carbon atoms per molecule; carbon
monoxide; carbon dioxide; mercaptans with one or two carbon atoms per molecule; hydrogen sulfide; and
air (nitrogen, oxygen, and argon). This test method is not applicable for the determination of constituents
that are present in amounts less than 0.1 mole %. This test method was developed on a specific type of
Mass Spectrometer, thus users of other instruments may have to modify operating parameters and the
calibration procedure and adapt it to their instrument.

The method sets out the experimental procedures, while the calculation procedures will depend on the
knowledge of qualitative mixture composition; errors due to components presumed absent; minimum cross

Pilot Version, September 2009 D-4

 interference between known components; maximum sensitivity to known components; low frequency and
complexity of calibration; and type of computing system available. The standard includes a tabulation of
calculation procedures that are recommended for stated applications.

f. ASTM D5291 – 02, Standard Test Methods for Instrumental Determination of Carbon, Hydrogen, and
Nitrogen in Petroleum Products and Lubricants, (2007)

This ASTM standard covers the simultaneous determination of carbon, hydrogen, and nitrogen in petroleum
products and lubricants. The results, expressed as total carbon, total hydrogen, and total nitrogen, are useful
in determining the complex nature of sample types covered by this test method, and can also be used to
determine the total carbon content that could be converted to CO2 upon combustion of the product.

These test methods are applicable to samples such as crude oils, fuel oils, additives, and residues for carbon
and hydrogen and nitrogen analysis. These test methods were tested in the concentration range of at least 75
to 87-wt% for carbon, at least 9 to 16-wt% for hydrogen, and 0.1 to 2-wt% for nitrogen. These test methods
are not recommended for the analysis of volatile materials such as gasoline, gasoline-oxygenate blends, or
gasoline type aviation turbine fuels.

D.2 Heating Value Measurement Methods

a. ASTM D4891 – 89, Test Method for Heating Value of Gases in Natural Gas Range by Stoichiometric
Combustion, (Reapproved 2006)

This test method covers the determination of the heating value of natural gases and similar gaseous mixtures
within the range of composition shown in Table D-4.

Table D-4. ASTM D4891-89 Range of Composition for Natural Gas Components

COMPOUND CONCENTRATION RANGE (MOLE, %)

Helium 0.01 to 5
Nitrogen 0.01 to 20
Carbon dioxide 0.01 to 10
Methane 50 to 100
Ethane 0.01 to 20
Propane 0.01 to 20
n-butane 0.01 to 10
isobutane 0.01 to 10
n-pentane 0.01 to 2
isopentane 0.01 to 2
Hexanes and heavier 0.01 to 2

Pilot Version, September 2009 D-5

 In this method air is mixed with the gaseous fuel to be tested. The mixture is burned and the air-fuel ratio is
adjusted so that essentially a stoichiometric proportion of air is present. The adjustment is made so that the
air-fuel ratio is in a constant proportion to the stoichiometric ratio that is a relative measure of the heating
value. To set this ratio, a characteristic property of the burned gas is measured, such as temperature or
oxygen concentration.

This test method provides an accurate and reliable procedure to measure the total heating value of a fuel gas,
on a continuous basis, which is used for regulatory compliance, custody transfer, and process control. Some
instruments which conform to the requirements set forth in this test method can have response times on the
order of 1 min or less and can be used for on-line measurement and control. The method is sensitive to the
presence of oxygen and unsaturated hydrocarbons. For components not listed and composition ranges that
fall outside those in Table 7, such as in process or refinery fuel gases, modifications in the method may be
required to obtain correct results.

In testing the precision and accuracy of this method, repeatability within a laboratory was shown to be 0.76
Btu/scf, with the corresponding 95% confidence of the repeatability interval being 2.1 Btu/scf.
Reproducibility between laboratories was determined to be 1.67 Btu/scf with the corresponding 95%
confidence reproducibility interval 5.1 Btu/scf. The average bias of all measurements agreed with the
average reference value to within 0.1%.

b. ASTM D1826 – 94, Standard Test Method for Calorific (Heating) Value of Gases in Natural Gas Range
by Continuous Recording Calorimeter, (Reapproved 2003)

This test method covers the determination – with the continuous recording calorimeter – of the total calorific
(heating) value of fuel gas produced or sold in the natural gas range from 900 to 1200 Btu/scf. The heating
value is determined by imparting the heat obtained from the combustion of the test gas to a stream of air and
measuring the rise of the air temperature. The streams of test gas and heat absorbing air are maintained in
fixed volumetric proportion to each other by metering devices similar to the ordinary wet test meters geared
together and driven from a common electric motor. The meters are mounted in a tank of water, the level of
which is maintained and the temperature of which determines the temperature of the entering gas and air.

The experimental set-up is such that the temperature rise produced in the heat-absorbing air is directly
proportional to the heating value of the gas. Since all the heat from the combustion of the test gas sample,
including the latent heat of vaporization of the water vapor formed in the combustion, is imparted to the
heat-absorbing air, the calorimeter makes a direct determination of total heating value. The temperature rise
is measured by nickel resistance thermometers and is translated into Btu/scf.

This test method provides an accurate and reliable method to measure the total calorific value of a fuel gas,
on a continuous basis, which is used for regulatory compliance, custody transfer, and process control. As
far as precision, the calorimeters tested were standardized with methane weekly, and a rigid control was

Pilot Version, September 2009 D-6

 maintained over the room temperature so that no errors were caused by a change in the tank water
temperature. The data indicate that one week after standardization about 95% of the errors were less than
0.3% with a few errors as high as 0.5%. It is expected that errors greater than these may be found if the
period between checking against the standard methane is greater than one week.

c. ASTM D7313 – 08, Standard Practice for Determination of the Heating Value of Gaseous Fuels using
Calorimetry and On-line/At-line Sampling (May 2008)

This practice is used for the determination of the heating value measurement of gaseous fuels using a
calorimeter with at-line and in-line instruments that are operated from time to time on a continuous basis.
This type of near-real time monitoring systems that measure fuel gas characteristics such as heating value
are prevalent in various gaseous fuel industries and in industries either producing or using gaseous fuel in
their industrial processes. The installation and operation of particular systems would vary depending on
process type, regulatory requirements, and the user’s objectives and performance requirements.

In operating the system, a representative sample of the gaseous fuel is extracted from a process pipe, a
pipeline, or other gaseous fuel stream and is transferred to an analyzer sampling system. After conditioning
that maintains the sample integrity, the sample is introduced into a calorimeter. Excess extracted process or
sample gas is vented to the atmosphere, a flare header, or is returned to the process in accordance with
applicable economic and environmental requirements and regulations. Post combustion gasses from the
calorimeter are typically vented to the atmosphere. The heating value is calculated based upon the
instrument’s response to changes in the heating value of the sample gas using an applicable computation
algorithm. This practice is providing guidance for standardized start-up procedures, operating procedures,
and quality assurance practices for calorimeter based on-line, at-line, in-line and other near-real time
heating-value monitoring systems.

d. ASTM D4809 – 06, Standard Test Method for Heat of Combustion of Liquid Hydrocarbon Fuels by
Bomb Calorimeter (Precision Method), (2006)

The heat of combustion is a measure of the energy available from a fuel, and it is essential when considering
the thermal efficiency of equipment for producing either power or heat. This procedure measures the mass
heat of combustion, namely, the heat of combustion per unit mass of fuel. The volumetric heat of
combustion, that is, the heat of combustion per unit volume of fuel, can be calculated by multiplying the
mass heat of combustion by the density of the fuel (mass per unit volume).

This test method covers the determination of the heat of combustion of hydrocarbon fuels, and is designed
for high precision with the difference between duplicate determinations in the order of 0.2%, for aviation or
turbine fuels, or greater differences for a wide range of volatile and nonvolatile materials.

In order to attain this high precision, strict adherence to all details of the procedure is essential since the
error contributed by each individual measurement that affects the precision ought to be kept below 0.04%,

Pilot Version, September 2009 D-7

 insofar as possible. Under normal conditions, the method is directly applicable to such fuels as gasoline,
kerosene, Nos. 1 and 2 fuel oil, Nos. 1-D and 2-D diesel fuel, and Nos. 0-GT, 1-GT, and 2-GT gas turbine
fuels.

e. ASTM D5865 – 07a, Standard Test Method for Gross Calorific Value of Coal and Coke, (2007)

The gross calorific value can be used to compute the total calorific content of the quantity of coal or coke,
and it can also be used for computing the calorific value versus sulfur content to determine whether the coal
meets regulatory requirements for industrial fuels.

This test method pertains to the determination of the gross calorific value of coal and coke by either an
isoperibol or adiabatic bomb calorimeter. The resulting values are to be regarded as standard, and no other
units of measurement are included in this standard.

Methods Cited

ASTM 1945 – 03, “Analysis of Natural Gas by Gas Chromatography”, current edition approved May 10, 2003.
Published July 2003. Originally approved in 1962. Last previous edition approved in 2001 as D1945–96(2001).
2ASTM D 1946 – 90 (Reapproved 2006), Standard Practice for Analysis of Reformed Gas by Gas
Chromatography, current edition approved June 1, 2006. Published June 2006 (Originally approved in 1962)
ASTM UOP UOP539-97, “Refinery Gas Analysis by Gas Chromatography”
ISO 6974, 2002, “Natural gas - Determination of composition with defined uncertainty by gas chromatography”
(in 6 parts); First edition 2002-10-15, Geneva, Switzerland
ASTM D2650, 2004, “Standard Test Method for Chemical Composition of Gases by Mass Spectrometry”,
Published November 1, 2004.
ASTM D5291 – 02, (2007) “Standard Test Methods for Instrumental Determination of Carbon, Hydrogen, and
Nitrogen in Petroleum Products and Lubricants”, Published 2007.
ASTM D 4891 – 89 (Reapproved 2006), Standard Test Method for Heating Value of Gases in Natural Gas
Range by Stoichiometric Combustion, Current edition approved June 1, 2006. Published June 2006. Originally
approved in 1989. Last previous edition approved in 2001 as D4891–89 (2001)
ASTM D 1826 – 94 (Reapproved 2003), “Standard Test Method for Calorific (Heating) Value of Gases in
Natural Gas Range by Continuous Recording Calorimeter”, Current edition approved May 10, 2003. Published
May 2003. Originally approved in 1961. Last previous edition approved in 1998 as D 1826 – 94 (1998).
ASTM D7313 – 08, “Standard Practice for Determination of the Heating Value of Gaseous Fuels using
Calorimetry and On-line/At-line Sampling”, Current edition approved May 1, 2008. Published May 2008.
ASTM D4809 – 06, “Standard Test Method for Heat of Combustion of Liquid Hydrocarbon Fuels by Bomb
Calorimeter (Precision Method)”, Published 2006.

ASTM D4809 – 06, “Standard Test Method for Heat of Combustion of Liquid Hydrocarbon Fuels by Bomb
Calorimeter (Precision Method)”, Published 2007.

Pilot Version, September 2009 D-8

 APPENDIX E

UNITS CONVERSION
 Appendix E

UNITS CONVERSION3

Energy units
Quantities
• 1.0 calorie = 4.187 J
• 1.0 gigajoule (GJ) = 109 joules = 0.948 million Btu = 239 million calories = 278 kWh
• 1.0 British thermal unit (Btu) = 1055 joules (1.055 kJ)
• 1.0 Quad = One quadrillion Btu (1015 Btu) = 1.055 exajoules (EJ)
− Approximately 172 million barrels of oil equivalent (boe)
• 1000 Btu/lb = 2.33 gigajoules per tonne (GJ/t)
• 1000 Btu/US gallon = 0.279 megajoules per liter (MJ/l)
Power
• 1.0 watt = 1.0 joule/second = 3.413 Btu/hr
• 1.0 kilowatt (kW) = 3413 Btu/hr = 1.341 horsepower
• 1.0 kilowatt-hour (kWh) = 3.6 MJ = 3413 Btu
• 1.0 horsepower (hp) = 550 foot-pounds per second = 2545 Btu per hour = 745.7watts = 0.746kW
Energy Costs
• $1.00 per million Btu = $0.948/GJ
• $1.00/GJ = $1.055 per million Btu

Common units of measure

• 1.0 U.S. ton (short ton) = 2000 pounds
• 1.0 imperial ton (long ton or shipping ton) = 2240 pounds
• 1.0 metric tonne (tonne) = 1000 kilograms = 2205 pounds
• 1.0 US gallon = 3.79 liter = 0.833 Imperial gallon
• 1.0 imperial gallon = 4.55 liter = 1.20 US gallon
• 1.0 liter = 0.264 US gallon = 0.220 imperial gallon

Fossil fuels4
• Barrel of oil equivalent (boe) = approx. 6.1 GJ (5.8 million Btu), equivalent to 1700 kWh
− "Petroleum barrel" is a liquid measure equal to 42 U.S. gallons (35 Imperial gallons or 159 liters);
about 7.2 barrels oil are equivalent to one tonne of oil (metric) = 42-45 GJ.
• Gasoline: LHV = 115,000 Btu/gallon = 121 MJ/gallon = 32 MJ/liter; HHV: 125,000 Btu/gallon = 132
MJ/gallon = 35 MJ/liter
• Metric tonne gasoline = 8.53 barrels = 1356 liter = 43.5 GJ/t (LHV); 47.3 GJ/t (HHV)

3
Based on data from the ORNL Bioenergy Feedstock Information Network; [Link]
4
The energy content (heating value) of petroleum products per unit mass is fairly constant, but their density differs
significantly, hence their energy content is different

Pilot Version, September 2009 E-1

 − Gasoline density (average) = 0.73 g/ml (= metric tonnes/m3)
• Petro-diesel = 130,500 Btu/gallon (36.4 MJ/liter or 42.8 GJ/t)
− Petro-diesel density (average) = 0.84 g/ml (= metric tonnes/m3)
• Metric tonne ethanol = 7.94 petroleum barrels = 1262 liters
− Ethanol energy content: LHV = 11,500 Btu/lb = 75,700 Btu/gallon = 26.7 GJ/t = 21.1 MJ/liter;
HHV = 84,000 Btu/gallon = 89 MJ/gallon = 23.4 MJ/liter
− Ethanol density (average) = 0.79 g/ml (= metric tonnes/m3)
• Metric tonne biodiesel = 37.8 GJ (33.3 - 35.7 MJ/liter)
− Biodiesel density (average) = 0.88 g/ml (= metric tonnes/m3)
• Metric tonne coal5 = 27-30 GJ (bituminous/anthracite); 15-19 GJ (lignite/sub-bituminous) (the above
ranges are equivalent to 11,500-13,000 Btu/lb and 6,500-8,200 Btu/lb).
• Natural gas: HHV = 1027 Btu/ft3 = 38.3 MJ/m3; LHV = 930 Btu/ft3 = 34.6 MJ/m3
• Therm (used for natural gas, methane) = 100,000 Btu (= 105.5 MJ)

Carbon content of selected fuels6

• Coal (average) = 25.4 metric tonnes carbon per terajoule (TJ)
− 1.0 metric tonne coal = 746 kg carbon
• Oil (average) = 19.9 metric tonnes carbon / TJ
• 1.0 US gallon gasoline (0.833 Imperial gallon, 3.79 liter) = 2.42 kg carbon
• 1.0 US gallon diesel/fuel oil (0.833 Imperial gallon, 3.79 liter) = 2.77 kg carbon
• Natural gas (methane) = 14.4 metric tonnes carbon / TJ
• 1.0 cubic meter natural gas (methane) = 0.49 kg carbon

5
The energy content (heating value) per unit mass varies greatly between different "ranks" of coal. "Typical" coal (rank
not specified) usually means bituminous coal, the most common fuel for power plants (27 GJ/t).
6
The average carbon content values above are indicative only. More accurate determinations should be sought from fuel
provider, or by testing fuels used.

