Addressing uncertainty in Climate

oil and natural gas industry Change

Water
February 2015
Efficiency in water use
2014
greenhouse gas inventories
www.ipieca.org
Technical considerations
Guidance document for the and calculation methods
upstream onshore oil and gas industry
Addressing uncertainty in
oil and natural gas industry
greenhouse gas inventories
Technical considerations and calculation methods

Acknowledgements

This document was prepared by The LEVON Group,

LLC and URS Corporation
IPIECA • API

Table of Contents
INTRODUCTION 7
DOCUMENT AT A GLANCE 8
1. INTRODUCTION 10
1.1 Importance of accurate and reliable GHG accounting 10
1.2 Overview of uncertainty terminology 11
1.3 Types of errors 11
1.4 Numerical determination of uncertainty intervals 12
1.5 Emissions inventory uncertainty assessment 12
2. ASSESSMENT OF UNCERTAINTY RELEVANT
TO GREENHOUSE GAS EMISSION QUANTIFICATION 13
2.1 Overview of emissions inventory uncertainty 13
2.2 Sources of measurement uncertainty 15
2.3 Emission calculation approaches 15
2.4 Inventory steps and data aggregation 18
2.5 Emissions inventory and uncertainty
assessment in the petroleum and natural gas industry 18
3. OVERVIEW OF MEASUREMENT PRACTICES FOR
ESTIMATING GREENHOUSE GAS EMISSIONS 22
3.1 Flow measurement practices 23
3.1.1 Measurements by Orifice Meters 23
3.1.2. Measurement of flow to flares 25
3.1.3 Flow measurements uncertainty analysis 26
3.1.4 Uncertainty specifications for “custody transfer” measurements 26
3.2 Flow measurements for GHG inventories 28
3.2.1 Flow meter types 28
3.2.2 Fuel gas measurement 28
3.2.3. Integrated measurement systems 30
3.3 Sampling and analysis for quantifying GHG emissions 32
3.3.1 Gaseous samples collection and handling 32
3.3.2 Quantifying sampling precision 33
3.4 Carbon content measurement practices 34
3.4.1 Laboratory-based measurements 34
3.4.2 On-line measurements 34
3.4.3 Carbon content calculation 36
3.5 Heat content determination 37
3.5.1 Direct measurements 38
3.5.2 Computational methods 38
3.6 Venting and fugitive emissions measurements 40
3.6.1 Vented emission measurements 41
3.6.2 Fugitive emission measurements 41
3.7 Laboratory management system 42

2
 ADDRESSING UNCERTAINTY IN OIL & NATURAL GAS INDUSTRY GREENHOUSE GAS INVENTORIES

4. OVERVIEW OF REPORTING PROGRAMS’

UNCERTAINTY REQUIREMENTS AND TOOLS 44
4.1 Global approaches applicable to petroleum and natural gas sector 44
4.2 Uncertainty targets for mandatory GHG programs 48
4.3 Tools for calculating uncertainty 50
4.3.1 Numerical methods for calculating uncertainty 50
4.3.2 Examples of program-based tools 51
5. UNCERTAINTY CALCULATION OVERVIEW 53
5.1 Measurement uncertainty 53
5.1.1 Precision and bias 53
5.1.2 Confidence intervals 54
5.1.3 Implementation of Pedigree Matrix approach 55
5.2 Overview of uncertainty propagation 56
5.2.1 Propagation equations 57
5.2.2 Comparison of uncertainty propagation and Monte Carlo 58
5.2.3 Correlation coefficient 58
5.3 Uncertainty aggregation examples 59
5.3.1 IPCC aggregation template 59
5.3.2 Pedigree Matrix example 61
5.3.3 Uncertainty assessment from API Technical Report 2571 61
5.4 Strategic reduction of uncertainty 62
5.4.1 Periodic updates 62
5.4.2 Reducing uncertainty of emission estimates 62
6. REFERENCES 66

APPENDICES (Available in a separate document)

A Glossary of Statistical and GHG Inventory Terms A-1

B List of Industry Measurement Standards B-1
C Operating Conditions, Inspection, Calibration and
Expected Uncertainties for Common Flow Meters C-1
D Select Measurement Methods Summaries D-1
E Units Conversion E-1
F Statistical Concepts and Calculation Methods: Tutorial F-1
G Uncertainty Estimation Details for an Example
Onshore Oil Field Inventory G-1
H Uncertainty Estimation Details for an Example Refinery Inventory H-1