Pilot Version, September 2009 E-2

 APPENDIX F

UNCERTAINTY ESTIMATION DETAILS FOR AN EXAMPLE INVENTORY

 Appendix F

UNCERTAINTY ESTIMATION DETAILS FOR AN EXAMPLE INVENTORY

As stated in Section 5.0, this Appendix provides the source-by-source calculation and aggregation of
uncertainty for a hypothetical facility.7

The following summarizes the characteristics of the onshore oil field example facility introduced in
Section 5. This example field is described in detail in the API Compendium (API, 2009). The calculation
exhibits that are reprinted directly from the API Compendium but have additional uncertainty information
added, are indicated by the highlighted text. Excerpts of Tables 5-1 and 5-2 are provided to reiterate
information used in the uncertainty calculations.

Facility Description: An onshore oil field in Texas consists of 320 producing oil wells.

Throughput: The average daily oil and gas production rates are 6,100 bbl/day and 30×106 scf/day,
respectively.

Operations: The facility operates approximately 343 days per year. The facility imports 917 MW-hr from
the eGRID subregion ERCOT all annually. The facility gas composition is presented in Table F-1 and
results in a heating value of 928 Btu/scf with an uncertainty of ±4%, based on engineering
judgment.

Table F-1. Gas Composition for Onshore Production Platform

(Reprinted from the API Compendium)

Uncertainty a
Gas Compound Produced Gas Mole % (±%)
CO2 12 4
N2 2.1 4
CH4 80 4
C2H6 4.2 4
C3H8 1.3 4
C4H10 0.4 4
Footnote:
a
Uncertainty is based on engineering judgment at a 95% confidence interval.
Note: the values shown above are for example only. They do not reflect average operations.

7
Note, the calculations shown in this Appendix are rounded to three significant figures. Actual calculations were
compiled in a spreadsheet with rounding reserved for the final results.

Pilot Version, September 2009 F-1

 In this hypothetical example, CO2 emissions from combustion are calculated using the gas composition
approach as presented in the API Compendium. The calculations for CO2 emissions from combustion are
reprinted directly from the API Compendium. Some uncertainty values were assigned as follows:

• Fuel consumption for boilers, turbines, and flares = ±15% (assigned by expert judgment in
Table 5-1).
• Heating Value of Gas Combusted = 928 Btu/scf ±4% (measured independent of gas composition;
assigned by expert judgment here).
• Gas Composition measurement = ±4% (determined by analysis of repeat samples using
techniques from Section 4.0; assigned by expert judgment here).
• Unit Capacity of Heaters and Reboilers = ±6.71% (Calculated for Table 5-1).
F.1 Uncertainty in Natural Gas Combustion Devices – CO2 Emissions

Carbon dioxide emissions are estimated based on the volume of fuel consumed and the fuel carbon
content. Boilers, heater/reboilers, and compressor engines-turbines use natural gas at this facility.
However, because CH4 and N2O emission factors are based on the type of equipment as well as the fuel
consumed, the CO2 emission estimates are also calculated for common equipment types.

We will first examine boilers and heaters/reboilers. The quantity of natural gas consumed by this
equipment is based on the parameters shown in Table F-2.

Table F-2. Operating Parameters for Boilers, Heaters and Reboilers

Annual
Average Activity
No. Unit Operation Factor (all
of Uncertainty Capacity Uncertainty (per unit Uncertainty units Uncertainty
Source Fuel Units (±%)a (per unit) (±%)a per year) (±%)a combined) (±%)a
Boilers Produced 40×106
6 ±0% units N/A N/A ±15% scf/yr
Gas scf/yr
Heaters/ Produced 2×106 343 53.2×106 ±6.71%
3 ±0% units ±5% Btu/hr ±2% days/yr
reboilers Gas Btu/hr days/yr scf/yr scf/yr

The volume of fuel combusted (V) for Boilers and Heaters/Reboilers is:

40 ×106 scf ⎛ 2 ×106 Btu 24hr 343 day scf ⎞

V= + ⎜ 3 units × × × × ⎟
yr ⎝ hr day yr 928 Btu ⎠
V = (40 + 53.2) ×106 scf/yr
V = 93.2 ×106 scf/yr

Uncertainty Estimate of Heater/Reboilers Activity Data: (Equation 4-6, relative uncertainties)

The uncertainty for the heaters/reboilers is examined first, then the combined natural gas uncertainty is
estimated.

Pilot Version, September 2009 F-2

The uncertainty for the heaters/reboilers activity value is calculated by applying Equation 4-6 and using the
relative uncertainty values.

U (rel ) heater = U(rel) 2Units +U(rel)Capacity

2 2
+U(rel) Days 2
+U(rel) Heating Value

U (rel ) heater = 02 + 52 + 22 + 42 = ±6.71% ( scf/yr )

The uncertainty for the boilers fuel consumption is 15% based on expert judgment.

Combined Uncertainty for Boilers and Heaters/Reboilers Activity Data: (Equation 4-4, Absolute
Uncertainties)

The natural gas usage is summed. Therefore, the uncertainty for the combined fuel consumption by the
boiler and heaters/reboiler is calculated by applying Equation 4-4 and using the absolute uncertainty values.

U (abs ) Boilers and heaters Gas Usage = (0.150×40×106 ) 2 +(0.0735×53.2×106 ) 2 =6.98×106scf/yr

6.98 × 106 scf/yr
U (rel ) Boilers and heaters Gas Usage = = ±7.49% ( scf/yr )
93.2 × 106 scf/yr

Uncertainty Estimate of Turbines Activity Data:

The uncertainty associated with the fuel consumed by turbines is assigned by expert judgment, as shown
in Table F-3. In this case, no further calculations are required for the turbine activity data.

Table F-3. Operating Parameters for Turbines

Annual
Unit Activity Factor
No. of Uncertainty Capacity Uncertainty (all units Uncertainty
Source Fuel Units (±%)a (per unit) (±%)a combined) (±%)a
Compressor
Produced
engines – 11 ±0% units N/A 250×106 scf/yr ±15% scf/yr
Gas
turbines

Fuel Composition and Carbon Content:

The carbon content of the natural gas is needed to complete the estimation of CO2 emissions from natural
gas combustion. The gas composition data for this example are sown in Table F-4. Calculation of the
uncertainty values is described below.

Pilot Version, September 2009 F-3

 Table F-4. Natural Gas Composition

Mole % Uncertainty MW Mole% Uncertainty Carbon Uncertainty

% ×MW % Content %
(wt%C)
CO2 12.0 4 44.01 5.28 4 6.94 4.82
N2 2.1 4 28.01 0.588 4 0 4.82
CH4 80.0 4 16.04 12.8 4 46.3 4.82
C2H6 4.2 4 30.07 1.26 4 4.86 4.82
C3H8 1.3 4 44.10 0.573 4 2.26 4.82
C4H10 0.4 4 58.12 0.232 4 0.925 4.82
Fuel 100 MW mixture ±2.69% lb/scf 61.24 ±3.71%
Mixture 20.77 lb/scf (lbC/lb total) (lbC/lb total)

For simplicity, we assume a 4% uncertainty on the mole% for all of the natural gas components. In reality,
the uncertainty of the gas composition should be determined through multiple samples, as shown in
Section 4.

Uncertainty Estimate of the Fuel Mixture Mole %:

The relative uncertainty of the mole% × MW for the individual components is the same as the relative
uncertainty for the mole%, since multiplication by a constant does not change the relative uncertainty. The
uncertainty for the sum of the mole% × MW is calculated by applying Equation 4-4 and using the absolute
uncertainties.

U (abs )
∑ (mole%×MW)
= ∑ U(abs) 2
mole%×MW

(0.04 × 5.28) 2 + (0.04 × 0.588) 2 + (0.04 ×12.8) 2

U (abs ) = = 0.558
∑ (mole%×MW) + (0.04 × 1.26) 2 + (0.04 × 0.573) 2 + (0.04 × 0.232) 2
0.558
U (rel ) = × 100% = ±2.69% ( lb/scf )
∑ (mole%×MW)
20.77

The uncertainty for the wt% C for the individual components is calculated by applying Equation 4-6 and
using the relative uncertainties.

U (rel ) wt%C = U(rel ) mole%

2
+U(rel ) 2∑ (Mole%×MW) = 42 + 2.692 = ±4.82% ( wt% C )

The uncertainty for the sum of the wt%C is calculated by applying Equation 4-4 and using the absolute
uncertainties.

Pilot Version, September 2009 F-4

U (abs )
∑ (Wt%C)
= ∑ U(abs) 2
Wt%C

(0.0482 × 6.94) 2 + (0.0482 × 0) 2 + (0.0482 × 46.3) 2

U (abs ) = = 2.27
∑ (Wt%C) +(0.0482 × 4.86) 2 + (0.0482 × 2.26) 2 + (0.0482 × 0.925) 2
2.27
U (rel ) = × 100% = ±3.71% ( lb C/lb total )
∑ (Mole%×MW) 61.24

Uncertainty Estimate of CO2 Emissions from Boilers and Heaters/Reboilers:

Carbon dioxide emissions can then be calculated using mass balance approach. Emissions are calculated
below, by equipment type.
Boilers and Heaters:
93.2 ×106scf fuel lbmole 20.77 lb fuel 0.6124 lb C lbmole C
E CO2 = × × × ×
yr 379.3 scf fuel lbmole fuel lb fuel 12.01 lb C
lbmole CO 2 44.01 lb CO 2 tonne
× × × = 5, 200 tonnes CO 2 / yr
lbmole C lbmole CO 2 2204.62 lb
As shown earlier, 93.2×106 scf/yr has an uncertainty of ±7.49%. The uncertainty in the carbon content of
the fuel mixture is 3.71% and the uncertainty of the MW of the mixture is 2.69%. The uncertainty in the
CO2 emissions is calculated by applying Equation 4-6 and using the relative uncertainty values.

U (rel ) Emissions = U(rel ) 2Boilers and Heaters Gas Usage +U(rel ) Carbon
2 2
Content +U(rel ) MW mix

U (rel ) Emissions = 7.492 + 3.712 + 2.692 = ±8.78% (tonnes CO 2 / yr)

Uncertainty Estimate of CO2 Emissions from Turbines:

The CO2 emissions from Turbines are:

250 ×106scf fuel lbmole 20.77 lb fuel 0.6124 lb C lbmole C
E CO2 = × × × ×
yr 379.3 scf fuel lbmole fuel lb fuel 12.01 lb C
lbmole CO 2 44.01 lb CO 2 tonne
× × × = 13,900 tonnes CO 2 / yr
lbmole C lbmole CO 2 2204.62 lb
The uncertainty for the activity factor 250×106 scf fuel/yr is 15%. Similarly, the uncertainty in the carbon
content of the fuel mixture is 3.71% and the uncertainty of the MW of the mixture is 2.67%. The
uncertainty in the CO2 emissions is calculated by applying Equation 4-6 and using the relative uncertainty
values.

U ( r e l ) E m is s io n s = U ( r e l ) 2A F + U ( r e l ) C2 a rb o n C o n te n t + U ( r e l ) 2M W m ix

U ( r e l ) E m is s io n s = 1 5 + 3 .7 1 + 2 .6 7 = ± 1 5 .7 % (to n n e s C O 2 /yr)
2 2 2

Pilot Version, September 2009 F-5

 F.2 Uncertainty in Diesel Combustion Devices – CO2 Emissions

The diesel firing rate is the sum of the activity factors for the generators and pumps as shown in
Table F-5.

Table F-5. Operating Parameters for Diesel Engines

Annual
Average Activity
No. Unit Operation Factor (all
of Uncertainty Capacity Uncertainty (per unit Uncertainty units Uncertainty
Source Fuel Units (±%)a (per unit) (±%)a per year) (±%)a combined) (±%)a
Emergency
±10% 2,912 ×106 ±12.3%
generator IC Diesel 1 ±0% units 1800 hp ±5% hr 200 hr/yr
(hr/yr) Btu/yr Btu/yr
engine
Fire water ±10%
24 hr/yr; 77.7 ×106 ±13.2%
pump IC Diesel 1 ±0% units 460 hp ±5% hr (hr/yr);
87% load Btu/yr Btu/yr
engine ±20% (load)

Uncertainty Estimate of Diesel Activity Data:

Because the diesel emission factor is provided on a heat input basis, the equipment ratings are converted
to volume of fuel consumed (V) on a heat input basis.

⎛ 1,800 hp 8,089 Btu 200 hr ⎞ ⎛ 460 hp 8,089 Btu 24 hr ⎞

V = ⎜ 1 Unit × × × ⎟ + ⎜1 Unit × × 0.87 × × ⎟
⎝ unit hp-hr yr ⎠ ⎝ unit hp-hr yr ⎠
V = (2,912+7.77)× 106 =2989.7×106 Btu/yr

We assume there is no uncertainty in the number of units, which here is the count of diesel engines. The
uncertainty in the unit capacity is 5%, the uncertainty in the energy efficiency of the engine is 5%, and the
uncertainty in the average hours of operation is 10% based on expert judgment. The uncertainty in the CO2
emissions is calculated by applying Equation 4-6 and using the relative uncertainty values.

U ( rel ) Activity Factor = U(rel ) 2Number Of Units +U(rel ) 2Unity Capacity +U(rel ) 2Efficiency +U(rel ) 2Average Operations

U ( rel ) ActivityFactor = 02 + 52 + 52 + 102 = ±12.3% ( Btu/yr )

The uncertainty for pumps includes one additional term: the uncertainty in the percent load is 20%,

U(rel ) 2Number Of Units +U(rel ) 2Unity Capacity +U(rel )Efficiency

2 2
+U(rel ) Percent Load
U (rel ) ActivityFactor =
+U(rel ) 2Average Operations

U (rel ) ActivityFactor = 02 + 52 + 52 + 202 + 102 = ±23.5% ( Btu/yr )

Pilot Version, September 2009 F-6

Combining the fuel usage for the two diesel fired equipment, the total uncertainty in the diesel firing rate
is calculated by applying Equation 4-4 and using the absolute uncertainty values.

2
U (abs ) Diesal Firing Rate = (U(abs )Generator +U(abs ) 2Pump

U (abs ) Diesal Firing Rate = (0.123 × 2912 × 106 ) 2 + (0.235 × 77.7 × 106 ) 2 = 356.85 ×106 Btu / yr
356.85×106 Btu/yr
U (rel ) = 100%× = ± 11.9% ( Btu/yr )
2989.7×106 Btu/yr

Uncertainty Estimate of CO2 Emissions from Diesel Combustion:

2,989.7 × 106 Btu 0.0732 tonnes CO 2
E CO2 = × = 219 tonnes CO 2 / yr
year 106 Btu

The uncertainty in the emissions factor is determined by engineering judgment to be 10%. The total
uncertainty for the CO2 emissions is then is calculated by applying Equation 4-6 and using the relative
uncertainty values.

2
U (rel )CO2 = U(rel ) AF 2
+U(rel ) EF = 11.92 + 10.02 = ±15.6% ( tonnes CO 2 /yr )

F.3 Uncertainty in Stationary Combustion Devices – CH4, N2O, and CO2e Emissions

In this hypothetical example, CH4 and N2O emissions from combustion are calculated using the fuel-based
emission factors presented in the API Compendium. The calculations for CH4 and N2O emissions from
combustion are reprinted directly from the API Compendium. Some uncertainty values were assigned as
follows:

• Fuel Consumption for Boilers = ±15% (assigned by expert judgment).

• Heating Value of Gas Combusted = ±4% (measured independent of gas composition; assigned by
expert judgment here).
• Gas Composition measurement = ±4% (determined by analysis of repeat samples using
techniques from Section 4; assigned by expert judgment here).
• The uncertainty in the emissions factor for N2O is determined by engineering judgment to be
+150%,–100%. The uncertainty in the emissions factor for methane is 25% based on expert
judgment.

Uncertainty Estimate of CH4 and N2O Emissions from Boilers and Heater/Reboilers:

Natural gas emission factors for CH4 and N2O are based on the energy consumed by the equipment. This
differs from the CO2 emission estimates above, which were based on the volume of natural gas consumed.
So, the fuel usage information from Table F-2 must first be converted to an energy input basis.

Pilot Version, September 2009 F-7

 ⎛ 40×106 scf 928 Btu ⎞ ⎛ 2×106 Btu 24 hr 343 days ⎞
V= ⎜ × +
⎟ ⎜ 3 Units× × × ⎟
⎝ yr scf ⎠ ⎝ hr-unit day yr ⎠
6
V = 86,512 ×10 Btu/yr

The uncertainty for the boiler activity factor is calculated by applying Equation 4-6 and using the relative
uncertainty values.

U (rel ) boiler AF = U(rel ) 2AF +U(rel )Heating Value = 15 + 4 = ±15.5% ( Btu/yr )

2 2 2

Uncertainty for heater/reboilers: (Equation 4-6, relative uncertainties)

U (rel ) heater = U(rel ) 2Units +U(rel )Capacity

2 2
+U(rel ) Days

U (rel ) heater = 02 + 52 + 22 = ±5.39% ( Btu/yr )

The uncertainty for the combined activity factor is calculated by applying Equation 4-4 and using the
absolute uncertainty values.

2 2
U (abs ) AF = (U(abs ) Boiler +U(abs ) Heater
U (abs ) AF = (0.155 × 37,120 × 106 ) 2 + (0.0539 × 49,392 × 106 ) 2 = 6,347 × 106 Btu/yr
6,347×106 Btu/yr
U (rel ) AF = 100% × = ±7.34% ( Btu/yr )
86,512×106 Btu/yr

As stated previously, the uncertainty in the emissions factor for N2O is determined by engineering judgment
to be 150%. The uncertainty in the emissions factor for methane is 25% based on expert judgment. The
uncertainty for emissions is calculated by applying Equation 4-6 and using the relative uncertainty values.

Because it is not possible to have negative emissions, the lower bound uncertainty will be truncated at zero.
As a result, the lower and upper bound uncertainties are estimated separately.

86,512 ×106 Btu 1.0 × 10-6 tonne CH 4

E CH4 : × = 0.0865 tonnes CH 4 /yr
yr 106 Btu

U (rel )emissions = U(rel ) 2AF +U(rel ) EF

2
= 7.342 + 252 = ±26.1% ( tonnes CH 4 /yr) )

86,512 × 106 Btu 2.8 × 10-7 tonne N 2 O

E N 2O = × = 0.0242 tonnes N 2 O/yr
yr 106 Btu

2
U (rel )emissions = U(rel ) AF 2
+U(rel )EF = 7.342 + 1502 = +150%, −100% ( tonnes N 2 O/yr )

Pilot Version, September 2009 F-8

Uncertainty Estimate of CO2e Emissions from Boilers and Heater/Reboilers:

The final step is to convert the emissions of CO2, CH4, and N2O to a CO2e. Note there are no uncertainties
for any of the GWPs, so the combined uncertainty is based on summing the individual emissions and
applying Equation 4-4 (using absolute uncertainty values). Also, because the N2O uncertainty estimate is
asymmetrical, separate upper and lower bound uncertainties are calculated for the CO2e. emissions.

5, 200 tonne CO2 ⎛ 0.0865 tonne CH4 21 tonne CO2e ⎞

E CO2e = +⎜ × ⎟
yr ⎝ yr tonne CH4 ⎠
⎛ 0.0242 tonne N2O 310 tonne CO2e ⎞
+⎜ × ⎟ = 5, 210 tonne CO2e/yr
⎝ yr tonne N2O ⎠

U (abs ) ECO e = (U(abs ) 2ECO +U(abs ) 2ECH +U(abs ) 2E N O

2 2 4 2

U (abs ) ECO e = (0.0878 × 5, 200) 2 + (0.261× 21× 0.0865) 2 + (1.50 × 310 × 0.0242) 2
2
Upper:
= 456 tonnes CO2e/yr
456 tonnes CO2e/yr
U (rel ) ECO e = 100% × = +8.77% ( tonnes CO2e/yr )
2
5, 210 tonnes CO2e/yr

U (abs ) ECO e = (U(abs ) 2ECO +U(abs ) 2ECH +U(abs ) 2E N O

2 2 4 2

U (abs ) ECO e = (−0.0878 × 5, 200) 2 + (−0.261× 21× 0.0865) 2 + (−1.00 × 310 × 0.0242) 2
2
Lower:
= 456 tonnes CO2e/yr
456 tonnes CO2e/yr
U (rel ) ECO e = 100% × = −8.77% ( tonnes CO2e/yr )
2
5, 210 tonnes CO2e/yr

Uncertainty Estimate of CH4 and N2O Emissions from Turbines:

Here also, the natural gas emission factors for CH4 and N2O are based on the energy consumed by the
equipment. The fuel usage information from Table F-3 is first converted to an energy input basis.

⎛ 250 ×106 scf 928 Btu ⎞

V =⎜ × ⎟ = 232,000 × 10 Btu/yr
6

⎝ yr scf ⎠

Then the emission factors are applied to estimated CH4 and N2O emissions:

Pilot Version, September 2009 F-9

 232,000 ×106 Btu 3.9 ×10-6 tonne CH 4
E CH4 = × = 0.905 tonnes CH 4 /yr
yr 106 Btu

232,000 ×106 Btu 1.4 × 10-6 tonne N 2 O

E N2O = × = 0.325 tonnes N 2 O/yr
yr 106 Btu

The uncertainty for the turbines’ activity factor is 15%. The uncertainty in the heating values is 5% based
on expert judgment. The uncertainty in the CH4 emissions factor is determined by engineering judgment to
be 25%, and +150%, -100% for the N2O emission factor. The uncertainty of the emissions is calculated by
applying Equation 4-6 and using the relative uncertainty values.

Because it is not possible to have negative emissions, the lower bound uncertainty will be truncated at zero.
As a result, the lower and upper bound uncertainties are estimated separately.

U (rel )emissions = U (rel ) 2AF + U (rel ) 2HeatingValue + U (rel ) 2EF

For CH 4 : U (rel ) = 152 + 42 + 252 = ±29.4% ( tonnes CH 4 /yr )

For N 2 O: U (rel ) = 152 + 42 + 1502 = +151%, −100% ( tonnes N 2 O/yr )
Uncertainty Estimate of CO2e Emissions from Turbines:

The final step is to convert the emissions of CO2, CH4, and N2O to a CO2e basis. Equation 4-4 is applied
using absolute uncertainty values. Because the N2O uncertainty estimate is asymmetrical, separate upper
and lower bound uncertainties are calculated for the CO2e emissions.

13,900 tonneCO 2 ⎛ 0.905 tonneCH 4 21 tonne CO 2 e ⎞

E CO2e = +⎜ × ⎟
yr ⎝ yr tonneCH 4 ⎠
⎛ 0.325 tonneN 2 O 310 tonne CO 2 e ⎞
+⎜ × ⎟ = 14,100 tonne CO 2 e/yr
⎝ yr tonneN 2 O ⎠

U (abs ) ECO e = (U (abs ) 2ECO + U (abs ) 2ECH + U (abs ) 2EN O

2 2 4 2

U (abs ) ECO e = (0.157 ×13,900) 2 + (0.294 × 21× 0.905) 2 + (1.51× 310 × 0.325)2
2
Upper:
U (abs ) ECO e = 2,190 tonnes CO 2 e/yr
2

2,190 tonnes CO 2 e/yr

U (rel ) ECO e = 100% × = +15.6% ( tonnes CO 2 e/yr )
2
14,100 tonnes CO 2 e/yr

Pilot Version, September 2009 F-10

 U (abs ) ECO e = (U (abs ) 2ECO + U (abs ) 2ECH + U (abs ) 2EN O
2 2 4 2

U (abs ) ECO e = (−0.157 ×13,900) 2 + (−0.294 × 21× 0.905) 2 + (−1.00 × 310 × 0.325) 2
2
Lower:
U (abs ) ECO e = 2,190 tonnes CO 2 e/yr
2

2,190 tonnes CO 2 e/yr

U (rel ) ECO e = 100% × = −15.6% ( tonnes CO 2 e/yr )
2
14,100 tonnes CO 2 e/yr

Uncertainty Estimate of CH4 and N2O Emissions from Diesel-fired Equipment:

For diesel engines, the emission factor is provided on a heat input basis. The equipment ratings provided in
Table F-5 are converted to volume of fuel consumed on a heat input basis using the conversion factor 8,089
Btu/hp-hr (from API Compendium Table 4-2). An uncertainty of 5% is assumed based on engineering
judgment for this conversion factor.

Diesel engine >600 hp:

1,800 hp 8,089 Btu 200 hr 3.7 × 10-6 tonne CH 4

E CH4 = 1 Unit × × × × = 0.0108 tonnes CH 4 /yr
unit hp-hr yr 106 Btu

1,800 hp 8,089 Btu 200 hr 6.01 × 10-7 tonne N 2 O

E N2O = 1 Unit × × × × = 0.00175 tonnes N 2 O/yr
unit hp-hr yr 106 Btu

As discussed earlier, we assume there is no uncertainty in the number of units, the count of diesel engines.
The uncertainty in the unit capacity is 5%, the uncertainty in the hp-hr to Btu conversion factor is 5%, and
the uncertainty in the average hours of operation is 10% based on expert judgment. The uncertainty in the
emissions factors are determined by engineering judgment to be 25% for CH4 and +150%, –100% for N2O.
The uncertainty of the emissions is calculated by applying Equation 4-6 and using the relative uncertainty
values.

Because it is not possible to have negative emissions, the lower bound uncertainty will be truncated at zero.
As a result, the lower and upper bound uncertainties are estimated separately.

U (rel )emissions = U(rel) 2NumberOfUnits +U(rel ) 2UnityCapacity +U(rel ) 2Btu/hp-hr conv +U(rel )OperationHours
2
+U(rel ) 2EF

For CH 4 : U (rel ) = 02 + 52 + 52 + 102 + 252 = ±27.8% ( tonnes CH 4 /yr )

For N 2 0 : U (rel ) = 02 + 52 + 52 + 102 + 1502 = +151%, −100% ( tonnes N 2 O/yr )

Pilot Version, September 2009 F-11

Diesel engine <600 hp (fire water pump):

460 hp 8,089 Btu 24 hr 0.00016 tonne TOC 0.09 tonne CH 4

E CH4 = 1 Unit × × 0.87 × × × ×
unit hp-hr yr 106 Btu tonne TOC
E CH4 = 0.00112 tonnes CH 4 /yr

460 hp 8,089 Btu 24 hr 6.01× 10-7 tonne N 2 O

E N2O = 1 Unit × × 0.87 × × ×
unit hp-hr yr 106 Btu
E N2O = 0.0000467 tonnes N 2 O/yr

As discussed earlier in Table 5-1, we assume there is no uncertainty in the number of units, the count of
diesel engines. The uncertainty in the unit capacity is 5%, the uncertainty in the percent load is 20%, the
uncertainty in the heating value is 5%, and the uncertainty in the average hours of operation is 10% based on
expert judgment. The uncertainty in the emissions factors are determined by engineering judgment to be
25% for CH4 and +150%,-100% for N2O. For CH4, the CH4 content of the TOC is an assumed value, so the
uncertainty with this composition is assigned by engineering judgment to be 100%. The uncertainty of the
emissions is calculated by applying Equation 4-6 and using the relative uncertainty values.