3
IPIECA • API

List of Tables
Table 2-1. Overview of methods used to estimate emissions uncertainty 14
Table 2-2. Summary of typical sources of measurement uncertainty 16
Table 2-3. Common emission estimation approaches and their error sources 17
Table 3-1. Example of flare flow measurement systems combined uncertainty 26
Table 3-2. Summary of Alberta ERCB accuracy requirements 27
Table 3-3. Compilation of specifications for common flow meters 29 - 30
Table 3-4. Summary of selected carbon content measurement methods 35
Table 3-5. Summary of selected heating value measurement methods 39
Table 3-6. GHGRP measurement methods for vented and fugitive emission sources 40
Table 4-1. Inter-Country comparison of select mandatory GHG reporting programs 45 - 47
Table 4-2. Comparison of reporting scope between EPA’s GHGRP and CDP 47
Table 4-3. EU regulation maximum permissible uncertainty for activity data tiers 48
Table 4-4. Highlights of calibration requirements under the EPA GHGRP 49
Table 4-5. Estimated uncertainty levels based on Australia National
Greenhouse Gas and Energy Reporting (NGER) 51
Table 5-1. Example of Pedigree Matrix for determining uncertainty
scaling factors based on data quality ratings 55
Table 5.2. Basic Pedigree Matrix uncertainty factors for selected pollutants 56
Table 5-3. Adapted IPCC Template: Approach for Aggregating Uncertainty 59
Table 5-4. Example uncertainty calculation and reporting table 60
Table 5-5. Pedigree Matrix example 60
Table 5-6. Example of flow measurement total uncertainty 61
Table 5-7. Emission uncertainty ranking for onshore oil production example 63

List of Figures
Section
2-1 Hierarchy of emission estimation approaches 19
5-1 Measurement error over time of an unbiased estimate 54
5-2 Measurement error over time of a biased estimate 54
5-3 Onshore oil field: summary of CO2 equivalent emissions and uncertainties 63
List of Figures in Appendices document
F-1 Decision diagram for emission factor uncertainty F-5
F-2 Decision diagram for measurement uncertainty F-6
F-3 Step C – decision diagram for uncertainty aggregation F-22

4
 ADDRESSING UNCERTAINTY IN OIL & NATURAL GAS INDUSTRY GREENHOUSE GAS INVENTORIES

Acronym List
ADEME French Environment and IPCC Intergovernmental Panel on
Energy Management Agency Climate Change
AF Activity Factor IPIECA Global oil and gas industry
AGA American Gas Association association for environmental
ANSI American National and social issues
Standards Institute ISO International Standards
API American Petroleum Institute Organization
ASTM American Society of kW Kilowatt
Testing and Materials kWh Kilowatt-hour
AWP Alternative Work Practice l Liter
BOE Barrels of Oil Equivalent LACT Lease Automatic Custody
BTU British Thermal Unit Transfer
CARB California Air Resources Board LHV Lower Heating Value
CDP Carbon Disclosure Project LPG Liquefied Petroleum Gas
CE Calibration Error MJ Megajoules
CEMS Continuous Emissions MRG Monitoring and Reporting
Monitoring System Guidelines
CER Clean Energy Regulator MPMS Manual of Petroleum
CH4 Methane Measurement Standards
CIP Chemical Injection Pumps NGER National Greenhouse and
CO2 Carbon Dioxide Energy Reporting
CORINAIR The Core Inventory of Air Nm3 Normal Cubic Meters
Emissions in Europe OBQ On-board Quantity
DP Differential Pressure OECD Organization for Economic
DQI Data Quality Indicators Cooperation
EDR Electronic Data Reporting IOGP International Oil and Gas
EF Emission Factor Producers Association
eGGRT Electronic GHG Reporting Tool OSCAR Online System
EIPP Emissions Inventory Comprehensive Activity
Improvement Program Reporting
EJ Exajoules OVA Organic Vapor Analyzer
EMEP European Monitoring and PD Positive Displacement
Evaluation Programme PEI Primary Element Inspection
EPA Environmental Protection PPM Parts Per Million
Agency ROB Remaining on Board
ERCB Energy Resources SAR Second Assessment Report
Conservation Board SI International System of Units
ETS European Union Emissions t tonne
Trading System TJ Terajoule
ETSG ETS Support Group TCD Thermal Conductivity
EU European Union Detector
EVCI Electronic Volume TVA Toxic Vapor Analyzer
Conversion Instrument UNFCCC United Nations Framework
FCCU Fluid Catalytic Cracking Unit Convention on Climate
FID Flame Ionization Detector Change
FFMS Flare Flow Measurement VEF Vessel Experience Factor
System VEFD Vessel Experience Factor on
GC Gas Chromatography Discharging
GHG Greenhouse Gas VEFL Vessel Experience Factor on
GHGRP GHG Reporting Program Loading
Gj Gigajoule VOC Volatile Organic Compounds
GOR Gas-to-Oil Ratio WBCSD World Business Council for
GPA Gas Processors Association Sustainable Development
GWP Global Warming Potential WCI Western Climate Initiative
HHV Higher Heating Value WRI World Resources Institute
Hp Horsepower
INGAA Interstate Natural Gas
Association of America 5
 ADDRESSING UNCERTAINTY IN OIL & NATURAL GAS INDUSTRY GREENHOUSE GAS INVENTORIES