U (rel ) 2NumberOfUnits + U (rel ) 2PercentLoad + U (rel )UnityCapacity

2
+ U (rel ) 2HeatingValue
For CH 4 U (rel )emissions =
+U (rel )OperationHours
2
+ U (rel ) 2EF + U (rel )Weight
2
%

U (rel )emissions = 02 + 202 + 52 + 52 + 102 + 252 + 1002 = +106%, −100% ( tonnes CH 4 /yr )

U (rel ) 2NumberOfUnits + U (rel ) 2PercentLoad + U (rel )UnityCapacity

2
+ U ( rel ) 2HeatingValue
For N 2 O: U (rel )emissions =
+U (rel )OperationHours
2
+ U (rel ) 2EF

U (rel )emissions = 02 + 202 + 52 + 52 + 102 + 1502 = +152%, −100% ( tonnes N 2 O/yr )

Uncertainty Estimate of CO2e Emissions from Diesel Engines:

The final step is to convert the emissions of CO2, CH4, and N2O to a CO2e. Equation 4-4 is applied using
absolute uncertainty values. Both CH4 and N2O uncertainty estimates are asymmetrical, so both upper and
lower bound uncertainties are calculated for the CO2e emissions.

Pilot Version, September 2009 F-12

 219 tonne CO 2 ⎛ 0.0108 tonne CH 4 21 tonne CO 2 e ⎞ ⎛ 0.00112 tonne CH 4 21 tonne CO2 e ⎞
E CO2e = +⎜ × ⎟+⎜ × ⎟
yr ⎝ yr tonne CH 4 ⎠ ⎝ yr tonne CH 4 ⎠
⎛ 0.00175 tonne N 2 O 310 tonne CO 2 e ⎞ ⎛ 0.0000467 tonne N 2 O 310 tonne CO 2 e ⎞
+⎜ × ⎟+⎜ × ⎟
⎝ yr tonne N 2 O ⎠ ⎝ yr tonne N 2 O ⎠
E CO2e = 220 tonne CO 2 e/yr

Upper:

U (abs ) ECO e = (U(abs ) 2ECO +U(abs ) 2ECH +U(abs ) 2ECH +U(abs ) 2E N O-Diesel>600hp +U(abs ) 2E N O-Diesel<600hp
2 2 4 -Diesel>600hp 4 -Diesel<600hp 2 2

(0.156 × 219) 2 + (0.278 × 21× 0.0108) 2 + (1.06 × 21× 0.00112) 2

U (abs ) ECO e =
2
+ (1.51× 310 × 0.00175) 2 + (1.52 × 310 × 0.0.0000467) 2
U (abs ) ECO e = 34.1 tonnes CO 2 e/yr
2

34.1 tonnes CO 2 e/yr

U (rel ) ECO e = 100% × = +15.5% ( tonnes CO 2e/yr )
2
220 tonnes CO 2 e/yr
Lower:

U (abs) ECO e = (U(abs ) 2ECO +U(abs ) 2ECH -Diesel>600hp +U(abs ) 2ECH -Diesel<600hp +U(abs ) 2E N O-Diesel>600hp +U(abs ) 2E N O-Diesel<600hp
2 2 4 4 2 2

(−0.156 × 219) 2 + (−0.278 × 21× 0.0108) 2 + (−1.00 × 21× 0.00112) 2

U (abs) ECO e =
2
+ (−1.00 × 310 × 0.00175) 2 + (−1.00 × 310 × 0.0.0000467) 2
U (abs) ECO e = 34.1 tonnes CO 2e/yr
2

34.1 tonnes CO 2 e/yr

U (rel ) ECO e = 100% × = −15.5% ( tonnes CO 2 e/yr )
2
220 tonnes CO 2e/yr

F.4 Uncertainty in Combustion Sources – Flares

Table F-6 summarizes the operating parameters for the flares.

Table F-6. Operating Parameters for Flares

Annual
Unit Activity Factor
No. of Uncertainty Capacity Uncertainty (all units Uncertainty
Source Fuel Units (±%)a (per unit) (±%)a combined) (±%)a
Emergency Produced
1 0 N/A 500×106 scf/yr ±15% scf/yr
flare Gas

For this example, we assume the flare gas composition is equal to the field gas composition (shown in
Table F-4).

Pilot Version, September 2009 F-13

Uncertainty Estimate of CH4 Emissions from Flares:

500 ×106scf gas lbmole gas 0.80 lbmole CH 4 16.04 lb CH 4

E CH4 = × × ×
yr 379.3 scf gas lbmole gas lbmole CH 4
0.02 lb noncombusted CH 4 tonne
× × = 153 tonnes CH 4 /yr
lb CH 4 2204.62 lb

For CH4, the uncertainty in the activity factor is 15%, and the uncertainty in the noncombusted methane is
20%, both assigned by expert judgment. The uncertainty in the mole % of CH4 is 4%. The uncertainty of
the emissions is calculated by applying Equation 4-6 and using the relative uncertainty values.

U (rel )emissions = U(rel ) 2AF +U(rel ) 2mole% +U(rel ) 2Noncombusted Methane

U (rel )emissions = 152 + 42 + 202 = ±25.3% ( tonnes CH 4 /yr )

Uncertainty Estimate of N2O Emissions from Flares:

N2O emissions from flares are estimated based on the volume of crude produced.
6,100 bbl 365 days 1.0 ×10−4 tonnes N 2 O
E N 2O = × × = 0.223 tonnes N 2 O/yr
day yr 1, 000 bbl

The uncertainty in the activity factor is 5% based on expert judgment. There is no uncertainty in the number
of days of operation. The uncertainty in the emissions factor is +200%,–100% based on expert judgment.
The uncertainty of the emissions is calculated by applying Equation 4-6 and using the relative uncertainty
values.

Because it is not possible to have negative emissions, the lower bound uncertainty will be truncated at zero.
As a result, the lower and upper bound uncertainties are estimated separately.

U (rel )emissions = U(rel) 2AF +U(rel) 2AverageOperations +U(rel) 2EF

U (rel )emissions = 52 + 02 + 2002 = +200%, −100% ( tonnes N 2 O/yr )

Uncertainty Estimate of CO2 Emissions from Flares:

CO2 emissions result from CO2 present in the gas and CO2 formed through combustion.
The flare emissions for CO2 in gas and CO2 formed are correlated, since they both use the same activity
factor. To calculate the uncertainty, first rearrange the CO2 flare emissions as follows to eliminate the
correlation.

Pilot Version, September 2009 F-14

 500 ×106 scf gas lbmole gas 44.01 lb CO 2 tonne
ECO2 = × × × ×
yr 379.3 scf gas lbmole CO 2 2204.62 lb
⎡⎛ 0.80 lbmole CH 4 1 lbmole C 0.042 lbmole C2 H 6 2 lbmole C ⎞⎤
⎢⎜ lbmole gas × lbmole CH + lbmole gas
×
lbmole C2 H 6 ⎟⎥
⎢⎜ 4 ⎟⎥
⎢⎜ 0.013 lbmole C3 H8 3 lbmole C 0.004 lbmole C4 H10 4 lbmole C ⎟ ⎥
⎢⎜ + × + × ⎟⎥
⎢⎝ lbmole gas lbmole C3 H8 lbmole gas lbmole C4 H10 ⎠ ⎥
⎢ 0.98 lbmole CO formed 0.12 lbmole CO ⎥
⎢× 2
+ 2
⎥
⎣ lbmole C combusted lbmole gas ⎦
ECO2 = 27,400 tonnes CO 2 /yr

For the uncertainty aggregation, start first with the uncertainty for the moles of carbon in the flared gas
stream (the terms in parenthesis) by applying Equation 4-4 and using the absolute uncertainties. The
uncertainty in the gas stream composition is 4%.

U (abs )
∑ lbmoleC
= ∑U (abs) 2
mole %

U (abs ) = (0.04 × 0.80 × 1) 2 + (0.04 × 0.042 × 2) 2 + (0.04 × 0.013 × 3) 2 + (0.04 × 0.004 × 4) 2

∑ lbmoleC
U (abs ) = 0.0322
∑ lbmoleC
0.0322
U (rel ) = 100% = 3.43%
∑ lbmoleC 0.939

Then calculate the uncertainty of the product of the composition and the combustion efficiency by applying
Equation 4-4 and using the relative uncertainty.

U (rel )1 = U (rel )composition

2
+ U (rel )CombustionEff
2
= 3.432 + 202 = 20.3%

Next, account for the uncertainty of the CO2 present in the gas by applying Equation 4-4 and using the
absolute uncertainty values. This will aggregate uncertainty for all the terms in the brackets.

U (abs ) 2 = U (abs )12 + U (abs ) 2mole % = (0.203 × 0.939 × 0.98) 2 + (0.04 × 0.120) 2 = 0.187
0.187
U (rel ) 2 = 100% × = 18.0%
( 0.939 × 0.98) + 0.120
Finally the uncertainty of the emissions is calculated by applying Equation 4-6 and using the relative
uncertainty values. For the CO2 that is in the flared gas, the uncertainty in the activity factor is 15% based
on expert judgment.

U (rel )emissions = U (rel ) 2AF + U (rel ) 22 = 152 + 18.02 = 23.4%

Pilot Version, September 2009 F-15

Uncertainty Estimate of CO2e Emissions from Flares:

The final step is to convert the emissions of CO2, CH4, and N2O to a CO2e. Equation 4-4 is applied using
absolute uncertainty values. Because the N2O uncertainty estimate is asymmetrical, separate upper and
lower bound uncertainties are calculated for the CO2e emissions.

27, 400 tonne CO 2 ⎛ 153 tonne CH 4 21 tonne CO 2 e ⎞

E CO2e = +⎜ × ⎟
yr ⎝ yr tonne CH 4 ⎠
⎛ 0.223 tonne N 2 O 310 tonne CO 2 e ⎞
+⎜ × ⎟ = 30, 700 tonne CO 2 e/yr
⎝ yr tonne N 2 O ⎠

Upper:

U (abs ) ECO e = (U(abs )2ECO +U(abs ) 2ECH +U(abs) 2E N O

2 2 4 2

U (abs ) ECO e = (0.234 × 27, 400) + (0.253 × 21×153) 2 + (2.00 × 310 × 0.233) 2 = 6, 460 tonnes CO 2 e/yr
2
2

6,460 tonnes CO 2 e/yr

U (rel )ECO e = 100% × = +21.1% ( tonnes CO 2 e/yr )
2
30, 700 tonnes CO 2 e/yr

Lower:

U (abs ) ECO e = (U(abs )2ECO +U(abs ) 2ECH +U(abs) 2E N O

2 2 4 2

U (abs ) ECO e = (−0.234 × 27, 400)2 + (−0.253 × 21×153)2 + (−1.00 × 310 × 0.233) 2 = 6, 460 tonnes CO2e/yr
2

6,460 tonnes CO2e/yr

U (rel )ECO e = 100% × = −21.1% ( tonnes CO2e/yr )
2
30, 700 tonnes CO2e/yr

F.5 Uncertainty in Combustion Sources – Fleet Vehicles

Table F-7 summarizes the operating parameters for the flares.

Table F-7. Operating Parameters for Fleet Vehicles

No. of Uncertainty Average Operation

Source Fuel Units (±%)a (per unit per year) Uncertainty (±%)a
Fleet
Vehicles Gasoline 5 ±0% units 40,000 mi/yr ea. ±15.0% mi/yr
(Trucks)

The calculations for fleet vehicle emissions are reprinted directly from the API Compendium. Some
uncertainty values were assigned as follows:

• Miles Driven per Vehicle = ±15% (assigned by expert judgment, based upon uncertainties in the
vehicle logs gathered by the company).

Pilot Version, September 2009 F-16

 • Fuel Economy (miles per gallon) = ±5% (assigned by expert judgment, based on some fuel
economy data for similar vehicle types).
• Heat Value of Gasoline = ±5% (determined by expert judgment).
• The number of trucks = ±0% (expert judgment determined that this count is perfectly known and
discreet).
• The CO2 emission factor for this category, expressed in tonne CO2/MMBtu, has a provided
uncertainty of ±10%. The reference source CH4 and N2O emission factors, while small, have
much larger uncertainties of +150%,–100%. Because it is not possible to have negative
emissions, the lower bound uncertainty will be truncated at zero. As a result, the lower and upper
bound uncertainties are estimated separately.
The energy content of the fuel consumed by the vehicles (V) must first be converted to an energy input
basis and is determined based on the annual miles and an average fuel economy.

40,000 mi gal bbl 5.25 ×106 Btu

V= × × ×
yr-truck 14 mi 42 gal bbl
V=1,790 × 106 Btu/yr-truck

Uncertainty Estimate for Fleet Activity Data:

Equation 4-6 is applied using the relative uncertainty values.

2 2 2
U (rel ) Energy of Fuel = U(rel ) AF +U(rel )fuel economy +U(rel ) Heat value of gasoline

U (rel ) Energy of Fuel = 152 + 52 + 52 = ±16.6% ( Btu/yr-truck )

Uncertainty Estimate for CO2 Emissions from Fleet:

Carbon dioxide emissions are then calculated by applying the emission factor for vehicles.

1,790 ×106 Btu 0.0709 tonne CO 2

E CO2 = 5 trucks × ×
yr-truck 1×106 Btu
E CO2 = 127 tonnes CO 2 /yr

Equation 4-6 is applied using the relative uncertainty values.

2 2 2
U (rel )CO2 Emissions = U(rel ) Vehicle +U(rel )Energy of Fuel +U(rel ) EF

U (rel )CO2 Emissions = 02 + 16.62 + 102 = ±19.4% ( tonnes CO 2 /yr )

Uncertainty Estimate for CH4 Emissions from Fleet:

Methane emissions are then calculated by applying the emission factor for vehicles.

Pilot Version, September 2009 F-17

 40,000 mi gal 4.5 ×10-4 tonne CH4 Mgal
E CH4 = 5 trucks × × × ×
yr-truck 14 mi Mgal 1000 gal
E CH4 = 0.00643 tonnes CH4/yr

Equation 4-6 is applied using the relative uncertainty values.

2 2 2
U (rel )CH 4Emissions = U(rel )Vehicle +U(rel ) Fuel Economy +U(rel ) EF

U (rel )CH 4Emissions = 02 + 52 + 1502 = +151%, −100% ( tonnes CH 4 /yr )

Uncertainty Estimate for N2O Emissions from Fleet:

Nitrous oxide emissions are then calculated by applying the emission factor for vehicles.

40,000 mi gal 6.1×10-4 tonne CH4 Mgal

E N2O = 5 trucks × × × ×
yr-truck 14 mi Mgal 1000 gal
E N 2O = 0.00871 tonnes N2O/yr

Equation 4-6 is applied using the relative uncertainty values.

2
U (rel ) N2OEmissions = U(rel ) Vehicle +U(rel ) 2Fuel Economy +U(rel ) 2EF

U (rel ) N2OEmissions = 02 + 52 + 1502 = +151%, −100% ( tonnes N 2 O/yr )

Uncertainty Estimate of CO2e Emissions from Fleet:

The final step is to convert the emissions of CO2, CH4, and N2O to a CO2e. Equation 4-4 is applied using
absolute uncertainty values. Both CH4 and N2O uncertainty estimates are asymmetrical, so both upper and
lower bound uncertainties are calculated for the CO2e emissions.

127 tonne CO 2 ⎛ 0.00643 tonne CH 4 21 tonne CO 2 e ⎞

E CO2e = +⎜ × ⎟
yr ⎝ yr tonne CH 4 ⎠
⎛ 0.00871 tonne N 2 O 310 tonne CO 2 e ⎞
+⎜ × ⎟ =129 tonne CO 2 e/yr
⎝ yr tonne N2O ⎠

Pilot Version, September 2009 F-18

 Upper:
U (abs ) ECO e = (U(abs ) 2ECO +U(abs) 2ECH +U(abs ) 2E N O
2 2 4 2

U (abs ) ECO e = (0.194 ×127) + (1.51× 21× 0.00643) 2 + (1.51× 310 × 0.00871)2 = 24.8 tonnes CO 2 e/yr
2
2

24.8 tonnes CO 2 e/yr

U (rel )ECO e = 100% × = +19.2% ( tonnes CO 2 e/yr )
2
129 tonnes CO 2 e/yr
Lower:
U (abs ) ECO e = (U(abs ) 2ECO +U(abs ) 2ECH +U(abs ) 2E N O
2 2 4 2

U (abs ) ECO e = (−0.194 × 127) + (−1.00 × 21× 0.00643)2 + (−1.00 × 310 × 0.00871) 2 = 24.7 tonnes CO2 e/yr
2
2

24.7 tonnes CO 2 e/yr

U (rel ) ECO e = 100% × = −19.1% ( tonnes CO 2 e/yr )
2
129 tonnes CO 2 e/yr

F.6 Uncertainty in Vented Sources – Gas Dehydration

Table F-8 provides the operating parameters for gas dehydration at the example facility.

Table F-8. Operating Parameters for Dehydration

Annual
Average Activity
No. Unit Operation Factor (all
of Uncertainty Capacity Uncertainty (per unit Uncertainty units Uncertainty
Source Fuel Units (±%)a (per unit) (±%)a per year) (±%)a combined) (±%)a
Dehydration
vents (also
Produced 30×106 343 10,290×106 ±5.39%
has Kimray 1 ±0% units ±5% scf/day ±2% days/yr
Gas scf/day days/yr scf/yr scf/yr
pump
emissions)

Uncertainty Estimate for CH4 Emissions from Dehydration:

Methane emissions from the gas dehydrator vents and for Kimray pumps are estimated using the production
segment emission factors provided in the API Compendium. For both of these sources, emissions are based
on the quantity of gas processed. Carbon dioxide emissions are estimated using the facility CO2 and default
CH4 concentrations in the gas.

EFCH 4 = ( EFDehydrator vents ) + ( EFKimray pumps )

⎛ 0.0052859 tonnes CH 4 ⎞ ⎛ 0.01903 tonnes CH 4 ⎞
EFCH4 = ⎜ ⎟+⎜ ⎟
⎝ 106 scf ⎠ ⎝ 106 scf ⎠
EFCH4 = 0.024326 tonnes CH 4 / 10 scf
6

For the uncertainty estimate, the emission factors should be combined to eliminate the correlation associated
with the common activity data. The uncertainty of the emissions factor is calculated by applying

Pilot Version, September 2009 F-19

Equation 4-4 and using the absolute uncertainty values. The uncertainty in the emissions factor for the
dehydrator is 191% from Table 5-2 of the API Compendium and 82.8% for the Kimray pump based on
Table 5-4 of the API Compendium.

U (abs ) Emission factors = (1.91×0.0052859) 2 +(0.828×0.01903) 2 =0.0187 tonnes CH 4 /106scf

0.0187 tonnes CH 4 /106scf
U (rel ) Emission factors =100%× 6
= ± 77.0% ( tonnes CH 4 /106scf )
0.024326 tonnes CH 4 /10 scf

30 × 106 scf 343 day 0.80 tonne mole CH 4 (facility) 0.024326 tonne CH 4
E CH4 = × × ×
day gas processed yr 0.788 tonne mole CH 4 (default) 106scf

E CH4 = 254 tonnes CH 4 /yr

The uncertainty of the emissions can be calculated by applying Equation 4-6 and using the relative
uncertainty values. As noted in Table 5-1, there is no uncertainty associated with the count of equipment.
The uncertainty associated with the volume of gas is 5% and the uncertainty associated with the days of
operation is 2%. The uncertainty in the CH4 content for the facility gas is 4%. The uncertainty for the
default CH4 content is 5.53% from the GRI/EPA Methane Study (Shires and Harrison, Volume 6,
Tables A-1, A-2, A-3, 1996), which is provided in Table E-4 of the API Compendium.

U(rel ) 2Dehydrator units +U(rel )Gas

2 2 2
processed +U(rel ) Days +U(rel ) Facility mole%
U (rel )emissions =
+U(rel ) 2Default mole% +U(rel ) 2Emission factors

U (rel )emissions = 02 + 52 + 22 + 42 + 5.532 + 77.02 = ±77.5% ( tonnes CH 4 /yr )

Uncertainty Estimate for CO2 Emissions from Dehydration:

30 ×106 scf 343 day 0.024326 tonne CH 4 tonne mole CH 4 0.80 tonne mole CH 4 (facility)
E CO2 = × × 6
× ×
day gas processed yr 10 scf 16.04 tonne CH 4 0.788 tonne mole CH 4 (default)
tonne mole gas 0.12 tonne mole CO 2 44.01 tonne CO 2
× × ×
0.8 tonne mole CH 4 tonne mole gas tonne mole CO 2

E CO2 =105 tonnes CO 2 /yr

The uncertainty of the CO2 emissions is calculated by applying Equation 4-6 and using the relative
uncertainty values. As noted in Table 5-1, there is no uncertainty associated with the count of equipment.
The uncertainty associated with the volume of gas is 5% and the uncertainty associated with the days of
operation is 2%. The uncertainty in the CO2 content for the facility gas is 4%. The uncertainty for the

Pilot Version, September 2009 F-20

 default CH4 content is 5.53% from the GRI/EPA Methane Study (Shires and Harrison, 1996), which is also
provided in Table E-4 of the API Compendium. This equates to the same result as shown above for CH4.
Note, the uncertainties for the CH4 and CO2 emissions would be different if different uncertainty values
were associated with the respective molar compositions.

U(rel ) 2Dehydrator units +U(rel )Gas

2 2 2
processed +U(rel ) Days +U(rel ) Facility mole%
U (rel )emissions =
+U(rel ) 2Default mole% +U(rel ) 2Emission factors

U (rel )emissions = 02 + 52 + 22 + 42 + 5.532 + 77.02 = ±77.5% ( tonnes CO 2 /yr )

Uncertainty Estimate of CO2e Emissions from Dehydration:

106 tonne CO 2 ⎛ 254 tonne CH 4 21 tonne CO 2 e ⎞

E CO2e = +⎜ × ⎟ = 5, 440 tonne CO 2 e/yr
yr ⎝ yr tonne CH4 ⎠

The uncertainty of the CO2e emissions is calculated by applying Equation 4-4, using the absolute
uncertainty values.

U (abs ) ECO e = (U(abs ) 2ECO +U(abs) 2ECH

2 2 4

U (abs ) ECO e = (0.775 ×106) + (0.775 × 21× 254) 2 = 4,130 tonnes CO 2e/yr
2
2

4,130 tonnes CO 2 e/yr

U (rel ) ECO e = 100% × = ±76.0% ( tonnes CO 2 e/yr )
2
5, 440 tonnes CO 2 e/yr

F.7 Uncertainty in Vented Sources – Storage Tanks

Table F-9 summarizes the operating parameters for storage tanks at the example facility.

Table F-9. Operating Parameters for Storage Tanks

Annual
Average Activity
No. Unit Operation Factor
of Uncertainty Capacity Uncertainty (per unit Uncertainty (all units Uncertainty
Source Fuel Units (±%)a (per unit) (±%)a per year) (±%)a combined) (±%)a
Central
6,100 343 2,092,300 ±5.39%
tank Crude Oil 1 ±0% units ±5% bbl/day ±2% days/yr
bbl/day days/yr bbl/yr bbl/yr
battery
Storage
Chemical 1 ±0% units N/A N/A N/A
tanks
Naphtha 1 ±0% units N/A N/A N/A
Glycol 1 ±0% units N/A N/A N/A

Pilot Version, September 2009 F-21

Uncertainty Estimate for CH4 Emissions from Storage Tanks:

Methane emissions from flashing losses are estimated using the simple emission factor provided in the API
Compendium. (Note that this simple emission factor approach is used in absence of the more detailed
information necessary for the other calculation approaches provided in the API Compendium).

6,100 bbl 343 day 8.86 × 10-4 tonne CH 4 ⎛ 0.80 mole CH 4 (facility) ⎞
E CH 4 = × × ×⎜ ⎟
day yr bbl ⎝ 0.788 mole CH 4 (default EF) ⎠
E CH 4 = 1,880 tonnes CH 4 /yr

As noted in Table 5-1, there is no uncertainty associated with the count of equipment. The uncertainty
associated with the volume of gas is 5% and the uncertainty associated with the days of operation is 2%.
The uncertainty in the emissions factor is 90% based on expert judgment and the uncertainty in the site
specific CH4 mole % is 4 % based on Exhibit 5-1. The uncertainty in the CH4 content is 5.53% from the
GRI/EPA Methane Study (Shires and Harrison), which is provided in Table E-4 of the API Compendium.
The uncertainty of the emissions is calculated by applying Equation 4-6 and using the relative uncertainty
values.

U(rel ) 2Number of Units +U(rel )Gas

2 2 2 2
processed +U(rel ) Days +U(rel ) EF +U(rel ) Facility mole%
U (rel )emissions =
+U(rel ) 2Default mole%

U (rel )emissions = 02 + 52 + 22 + 902 + 42 + 5.532 = ±90.4% ( tonnes CH 4 /yr )

Uncertainty Estimate for CO2 Emissions from Storage Tanks:

6,100 bbl 343 day 8.86 × 10-4 tonne CH 4 tonne mole CH 4 0.80 mole CH 4 (facility)
E CO2 = × × × ×
day yr bbl 16.04 tonne CH 4 0.788 mole CH 4 (default EF)
tonne mole gas 0.12 tonne mole CO 2 44.01 tonne CO 2
× × × = 775 tonnes CO 2 /yr
0.80 tonne mole CH 4 tonne mole gas tonne mole CO 2

The same approach applies to estimating the uncertainty for the CO2 emissions. The uncertainty associated
with the mole% of CO2 is 4%, so the results are the same as for CH4.

U(rel ) 2Number of Units +U(rel )Gas

2 2 2 2
processed +U(rel ) Days +U(rel ) EF +U(rel ) Facility mole%
U (rel )emissions =
+U(rel ) 2Default mole%

U (rel )emissions = 02 + 52 + 22 + 902 + 42 + 5.532 = ±90.4% ( tonnes CO2 /yr )

Pilot Version, September 2009 F-22

 Uncertainty Estimate of CO2e Emissions from Storage Tanks:

775 tonneCO 2 ⎛ 1,880 tonne CH 4 21 tonne CO 2 e ⎞

E CO2e = +⎜ × ⎟ = 40,300 tonne CO 2 e/yr
yr ⎝ yr tonne CH 4 ⎠

U (abs ) ECO e = (U(abs ) 2ECO +U(abs ) 2ECH

2 2 4

U (abs ) ECO e = (0.775 × 775) 2 + (0.775 × 21×1,880) 2 = 35, 700 tonnes CO 2 e/yr
2

35,700 tonnes CO2 e/yr

U (rel ) ECO e = 100% × = ±88.7% ( tonnes CO 2 e/yr )
2
40,300 tonnes CO 2 e/yr

F.8 Uncertainty in Vented Sources – Acid Gas Removal (Amine Unit Emissions)
Table F-10 summarizes the operating parameters for acid gas removal at the example facility.

Table F-10. Operating Parameters for Storage Tanks

Annual
Average Activity
Operation Factor (all
Unit Capacity Uncertainty (per unit per Uncertainty units Uncertainty
Source Fuel (per unit) (±%)a year) (±%)a combined) (±%)a
6
Inlet Produced Gas 30×10 ±5% scf/day 343 days/yr ±2% days/yr 10,290×106 ±5.39%
scf/day scf/yr scf/yr
Outlet Outlet Gas N/A N/A 8,997×106 ±5% scf/yr
(0.5 mole% CO2) scf/yr

Uncertainty Estimate for CH4 Emissions from Acid Gas Removal:

Methane emissions from the amine unit are estimated based on the emission factor provided in the API
Compendium and the quantity of gas processed.