INTRODUCTION
The global petroleum and natural gas industry The uncertainties inherent in the data used for
has been active in promoting consistency emission inventories help inform and improve
and harmonization for industry greenhouse understanding of the data’s use. The uncertainty
gas (GHG) emission inventories. Industry of petroleum and natural gas companies’ GHG
associations and their members have emission inventories, or of its quantified emission
been contributing to the development of reductions, is determined largely by uncertainties
guidance for accounting and reporting of in the estimates of the key (largest) contributing
GHG emissions (API/IPIECA/OGP, 2011), and sources. In turn, each of these uncertainties
compiling methodologies that are appropriate depends on the quality and availability of
for estimating GHG emissions from industry sufficient data to estimate emissions. The
operations (API, 2009). This guidance has robustness of industry data disclosures is
been recently augmented with guidelines to receiving increased attention with special
account for reductions associated with GHG emphasis on the need to understand how GHG
projects (API/IPIECA, 2007a). emissions and emission reductions are quantified.

Background
The American Petroleum Institute (API), The 2007 workshop served as the first step in
Concawe (the scientific and technical addressing uncertainty and accuracy issues. In
organization of the European petroleum refining the ensuing industry discussion, a list of priority
industry) and the The global oil and gas industry issues was prepared. This list is comprised of
association for environmental and social issues items that industry experts ought to address
(IPIECA) convened an international workshop in a systematic fashion. As presented in the
on the topic on 16 January 2007, in Brussels, workshop summary report, the issues listed by
Belgium. The goals of this workshop were to: industry members fall into three thematic areas:
• develop an understanding of the relative 1. Measurement methods
importance of the key factors that
contribute to uncertainty; 2. Computational methods
• review and discuss emerging techniques 3. External communications
for quantitative assessment of the
uncertainty and accuracy of GHG emissions Because the industry recognizes the need
estimates; for meeting regulatory mandates and
stakeholders’ expectations, follow-up activities
• identify emission sources and methods will be designed to provide opportunities for
where petroleum and natural gas industry continued dialogue and collaborative activities
efforts are needed to improve accuracy with stakeholders.
and reduce uncertainty to acceptable
levels; and A pilot version of this document was published
• create a prioritized list of topics to be in September 2009 to allow broad review
addressed by the petroleum and natural gas and implementation of guidance provided
industry to minimize emissions estimation by all stakeholders. Comments received,
lessons learned, and new developments that
uncertainty and improve data accuracy. have occurred in the area since then are now
A summary report, as well as all the workshop summarized in this final guideline document.
presentations, is posted on the IPIECA website
(API/IPIECA, 2007b, http://www.ipieca.org/
publication/greenhouse-gas-emissions-
estimation-and-inventories-addressing-
uncertainty-and-accuracy ).

7
IPIECA • API

Document goals operations are quite complex; they are

based on a combination of measured and
The goals of these guidelines are to: estimated emissions data, according to local
• Summarize in a single document requirements and available information. The
guidance for meeting data needs of overall range of uncertainty associated with an
a range of voluntary and mandatory entity GHG inventory is determined primarily
initiatives, as well as the requirements of by the uncertainty associated with the largest
diverse GHG regimes; (“key”) sources of emissions. In turn, the
• Supplement existing industry guidance confidence interval associated with each
and provide technically valid approaches individual source depends on the availability
for use by the global petroleum and of sufficient data to estimate emissions, or on
natural gas industry to improve GHG the quality of that data, in order to properly
emissions estimation robustness and account for emission variability.
data quality; and Uncertainty analysis is a potential tool to not
• Provide a practical and easy to only assess confidence intervals, but more
implement approach for a range of importantly to allow the targeting of specific
applications starting with establishing areas for enhanced data collection. Such
a corporate carbon footprint, publicly an analysis will enable a user to prioritize
reporting GHG emissions, and assessing emission sources in terms of their overall
lifecycle emissions. contribution to the emissions inventory and
its overall uncertainty range.
The technical considerations and statistical This document is a companion to the API
calculation examples included in this Compendium of Greenhouse Gas Emissions
document are designed to provide technically Methodologies for the Oil and Natural Gas
sound approaches for assessing uncertainty Industry (API, 2009). It provides a range of
in different situations, but they do not background information on industry practices
represent an industry standard. The discussion and specific calculation methods that will
will include: clarification of the sources of enable inventory developers to quantify and
uncertainty in GHG inventories; information better understand the uncertainty associated
on measurement practices and their with the resultant GHG emissions.
associated uncertainties; overview of different Section 1 introduces some basic concepts
emerging regulatory approaches to ensuring and terms that provide a foundation for
quality of emission estimation data; and understanding GHG emissions inventory
explanation of statistical procedures and tools uncertainty. This terminology is used throughout
that can be used to quantify uncertainties. the document. Section 1 covers: the importance
The case studies included are for illustrative of reliable GHG accounting; a terminology
purposes and demonstrate options that may overview; definition of error types; and a
be available for practical implementation of description of the determination of uncertainty
the recommended approaches. ranges (also known as confidence intervals).
Section 2 discusses the major sources of
DOCUMENT AT A GLANCE uncertainty in GHG inventories. It moves from
general concepts to issues that are relevant
This document is designed to provide a
to GHG inventories in the petroleum and
summary of technical considerations that are
natural gas industry. It also describes factors
important for understanding and calculating
that could introduce errors into the emission
GHG emission inventory uncertainty. The
measurement process and contribute to the
document provides technical background and
range of uncertainties of estimated emissions.
specific calculation methods to determine
It introduces the categories of emission
uncertainties with targeted measurements
estimation approaches and their uncertainty
and emission factors, and to determine how
implications, and concludes with a short
to aggregate these individual terms to derive
description of emission inventory steps and
uncertainty ranges (at a pre-designated
data aggregation.
probability level) for entire GHG inventories at
any given level. Section 3 provides an overview of measurement
practices, focusing on gas flow measurements
These emission inventories of typical
and the determinations of carbon content and
petroleum and natural gas systems
heating values of combusted fuels. The section
8
 ADDRESSING UNCERTAINTY IN OIL & NATURAL GAS INDUSTRY GREENHOUSE GAS INVENTORIES

recognizes industry-recommended practices Detailed technical information is organized in

and standards that have traditionally applied to appendices (available in a separate document),
“custody transfer”. This section goes on to discuss as follows:
data considerations when collecting information
on activity levels and applicable GHG emissions.
• Appendix A: Glossary of statistical and GHG
inventory terms
It includes an overview of measurement
practices that would result in high quality data • Appendix B: A comprehensive list of
when properly implemented, while focusing applicable industry measurement
on measurements that are applicable to the standards
key contributing sources, i.e. carbon dioxide • Appendix C: Operating conditions,
(CO2) emissions from combustion devices. inspection, calibration and manufacturers’
Topics discussed include: flow measurement reported measurement errors for common
practices; uncertainties of flow measurements flow meters
for GHG inventories; uncertainty of sampling • Appendix D: Measurement method
and analysis for GHG estimation; and laboratory summaries for carbon content
management systems. measurement methods and heating value
Section 4 provides an overview of the measurement methods
uncertainty requirements from various • Appendix E: Unit conversions including
GHG reporting programs applicable to energy units, common units of measure
the petroleum and natural gas sector and for fossil fuel heating content values, and
compares elements of programmes and their carbon content of selected fuels
uncertainty targets. Where available, calculation • Appendix F: Tutorial on statistical concepts
tools available to assist reporters in quantifying and calculations
the uncertainty of GHG emission estimates for
these programs are also introduced.
• Appendix G: Calculation details for
uncertainty estimation for an example oil
Section 5 outlines uncertainty principles and natural gas production facility GHG
and introduces general calculation and inventory
aggregation approaches. This section also • Appendix H: Calculation details for
discusses reducing uncertainty in the context uncertainty estimation for an example
of improving GHG data quality. refinery GHG inventory.