30 × 106 scf 343 day 0.0185 tonne CH 4 0.80 tonne mole CH 4 (facility)
E CH4 = × × ×
day processed yr 106scf 0.788 tonne mole CH 4 (default)
E CH4 = 193 tonnes CH 4 /yr

We assume there is no uncertainty in the number of units. The uncertainty in the daily volume of gas
processed is 5% and the uncertainty in the days of operation is 2% based on expert judgment. The
uncertainty in the emission factor is 119% based on Table 5-5 of the API Compendium. Because it is not
possible to have negative emissions, the lower bound uncertainty will be truncated at zero. As a result, the
lower and upper bound uncertainties are estimated separately.

The uncertainty in the facility mole% is 4%, the uncertainty in the default mole% is 5.53% from the
GRI/EPA Methane Study (Shires and Harrison, 1996), which is also provided in Table E-4 of the API

Pilot Version, September 2009 F-23

Compendium. The uncertainty of the activity factor is calculated by applying Equation 4-6 and using the
relative uncertainties.

Upper:
U(rel ) 2Number of Units +U(rel ) Daily
2 2 2 2
Volume +U(rel ) Days of Operation +U(rel ) EF +U(rel ) Facility mole%
U (rel )emissions =
+U(rel ) 2Default mole%

U (rel )emissions = 02 + 52 + 22 + 1192 + 42 + 5.532 = +119% ( tonnes CH 4 /yr )

Lower:
U(rel ) 2Number of Units +U(rel ) Daily
2 2 2 2
Volume +U(rel ) Days of Operation +U(rel ) EF +U(rel ) Facility mole%
U (rel )emissions =
+U(rel ) 2Default mole%

U (rel )emissions = 02 + ( −5 ) + ( −2 ) + ( −100 ) + ( −4 ) + ( −5.53) = −100% ( tonnes CH 4 /yr )

2 2 2 2 2

Uncertainty Estimate for CO2 Emissions from Acid Gas Removal:

CO2 emissions are calculated based on a material balance approach.

⎡ ⎛ 30 ×106 scf 343 day ⎞ ⎛ 8,997×106 scf ⎞ ⎤

E CO2 =⎢ ⎜ × ×12 mole% CO 2 ⎟ - ⎜ ×0.5 mole% CO 2 ⎟ ⎥
⎣⎢ ⎝ day processed yr ⎠sour ⎝ yr ⎠sweet ⎦⎥
lbmole lb tonne
× × 44 ×
379.3 scf lbmole 2204.62 lb

scf mole% CO 2 lbmole 44.01 lb tonne

E CO2 = [1,234.8 − 44.985] ×106 × × ×
yr 379.3 scf lbmole 2204.62 lb
E CO2 = 62, 620 tonne CO 2 /yr

To calculate the uncertainty for the CO2 emissions, first calculate the uncertainty of the sour and sweet gas
components and then aggregate those uncertainties. For the uncertainty in the sour gas flow rate we assume
there is no uncertainty in the number of units, the uncertainty in the daily volume of gas processed is 5%,
and the uncertainty in the days of operation is 2% based on expert judgment. The uncertainty in the sweet
gas flow rate scf/yr is 5% based on expert judgment. The uncertainty in the CO2 concentrations for both the
sour and sweet gases are assumed to be 4%. The uncertainties of the first and second terms of the emissions
equation are calculated by applying Equation 4-6 and using the relative uncertainty values.

U (rel )1 = U (rel )Sour = U(rel ) 2Number of Units +U(rel ) Daily

2 2 2
Volume +U(rel ) Days of Operation +U(rel ) %Mole

U (rel )Sour = 02 + 52 + 22 + 42 = ±6.71% ( scf mole% CO2 /yr )

Pilot Version, September 2009 F-24

U (rel ) 2 = U (rel ) Sweet = U (rel ) 2Flow + U (rel )%2 Mole = 52 + 42 = ±6.40% ( scf mole% CO 2 /yr )

The aggregated uncertainty of the emissions is then calculated by applying Equation 4-4 and using the
absolute uncertainty values.

U (abs )1 − U (abs ) 2 = U (abs )12 + U (abs ) 22

U (abs )1 − U (abs ) 2 = (0.0671× 1, 234.8 ×106 ) 2 + (0.0640 × 44.985 ×106 )2 = 82.9×106 scf CO 2
82.9 ×106 scf CO 2
U (rel ) = = ±6.97% ( scf CO 2 )
1,189.8 ×106 scf CO 2

Uncertainty Estimate of CO2e Emissions from Acid Gas Removal:

⎛ 190 tonne CH 4 21 tonne CO 2 e ⎞

E CO2e = 62,600 tonne CO 2 /yr + ⎜ × ⎟
⎝ yr tonne CH 4 ⎠

E CO2e = 66, 700 tonne CO 2 e/yr

U (abs ) ECO e = (U (abs ) 2ECO + U (abs ) 2ECH

2 2 4

Upper: U (abs ) = (0.0697 × 62, 600) + (1.19 × 21× 190) 2 = 6,510 tonnes CO 2e/yr
2
E CO2e

6,510 tonnes CO 2 e/yr

U (rel ) ECO e = 100% × = +9.77% ( tonnes CO 2 e/yr )
2
66, 700 tonnes CO 2e/yr

U(abs ) ECO e = U (abs ) 2ECO + U (abs ) 2ECH

2 2 4

( −0.0697 × 62, 600 ) + ( −1.00 × 21×190 )

2 2
Lower: U(abs ) E = = 5,960 tonnes CO 2e/yr
CO2 e

5,960 tonnes CO 2 e/yr

U(rel ) ECO e = 100% × = −8.94% ( tonnes CO 2e/yr )
2
66,700 tonnes CO 2 e/yr

F.9 Uncertainty in Vented Sources – Pneumatics Devices and Chemical Injection Pumps

Table F-11 summarizes the operating parameters for acid gas removal at the example facility.

Table F-11. Operating Parameters for Pneumatic Devices and Chemical Injection Pumps

Annual Activity Factor Uncertainty

Source Fuel (all units combined) (±%)a
Pneumatic devices Produced Gas 64 pneumatic devices ±5% devices
Chemical injection pumps (CIPs) Produced Gas 67 CIPs ±5% pumps

Pilot Version, September 2009 F-25

The pneumatic devices and chemical injection pumps (CIPs) at the facility are actuated by natural gas.
Methane emissions from pneumatic device and CIP vents are estimated using CH4 emission factors
presented in API Compendium Tables 5-15 and 5-16, respectively. The type of pneumatic device and CIP
are not specified, so the “Production Average” device and the "Average Pump" emission factor are used.
Carbon dioxide emissions from pneumatic devices and chemical injection pumps are estimated using the
ratio of CO2 to CH4 in the gas.

Uncertainty Estimate for CH4 and CO2 Emissions from Pneumatic Devices:

2.415 tonne CH 4 ⎛ 0.80 tonne mole CH 4 (facility) ⎞

E CH4 = (64 pneumatic devices) × ×⎜ ⎟
device - yr ⎝ 0.788 tonne mole CH 4 (default EF) ⎠
E CH4 = 157 tonnes CH 4 /yr

tonne mole CH 4 ⎛ 0.80 tonne mole CH 4 (facility) ⎞

2.415 tonne CH 4
E CO2 = (64 pneumatic devices) × ×⎜ × ⎟
device - yr
16.04 tonne CH 4 ⎝ 0.788 tonne mole CH 4 (default) ⎠
tonne mole gas 0.12 tonne mole CO 2 44.01 tonne CO 2
× × × = 64.6 tonnes CO 2 /yr
0.8 tonne mole CH 4 tonne mole gas tonne mole CO 2

64.6 tonne CO 2 ⎛ 157 tonne CH 4 21 tonne CO 2 e ⎞

E CO2e = +⎜ × ⎟ = 3,360 tonne CO 2 e/yr
yr ⎝ yr tonne CH 4 ⎠

A 5% uncertainty was assigned to the count of pneumatic devices based on expert judgment. The
uncertainty in the emission factor is 49.5% per Table 5-15 of the API Compendium. The uncertainty in the
default methane content is 5.53% from the GRI/EPA Methane Study (Shires and Harrison), which is also
provided in Table E-4 of the API Compendium. The uncertainty in the site specific mole % is 4%. The
uncertainty of both the CO2 and CH4 emissions are calculated by applying Equation 4-6 and using the
relative uncertainty values.

U (rel ) ECO ,E CH4 = U(rel ) 2AF +U(rel ) 2EF +U(rel )3Facility mole% +U(rel )3Default mole%
2

U (rel ) ECO ,E CH4 = 52 + 49.52 + 42 + 5.532 = ±50.2% ( tonnes/yr )

2

Uncertainty Estimate of CO2e Emissions from Pneumatic Devices:

U (abs ) ECO e = (U (abs ) 2ECO + U (abs ) 2ECH

2 2 4

U (abs ) ECO e = (0.502 × 64.6) 2 + (0.502 × 21×157) 2 = 1, 650 tonnes CO 2e/yr

2

1,650 tonnes CO 2 e/yr

U (rel ) ECO e = 100% × = ±49.2% ( tonnes CO 2 e/yr )
2
3,360 tonnes CO 2e/yr

Pilot Version, September 2009 F-26

Uncertainty Estimate for CH4 and CO2 Emissions from Chemical Injection Pumps:

1.737 tonne CH 4 ⎛ 0.80 tonne mole CH 4 (facility) ⎞

E CH4 = (67 CIPs) × ×⎜ ⎟ = 118 tonnes CH 4 /yr
CIP - yr ⎝ 0.788 tonne mole CH 4 (default EF) ⎠
1.737 tonne CH 4 tonne mole CH 4 0.80 tonne mole CH 4 (facility)
E CO2 = (67 CIPs) × × ×
CIP - yr 16.04 tonne CH 4 0.788 tonne mole CH 4 (default)
tonne mole gas 0.12 tonne mole CO 2 44.01 tonne CO 2
× × × = 48.6 tonnes CO2 /yr
0.8 tonne mole CH 4 tonne mole gas tonne mole CO2

48.6 tonne CO 2 ⎛ 118 tonne CH 4 21 tonne CO2 e ⎞

E CO∂ e = +⎜ × ⎟ = 2,530 tonne CO 2 e/yr
yr ⎝ yr tonne CH 4 ⎠
A 5% uncertainty was assigned to the count of chemical injection pumps based on expert judgment. The
uncertainty in the emissions factor is 108% per Table 5-16 of the API Compendium. Because it is not
possible to have negative emissions, the lower bound uncertainty will be truncated at zero. As a result, the
lower and upper bound uncertainties are estimated separately.

The uncertainty in the default methane content is 5.53% from the GRI/EPA Methane Study (Shires and
Harrison), which is provided in Table E-4 of the API Compendium. The uncertainty in the site specific
mole % is 4%. The uncertainty of both the CO2 and CH4 emissions are calculated by applying Equation 4-6
and using the relative uncertainty values.

= U(rel ) 2AF +U(rel ) 2EF +U ( rel )Facility mole% +U ( rel )Default mole%
2 2
U (rel ) ECO ,E CH4
Upper: 2

U (rel ) ECO ,E CH4 = 5.002 + 1082 + 4.002 + 5.532 = +108% ( tonnes/yr )

2

U(rel) ECO & E CH4 = U(rel ) 2AF + U(rel ) 2EF + U(rel ) 2Facility mole% + U(rel )Default
2
mole%
2
Lower:
( −5.00 ) + ( −100 ) + ( −4.00 ) + ( −5.52 ) = −100% ( tonnes/yr )
2 2 2 2
U(rel)ECO & E CH4 =
2

Uncertainty Estimate of CO2e Emissions from Chemical Injection Pumps:

U (abs ) ECO e = U (abs ) 2ECO +U (abs) 2ECH

2 2 4

Upper: U (abs ) = (1.08 × 48.6) 2 + (1.08 × 21×118) 2 = 2, 690 tonnes CO 2e/yr

ECO2e

2,690 tonnes CO 2 e/yr

U (rel ) ECO e = 100% × = +106% ( tonnes CO 2 e/yr )
2
2,530 tonnes CO 2e/yr

Pilot Version, September 2009 F-27

 U(abs ) ECO e = U (abs ) 2ECO + U (abs ) 2ECH
2 2 4

( −1.00 × 48.6 ) + ( −1.00 × 21×118)

2 2
Lower: U(abs ) ECO e = = 2, 480 tonnes CO 2 e/yr
2

2, 480 tonnes CO 2 e/yr

U(rel ) ECO e = 100% × = −98.1% ( tonnes CO 2e/yr )
2
2,530 tonnes CO 2e/yr

F.10 Uncertainty in Vented Sources – Maintenance/Turnaround Emissions and Pressure

Relief Valves

Table F-12 summarizes the equipment parameters for PRVs and other equipment contributing to
maintenance/turnaround emissions at the example facility.

Table F-12. Operating Parameters for Maintenance/Turnaround Emissions and PRVs

Annual Activity
Factor (all units
Source Fuel combined) Uncertainty (±%)a
Vessel blowdowns (non-routine) Produced Gas 112 vessels ±0% vessels
Compressor starts (non-routine) Produced Gas 11 compressors ±0% compressors
Compressor blowdowns (non-routine) Produced Gas 11 compressors ±0% compressors
Well workovers (non-routine) Produced Gas 24 well workovers ±0% well workovers
Pressure relief valves Produced Gas 482 PRVs ±1% PRVs

Methane emissions from vessel blowdowns, compressor starts, compressor blowdowns, and oil well
workovers are estimated using the emission factors presented in API Compendium Table 5-23. Methane
emissions from PRVs are estimated using an emission factor from API Compendium Table 5-24. Carbon
dioxide emissions are estimated using the ratio of the facility gas CO2 to the default composition.

There is no uncertainty in the count of the major equipment (vessels, compressors, and wells). An
uncertainty of 1% was assigned to the count of PRVs based on expert judgment. The uncertainty in the
emissions factor for vessel blowdowns is 326% based on Table 5-23 of the API Compendium. The
uncertainty in the emissions factor for PRVs is 310% based on Table 5-24 of the Compendium. The
uncertainty in the emissions factor for compressor starts is 190% based on Table 5-23 of the API
Compendium. The uncertainty in the emissions factor of oil well workovers is 300% based on expert
judgment. The uncertainty in the emission factor of compressor blowdowns is 179% based on Table 5-23 of
the API Compendium. Because it is not possible to have negative emissions, the lower bound uncertainties
will be truncated at zero. As a result, the lower and upper bound uncertainties are estimated separately.

The uncertainty in the default methane content is 5.53% based on the GRI/EPA Methane Study (Shires and
Harrison, 1996), which is also provided in Table E-4 of the API Compendium. The uncertainty in the site

Pilot Version, September 2009 F-28

specific mole % for CH4 and CO2 is 4%. The uncertainty of both the CO2 and CH4 emissions are calculated
by applying Equation 4-6 and using the relative uncertainty values.

Uncertainty Estimate for CH4 and CO2 Emissions from Vessel Blowdowns:

0.0015 tonne CH 4 ⎛ 0.80 tonne mole CH 4 (facility) ⎞

E CH4 = (112 vessels) × ×⎜ ⎟
vessel - yr ⎝ 0.788 tonne mole CH 4 (default EF) ⎠
E CH4 = 0.171 tonnes CH 4 /yr

0.0015 tonne CH 4 tonne mole CH 4 0.80 tonne mole CH 4 (facility)

E CO2 = (112 vessels) × × ×
vessel - yr 16.04 tonne CH 4 0.788 tonne mole CH 4 (default)
tonne mole gas 0.12 tonne mole CO 2 44.01 tonne CO 2
× × × = 0.0702 tonnes CO 2 /yr
0.8 tonne mole CH 4 tonne mole gas tonne mole CO 2
0.0702 tonne CO 2 ⎛ 0.171 tonne CH 4 21 tonne CO 2 e ⎞
E CO2e = +⎜ × ⎟ = 3.65 tonne CO 2 e/yr
yr ⎝ yr tonne CH 4 ⎠

U (rel ) ECO ,E CH4 = U(rel ) 2AF +U(rel ) 2EF +U(rel ) 2Facility mole% +U(rel ) 2Default mole%
2

U (rel ) ECO ,E CH4 = 02 + 3262 + 42 + 5.532 = +326%, −100% ( tonnes/yr )

2

Because the emissions cannot be less than zero, the lower bound uncertainty value for this emission source
is truncated at 100%. Therefore, separate upper and lower bound uncertainties are estimated for the CO2e
emissions.

U (abs ) ECO e = (U (abs ) 2ECO + U (abs) 2ECH

2 2 4

Upper: U (abs ) = (3.26 × 0.0702) + (3.26 × 21× 0.171) 2 = 11.7 tonnes CO 2 e/yr
2
ECO2e

11.7 tonnes CO 2 e/yr

U (rel ) ECO e = 100% × = +319% ( tonnes CO 2 e/yr )
2
3.65 tonnes CO 2 e/yr

U(abs ) ECO e = U (abs ) 2ECO + U (abs ) 2ECH

2 2 4

( −1.00 × 0.0702 ) + ( −1.00 × 21× 0.171)

2 2
Lower: U(abs ) ECO e = = 3.58 tonnes CO 2 e/yr
2

3.58 tonnes CO 2 e/y

U(rel ) ECO e = 100% × = −98.1% ( tonnes CO 2 e/yr )
2
3.65 tonnes CO 2 e/y

Pilot Version, September 2009 F-29

Uncertainty Estimate for CH4 and CO2 Emissions from Compressor Starts:

0.1620 tonne CH 4 ⎛ 0.80 tonne mole CH 4 (facility) ⎞

E CH4 = (11 compressors) × ×⎜ ⎟
compressor - yr ⎝ 0.788 tonne mole CH 4 (default EF) ⎠
E CH4 = 1.81 tonnes CH 4 /yr

0.1620 tonne CH 4 tonne mole CH 4 0.80 tonne mole CH 4 (facility)

E CO2 = (11 compressors) × × ×
compressor - yr 16.04 tonne CH 4 0.788 tonne mole CH 4 (default)
tonne mole gas 0.12 tonne mole CO 2 44.01 tonne CO 2
× × × = 0.745 tonnes CO 2 /yr
0.8 tonne mole CH 4 tonne mole gas tonne mole CO 2
0.745 tonne CO 2 ⎛ 1.81 tonne CH 4 21 tonne CO 2 e ⎞
E CO2e = +⎜ × ⎟ = 38.7 tonne CO 2 e/yr
yr ⎝ yr tonne CH 4 ⎠

U (rel ) ECO , ECH 4 = U ( rel ) 2AF + U (rel ) 2EF + U (rel ) 2Facility mole % + U Default
2
mole %
2

U (rel ) ECO , ECH 4 = 02 + 1902 + 42 + 5.532 = +190%, −100% ( tonnes/yr )

2

Because the emissions cannot be less than zero, the lower bound uncertainty value for this emission source
is truncated at 100%. Therefore, separate upper and lower bound uncertainties are estimated for the CO2e
emissions.

U (abs)ECO e = (U(abs)2ECO +U(abs)2ECH

2 2 4

Upper: U (abs) = (1.90 × 0.745) + (1.90 × 21×1.81)2 = 72.3 tonnes CO2 e/yr
2
ECO2e

72.3 tonnes CO2e/yr

U (rel )ECO e = 100% × = +187% ( tonnes CO2e/yr )
2 38.7 tonnes CO2 e/yr

U(abs ) ECO e = U (abs ) 2ECO + U (abs ) 2ECH

2 2 4

Lower: U(abs ) E ( −1.00 × 0.745) + ( −1.00 × 21×1.81)

2 2
CO2 e
= = 38.0 tonnes CO 2 e/yr
38.0 tonnes CO 2 e/y
U(rel ) ECO e = 100% × = −98.1% ( tonnes CO 2 e/yr )
2
38.7 tonnes CO 2 e/y
Uncertainty Estimate for CH4 and CO2 Emissions from Compressor Blowdowns:

0.07239 tonne CH 4 ⎛ 0.80 tonne mole CH 4 (facility) ⎞

E CH4 = (11 compressors) × ×⎜ ⎟
compressor - yr ⎝ 0.788 tonne mole CH 4 (default EF) ⎠
E CH4 = 0.808 tonnes CH 4 /yr

Pilot Version, September 2009 F-30

 0.07239 tonne CH 4 tonne mole CH 4 0.80 tonne mole CH 4 (facility)
E CO2 = (11 compressors) × × ×
compressor - yr 16.04 tonne CH 4 0.788 tonne mole CH 4 (default)
tonne mole gas 0.12 tonne mole CO 2 44.01 tonne CO 2
× × × = 0.333 tonnes CO 2 /yr
0.8 tonne mole CH 4 tonne mole gas tonne mole CO 2

0.333 tonne CO 2 ⎛ 0.808 tonne CH 4 21 tonne CO 2 e ⎞

E CO2e = +⎜ × ⎟ = 17.3 tonne CO 2 e/yr
yr ⎝ yr tonne CH 4 ⎠

U (rel ) ECO ,E CH4 = U(rel ) 2AF +U(rel ) 2EF +U(rel ) 2Facility mole% +U(rel ) 2Default mole%
2

U (rel ) ECO ,E CH4 = 02 + 1792 + 42 + 5.532 = +179%, −100% ( tonnes/yr )

2

Because the emissions cannot be less than zero, the lower bound uncertainty value for this emission source
is truncated at 100%. Therefore, separate upper and lower bound uncertainties are estimated for the CO2e
emissions.

U (abs ) ECO e = (U (abs ) 2ECO + U (abs ) 2ECH

2 2 4

Upper: U (abs ) = (1.79 × 0.333) 2 + (1.79 × 21× 0.808) 2 = 30.4 tonnes CO 2 e/yr
ECO2e

30.4 tonnes CO 2e/yr

U (rel ) ECO e = 100% × = +175% ( tonnes CO 2 e/y )
2
17.3 tonnes CO 2 e/yr

U(abs ) ECO e = U (abs ) 2ECO + U ( abs ) 2ECH

2 2 4

( −1.00 × 0.333) + ( −1.00 × 21× 0.808)

2 2
Lower: U(abs ) E = = 17.0 tonnes CO 2 e/yr
CO2 e

17.0 tonnes CO 2 e/y

U(rel ) ECO e = 100% × = −98.1% ( tonnes CO 2 e/yr )
2
17.3 tonnes CO 2 e/y

Uncertainty Estimate for CH4 and CO2 Emissions from Oil Well Workovers:

24 well workovers 0.0018 tonne CH 4 ⎛ 0.80 tonne mole CH 4 (facility) ⎞

E CH4 = × ×⎜ ⎟
yr workovers ⎝ 0.788 tonne mole CH 4 (default EF) ⎠
E CH4 = 0.0439 tonnes CH 4 /yr
24 well workovers 0.0018 tonne CH 4 tonne mole CH 4 0.80 tonne mole CH 4 (facility)
E CO2 = × × ×
yr workovers 16.04 tonne CH 4 0.788 tonne mole CH 4 (default)
tonne mole gas 0.12 tonne mole CO 2 44.01 tonne CO 2
× × × = 0.0181 tonnes CO 2 /yr
0.8 tonne mole CH 4 tonne mole gas tonne mole CO 2

Pilot Version, September 2009 F-31

 0.0181 tonne CO 2 ⎛ 0.0439 tonne CH 4 21 tonne CO 2 e ⎞
E CO2e = +⎜ × ⎟ = 0.939 tonne CO 2 e/yr
yr ⎝ yr tonne CH 4 ⎠

U (rel ) ECO ,E CH4 = U(rel ) 2AF +U(rel ) 2EF +U(rel ) 2Facility mole% +U(rel ) 2Default mole%
2

U (rel ) ECO ,E CH4 = 02 + 3002 + 42 + 5.532 = +300%, −100% ( tonnes/yr )

2

Because the emissions cannot be less than zero, the lower bound uncertainty value for this emission source
is truncated at 100%. Therefore, separate upper and lower bound uncertainties are estimated for the CO2e
emissions.

U (abs)ECO e = (U(abs)2ECO +U(abs)2ECH

2 2 4

Upper: U (abs) = (3.00 × 0.0181)2 + (3.00 × 21× 0.0439)2 = 2.77 tonnes CO2e/yr
ECO2e

2.77 tonnes CO2e/yr

U (rel )ECO e = 100% × = +294% ( tonnes CO2e/yr )
2 0.939 tonnes CO2e/yr

U(abs ) ECO e = U(abs ) 2ECO + U(abs ) 2ECH

2 2 4

( −1.00 × 0.0181) + ( −1.00 × 21× 0.0439 )

2 2
Lower: U(abs ) E = = 0.921 tonnes CO 2 e/yr
CO2 e

0.921tonnes CO 2 e/yr
U(rel ) ECO e = 100% × = −98.1% ( tonnes CO 2 e/yr )
2
0.939 tonnes CO 2 e/yr

Uncertainty Estimate for CH4 and CO2 Emissions from Pressure Relief Valves:

0.00065 tonne CH 4 ⎛ 0.80 tonne mole CH 4 (facility) ⎞

E CH4 = (482 PRVs) × ×⎜ ⎟
PRV - yr ⎝ 0.788 tonne mole CH 4 (default EF) ⎠
E CH4 = 0.318 tonnes CH 4 /yr

0.00065 tonne CH 4 tonne mole CH 4 0.80 tonne mole CH 4 (facility)

E CO2 = (482 PRVs) × × ×
PRV - yr 16.04 tonne CH 4 0.788 tonne mole CH 4 (default)
tonne mole gas 0.12 tonne mole CO 2 44.01 tonne CO 2
× × × = 0.131 tonnes CO 2 /yr
0.8 tonne mole CH 4 tonne mole gas tonne mole CO 2

⎛ 0.318 tonne CH 4 21 tonne CO 2 e ⎞

E CO2e = 0.131 tonne CO 2 /yr + ⎜ × ⎟ = 6.81 tonne CO 2 e/yr
⎝ yr tonne CH 4 ⎠

Pilot Version, September 2009 F-32

U (rel ) ECO ,E CH4 = U(rel ) 2AF +U(rel ) EF
2 2
+U(rel ) Facility 2
mole% +U(rel ) Default mole%
2

U (rel ) ECO ,E CH4 = 1.002 + 3102 + 4.002 + 5.532 = +310%, −100% ( tonnes/yr )
2

Because the emissions cannot be less than zero, the lower bound uncertainty value for this emission source
is truncated at 100%. Therefore, separate upper and lower bound uncertainties are estimated for the CO2e
emissions.

U (abs)ECO e = (U(abs)2ECO +U(abs)2ECH

2 2 4

Upper: U (abs) = (3.10 × 0.131) + (3.10 × 21× 0.318)2 = 20.7 tonnes CO2e/yr
2
ECO2e

20.7 tonnes CO2e/yr

U (rel )ECO e = 100% × = +304% ( tonnes CO2e/yr )
2 6.81 tonnes CO2e/yr

U(abs ) ECO e = U(abs ) 2ECO + U(abs ) 2ECH

2 2 4

( −1.00 × 0.131) + ( −1.00 × 21× 0.318)

2 2
Lower: U(abs ) E = = 6.68 tonnes CO 2 e/yr
CO2 e

6.68 tonnes CO 2 e/yr

U(rel ) ECO e = 100% × = −98.1% ( tonnes CO 2 e/y )
2
6.81tonnes CO 2 e/yr

F.11 Uncertainty in Fugitive Sources - Equipment Leaks

Table F-13 provides fugitive component counts associated with the high CO2 content onshore oil field
facility. The corresponding average component emission factors are also provided based on the API
average oil and gas production emission factors given in Table 6-14 of the API Compendium for light crude.
Because the emission factors are taken from API 4615, the weight fraction of CH4 is taken from the
“Generic” weight fraction for light crude give in the API Compendium, Table C-6, which assigns
composition by facility type for an aggregate of all streams associated with that facility. Note that the API
Compendium Table C-6 does not provide CO2 speciation data; therefore, these emissions are not calculated.