9
IPIECA • API

1. INTRODUCTION
Policymakers use entity GHG inventories and SECTION FOCUS
reported facility-level GHG emissions to develop
strategies and policies for emission reductions This is an introductory section that
and to track the progress of these policies. introduces some basic concepts and
Both regulatory agencies and corporations rely terms that are the foundation for
on inventories to better understand emission understanding GHG emissions inventory
sources and trends. GHG inventory data are uncertainty. This terminology will be
associated with varying degrees of uncertainty, further expanded throughout the next
and such actual uncertainties have both sections of the document.
technical and policy implications.
The subsections include:
“Uncertainty analysis” has been increasingly
recognized as an important tool for improving • Importance of accurate and reliable
national, sectoral, and corporate inventories of GHG accounting;
GHG emissions and removals (IPCC, 2000). This • Overview of uncertainty
increased attention on accurate inventories terminology;
has resulted in the need to provide guidance • Types of errors;
to industry on technical considerations and
calculation methods for consistent estimation
• Numerical determination of
uncertainty intervals; and
of GHG inventory uncertainty. This typically
would consist of: • Emissions inventory uncertainty
assessment.
• determination of the uncertainties
associated with the individual
measurements and factors used in
constructing the emissions inventory; and Since an understanding of the magnitude and
• propagation and aggregation of these sources of GHG emissions is critical for properly
individual terms to derive uncertainty managing these emissions, using a consistent
intervals (at a pre-designated probability approach can significantly improve industry-
level) for the whole inventory. wide, comparable estimates of emissions and
The extent and scope of such analysis will emission reductions.
depend on the likely uses of this information. Higher-quality GHG data lead to higher
For example, the uncertainty analysis required certainty of emission assessments and improved
for data that are merely used for relative confidence in the data reported. This is true for
ranking or comparison of trends would be national and government assessments, and is also
different from that required to demonstrate important at the entity or facility level. To ensure
attainment of GHG emission limits or progress that a company’s strategies and forward-looking
made towards meeting GHG emission actions are based on the most robust data set and
reduction targets. most appropriate computational methods, it is
important to address the following factors:
a. Comparability: Uncertainty analysis enables
1.1 Importance of accurate and comparison of data trends and between
reliable GHG accounting entities.
Key areas that benefit from reliable GHG a. Consistency: Science-based estimation and
accounting include: measurement methods should include
• focused GHG emissions management; consideration of accuracy and precision.
• reduced business risk and reputation a. Certainty: Emission inventories are estimates;
management; and uncertainty analysis provides a likely range of
• participation in GHG emissions mitigation those estimates.
programs.
a. Confidence: Users of the inventory need to
understand the reliability of the estimated
emissions, especially when they are used for
Comparability, Consistency, Certainty, Confidence policy development, or to target inventory
improvements.