Table F-13. Onshore Oil Field (High CO2 Content) Fugitive Emission Factors
Average Uncertainty b Component EF, tonnes Uncertainty b
Component Service Component Count % TOC/comp./hr a %
Valves Liquid and Gas 2,740 75.0 1.32E-06 100
Pump Seals Liquid and Gas 185 75.0 3.18E-07 100
Connectors Gas 110 75.0 1.64E-07 100
Flanges Liquid and Gas 10,000 75.0 7.69E-08 100
OELs Gas 6 75.0 1.21E-06 100
Others Liquid and Gas 710 75.0 7.50E-06 100
Footnotes:
a
Note that for this example, the TOC weight fraction of the gas stream is not 100%. However, the TOC emissions are not adjusted here since
such an adjustment cancels out when calculating CH4 emissions as shown in Equation 6-9 of the API Compendium.
b
Uncertainty based on engineering judgment at a 95% confidence interval.

Pilot Version, September 2009 F-33

Methane and CO2 emissions are calculated for each component by multiplying the component emission
factor by the component count, the annual hours of operation (8760 hours/year, assuming the equipment
remains pressurized year-round), and the weight fraction of CH4. The results for these calculations are
shown below. The total fugitive emissions are then the sum of each of the component emissions.

The uncertainty in the number of devices is 75.0%, the uncertainty in the emission factors is 100%, and the
uncertainty in the weight fraction of CH4 is 15% based on expert judgment. The uncertainty of the
emissions estimate is calculated by applying Equation 4-6 and using the relative uncertainty values. The
calculation is demonstrated for the liquid and gas pump seals.

3.18 ×10-7 tonne TOC 8,760 hr 0.613 tonne CH 4

E CH4 = (185 seals) × × ×
seal - hr yr tonne TOC
E CH4 = 0.316 tonnes CH 4 /yr

0.316 tonne CH 4 21 tonne CO 2 e

E CO2e = × = 6.63 tonne CO 2 e/yr
yr tonne CH 4

U (rel ) ECH = U (rel ) 2Devices + U (rel ) 2EF + U (rel )2hr / yr + U (rel )Weight
2
%
4

U (rel ) ECH = 752 + 1002 + 02 + 152 = +126%, −100% ( tonnes/yr )

4

Because the emissions cannot be less than zero, the lower bound uncertainty value for this emission source
is truncated at 100%. Therefore, separate upper and lower bound uncertainties are estimated for the CO2e
emissions.

U (abs ) ECO e = U(abs ) 2ECH

2 4

Upper: U (abs ) = (1.26 × 21× 0.316) 2 = 8.36 tonnes CO 2 e/yr

E CO2 e

8.36 tonnes CO 2 e/yr

U (rel )ECO e = 100% × = +126% ( tonnes CO 2 e/yr )
2
6.63 tonnes CO 2 e/yr

U (abs ) ECO e = U(abs ) 2ECH

2 4

Lower: U (abs ) = (−1.00 × 21× 0.316) 2 = 6.64 tonnes CO 2 e/yr

E CO2e

6.64 tonnes CO 2 e/yr

U (rel )ECO e = 100% × = −100% ( tonnes CO 2 e/y )
2
6.63 tonnes CO 2 e/yr

The uncertainty of total fugitive emission estimate is calculated by applying Equation 4-4 using the absolute
uncertainties. This is demonstrated for the upper bound uncertainty of CH4 below.

Pilot Version, September 2009 F-34

 Table F-14. Fugitive Emission and Uncertainty Estimates
CH4 Lower Upper CO2e Lower Upper
Emissions Uncertainty Uncertainty Emissions Uncertainty Uncertainty
Component Service (tonnes/yr) % % (tonnes/yr) % %
Liquid
Valves 19.4 -100 126 408 -100 126
and Gas
Liquid
Pump seals 0.316 -100 126 6.63 -100 126
and Gas
Connectors Gas 0.0969 -100 126 2.03 -100 126
Liquid
Flanges 4.13 -100 126 86.7 -100 126
and Gas
OELs Gas 0.0390 -100 126 0.819 -100 126
Liquid
Others 28.6 -100 126 600 -100 126
and Gas
52.6 66.2 83.3 1,105 66.2 83.3

CH 4 : U (abs )
∑ ( Fugitive )
= ∑U (abs) 2
Fugitive

(1.26 ×19.4) 2 + (1.26 × 0.316) 2 + (1.26 × 0.0969) 2

CH4 Upper Bound Uncertainty = = 43.8
+ (1.26 × 4.13) 2 + (1.26 × 0.0390) 2 + (1.26 × 28.6) 2
43.8
U (rel ) = 100% = 83.3%
∑ (Wt %) 52.6

F.12 Propagating Uncertainty in Vented Sources – Fleet Vehicle Refrigerants

Emissions associated with the air conditioning units in fleet vehicles for this example are calculated based
assuming the refrigerant is R-134a. As stated in Section F.5, the uncertainty associated with the count of
vehicles is 0%. The uncertainties associated with the refrigerant capacity and the emission factor are
assumed to be 100% and 50%, respectively. The calculations for fleet vehicle refrigeration emissions are
shown below.

1.0 kg tonne R-134a

E R −134a = 5 vehicles × × 20% ×
vehicle 1000 kg
E R −134a = 0.00100 tonne R-134a/yr

U (rel )emissions = U(rel ) 2Vehicles +U(rel )Capacity

2
+U(rel ) 2EF

U (rel )emissions = 02 + 1002 + 502 = ±112% ( tonnes R-134a/yr )

0.00100 tonne R-134a 1300 tonne CO 2 e

E CO2e = ×
yr tonne R-135a
E CO2e = 1.30 tonne CO 2 e/yr

Pilot Version, September 2009 F-35

Uncertainty Estimate of CO2e Emissions from Fleet Vehicle Refrigerants:

The global warming potential for R-134a is treated as a constant, so the uncertainty in the CO2e emissions is
the same as for the refrigerant emissions.

U (rel )CO2e emissions = U(rel ) 2R-134a emissions

U (rel )CO2e emissions = 1122 = ±112% ( tonnes CO 2 e/yr )

F.13 Uncertainty in Indirect Sources – Electricity Consumption

Emissions associated with the electricity purchased by the facility are calculated using emission factors in
the API Compendium.

917 MW-hr 0.601 tonne CO 2

E CO2 = × = 551 tonnes CO 2 /yr
yr MW-hr

917 MW-hr 8.46 × 10-6 tonne CH 4

E CH4 = × = 0.00776 tonnes CH 4 /yr
yr MW-hr

917 MW-hr 6.85 ×10-6 tonne N 2 O

E N2O = × = 0.00628 tonnes N 2 O/yr
yr MW-hr

551 tonne CO 2 ⎛ 0.00776 tonne CH 4 21 tonne CO 2 e ⎞

E CO2e = +⎜ × ⎟
yr ⎝ yr tonne CH 4 ⎠
⎛ 0.00628 tonne N 2 O 310 tonne CO 2 e ⎞
+⎜ × ⎟ = 553 tonne CO 2 e/yr
⎝ yr tonne N 2 O ⎠

Based on expert judgment, the uncertainty in the activity factor is 2%, the uncertainty in the CO2 emission
factor is 10.0%, and the uncertainty in the CH4 and N2O emission factors is 100%. The uncertainty of the
emissions estimate is calculated by applying Equation 4-6 and using the relative uncertainty values.

U (rel )CO2 = U (rel ) 2AF + U (rel ) 2EF = 22 + 102 = ±10.2% ( tonnes CO 2 /yr )

U (rel )CH 4 and N2O = U (rel ) 2AF + U (rel ) 2EF = 22 + 1002 = ±100% ( tonnes/yr )

Uncertainty Estimate of CO2e Emissions from Purchased Electricity:

The final step is to convert the emissions of CO2, CH4, and N2O to a CO2e. Note there are no uncertainties
for any of the GWPs, so the combined uncertainty is based on summing the individual emissions and
applying Equation 4-4 (using absolute uncertainty values).

Pilot Version, September 2009 F-36

U (abs ) ECO e = (U(abs ) 2ECO +U(abs) 2ECH +U(abs ) 2E N O
2 2 4 2

U (abs ) ECO e = (0.102 × 551) 2 + (1.00 × 21× 0.00776) 2 + (1.00 × 310 × 0.00628)2 = 56.2 tonnes CO 2 e/yr
2

56.2 tonnes CO 2 e/yr

U (rel )ECO e = 100% × = ±10.2% ( tonnes CO 2 e/yr )
2
553 tonnes CO 2 e/yr

F.14 SUMMARY UNCERTAINTY – Propagating Uncertainty for the Example Facility

Total emissions for this facility are summarized in Table F-15, along with the summary uncertainty
propagation. The uncertainty of the total emissions is calculated by applying Equation 4-4 and using the
absolute uncertainty values. In practice, uncertainty values are usually displayed in relative terms, which are
provided in terms of upper and lower % uncertainty values in Table F-15 for those emissions sources with
an asymmetrical uncertainty.

As an example, the summed upper bound uncertainties for N2O combustion emissions for the facility are
calculated as follows, using absolute uncertainties:

U (abs ) N 2O CombustionTotal = U (abs ) 2BoilersandHeaters + U (abs )Turbines

2
+ U (abs ) 2Flare + U (abs ) 2IC + U (abs ) 2Fleet

(0.0242 × 1.50) 2 + (0.325 ×1.51) 2 + (0.223 × 2.00) 2 + (0.00175 × 1.51) 2

U (abs ) N 2O CombustionTotal = = 0.663
+ (0.0000467 × 1.51) 2 + (0.00871× 1.51) 2
0.663
U (rel ) N 2O Combustion Total = 100% × = 114%
0.582

U (abs ) N 2OTotal = U (abs )Combustion

2
+ U (abs )Vented
2
+ U (abs ) 2Fugitive + U (abs ) 2Indirect

U (abs ) N 2OTotal = (0.582 × 1.14)2 + 02 + 02 + (0.00628 ×1.00) 2 = 0.663

0.663
U (rel ) N 2O Total = 100% × = 113%
0.582

This calculation was repeated for the sum of each category, and for the totals in that GHG gas type. The
gases were each converted to CO2e using the GWPs. As was stated previously, for the purpose of compiling
a facility-level GHG inventory, the GWP’s are assumed to have no uncertainty. The bottom row of
Table F-15 shows the emissions in CO2e.

Pilot Version, September 2009 F-37

 Table F-15. Onshore Oil Field (High CO2 Content) Emissions

CO2 CH4 N2O and Other GHG Total Emissions, CO2 Eq

Uncertainty Uncertainty Uncertainty Uncertainty
Emissions % Emissions % Emissions % Emissions %
Source Type Source (tonnes/yr) Lower Upper (tonnes/yr) Lower Upper (tonnes/yr) Lower Upper (tonnes/yr) Lower Upper
Combustion Boiler/heaters 5,200 8.78 8.78 0.0865 26.1 26.1 0.0242 100 150 5,210 8.77 8.77
Sources Natural gas engines 13,900 15.7 15.7 0.904 29.4 29.4 0.325 100 151 14,100 15.6 15.6
Emergency generator IC
0.0108 27.8 27.8 0.00175 100 151
engine 219 15.6 15.6 220 15.5 15.5
Fire water pump IC engine 0.00112 100 106 0.0000467 100 151
Flares 27,400 23.4 23.4 153 25.3 25.3 0.223 100 200 30,700 21.1 21.1
Fleet vehicles 127 19.4 19.4 0.00643 100 151 0.00871 100 151 129 19.1 19.2
Combustion Total 46,800 14.5 14.5 154 25.2 25.2 0.582 68.9 114 49,900 13.7 13.7
Vented sources Dehydration and Kimray
105 77.5 77.5 254 77.5 77.5 NA NA NA 5,440 76.0 76.0
pump vents
Tanks – flashing losses 775 90.4 90.4 1,880 90.4 90.4 NA NA NA 40,300 88.7 88.7
Amine unit 62,600 6.97 6.97 193 100 119 NA NA NA 66,700 8.94 9.77
Pneumatic devices 64.6 50.2 50.2 157 50.2 50.2 NA NA NA 3,360 49.2 49.2
Chemical injection pumps 48.6 100 108 118 100 108 NA NA NA 2,530 98.1 106
Vessel blowdowns 0.0702 100 326 0.171 100 326 NA NA NA 3.65 98.1 319
Compressor starts 0.745 100 190 1.81 100 190 NA NA NA 38.7 98.1 187
Compressor blowdowns 0.333 100 179 0.808 100 179 NA NA NA 17.3 98.1 175
Well workovers 0.0181 100 300 0.0439 100 300 NA NA NA 0.939 98.1 294
Other non-routine (PRVs) 0.131 100 310 0.318 100 310 NA NA NA 6.81 98.1 319
Vented Total 63,600 6.95 6.95 2,610 66.3 66.5 NA NA NA 118,000 30.9 31.0

Pilot Version, September 2009 F-38

 Table F-15. Onshore Oil Field (High CO2 Content) Emissions, continued

CO2 CH4 N2O and Other GHG Total Emissions, CO2 Eq

Uncertainty Uncertainty Uncertainty Uncertainty
Emissions % Emissions % Emissions % Emissions %
Source Type Source (tonnes/yr) Lower Upper (tonnes/yr) Lower Upper (tonnes/yr) Lower Upper (tonnes/yr) Lower Upper
Fugitive Sources Fugitive components NA NA NA 52.6 66.2 83.3 NA NA NA 1,100 66.2 83.3
Fleet vehicle
NA NA NA NA NA NA 0.00100 100 112 1.30 100 112
refrigeration, R-314a
Fugitive Total NA NA NA 52.6 66.2 83.3 0.00100 100 112 1,100 66.1 83.2
Indirect Sources Electricity consumed 551 10.2 10.2 0.00776 100 100 0.00628 100 100 553 10.2 10.2
Indirect Total 551 10.2 10.2 0.00890 100 100 0.00628 100 100 553 10.2 10.2
TOTAL (tonnes of each gas) 111,000 7.29 7.29 2,820 61.4 61.7 N2O:0.588 67.1 113
R134a:
0.00100 100 112
TOTAL (CO2 Equivalents) 111,000 7.29 7.29 59,100 61.4 61.7 184 66.6 112 170,300 21.9 21.9
Values may not sum due to rounding.

Pilot Version, September 2009 F-39

INTERNATIONAL PETROLEUM INDUSTRY
ENVIRONMENTAL CONSERVATION ASSOCIATION