10
 ADDRESSING UNCERTAINTY IN OIL & NATURAL GAS INDUSTRY GREENHOUSE GAS INVENTORIES

1.2 Overview of uncertainty

terminology EXHIBIT 1-1: SELECTED TERMINOLOGY
In Chapter 13 of its Manual of Petroleum • Accuracy – Ability to indicate values that closely
Measurement Standards (MPMS), the API approximate the true value of the measured variable.
provides detailed guidance on concepts
and procedures for addressing the statistical
• Bias – Any influence on a result that produces an
incorrect approximation of the true value of the
procedures that should be followed when
variable being measured. Bias is the result of a
estimating a true quantity from measurements
predictable systematic error.
- or models – and when deriving the confidence
interval of the results (API, 1985). That chapter • Confidence interval (or range of uncertainty) –
also examines sources of error and recommends The range or interval within which the true value is
how to develop a statement of the overall range expected to lie with a stated degree of confidence.
of uncertainty of the results obtained. Some • Confidence level – The degree of confidence that
of the key terms used in the API MPMS are may be placed on an estimated range of uncertainty.
presented in Exhibit 1-1. • Error – The difference between true and
Appendix A presents an expanded glossary observed values.
of statistical terms with comments on how • Precision – The degree to which data within a set
these terms are used in the context of GHG cluster together.
emission inventories. • Random error – An error that varies in an
This terminology is also used by the unpredictable manner when a large number of
Intergovernmental Panel on Climate Change measurements of the same variable are made under
(IPCC) as part of their ‘Good Practice Guidelines’ effectively identical conditions.
for managing and estimating uncertainty • Spurious error – A gross error in procedure (for
in national emission inventories (IPCC, example, human errors or machine malfunctions).
2000), which provides a conceptual basis for • Systematic error – An error that, in the course of
uncertainty analyses. The IPCC guidelines a number of measurements made under the same
introduce a structured approach to estimating conditions on material having the same true value
GHG inventory uncertainty by incorporating of a variable, either remains constant in absolute
methods used to determine uncertainties value and sign, or varies in a predictable manner.
of individual terms and aggregate them to Systematic errors result in a bias.
the total inventory. The IPCC also recognizes
that other uncertainties may exist, such as
• Variance – The measure of the dispersion or scatter
of the values of the random variable about the mean.
those arising from inaccurate definitions or
procedures, which cannot be addressed by Source: API MPMS Chapter 13.1 (API, 1985 and 2011)
statistical means.

1.3 Types of errors

The difference between the observed value should not be confused with the overall
of a variable and its true value includes all inventory uncertainty. The overall uncertainty
errors associated with a given measurement or of the emission inventory is comprised of the
estimation process. Such errors are comprised weighted individual errors as further discussed
of instrumentation errors, errors resulting in Section 5.2 with examples in Section 5.3.
from faulty sampling procedures, changes in
Indicators such as the range, confidence
conditions during the measurement period, or
interval or other error bounds are typically used
use of improper methods. Three basic types of
to quantify an emission estimate uncertainty.
errors should be considered:
Errors may be due to the inherent variability
• spurious errors; of the emission processes and the bias – or
• systematic errors; and imprecision – in the terms typically used to
define them. Bias is the result of a systematic
• random errors.
error in some aspect of the emissions
One or all of these errors could be associated inventory process. In contrast, imprecision is
with individual measurements or input due to random errors or fluctuations in the
variables used for deriving an emissions measurement process.
inventory, though such individual errors

11
IPIECA • API

1.4 Numerical determination of 1.5 Emissions inventory uncertainty

uncertainty intervals assessment
The uncertainty intervals associated with Several types of parameters are closely linked in
input quantities (such as measured emission general with emissions inventory development
rates, activity data or emission factors) and the assessment of its uncertainty:
are characterized by the dispersion of the
respective values that are used in their
• direct emissions data;
derivation. Mathematically these intervals are • activity data; and
defined as either the standard deviation of the • emission factors.
sample populations or the standard deviation The uncertainty of these parameters can be
of the sample means. The standard deviation represented by either a probability distribution
of the mean, known also as the standard error or as a range. Common distributions include,
of the mean, is the standard deviation of the but are not limited to, the normal distribution,
sample data set divided by the square root of lognormal distribution, uniform distribution,
the number of data points. While the standard and triangular distribution (Lloyd, 2007). For
deviation and variance of the data set do activity data and emission factor data, the log-
not change systematically with the number normal distribution is often determined to be a
of observations, the standard deviation reasonable fit.
of the mean decreases as the number of
observations increases. Different approaches to quantifying parameter
uncertainty may include:
Estimating uncertainties in emission inventories
is based on the characteristics of the variable(s) • measured uncertainty (represented by
of interest (input quantities) as estimated from standard deviations);
the corresponding data set. The statistical • uncertainty factors for specific activities or
computations could entail the determination of: sector data (reported in literature);
• the arithmetic mean (mean) of the data set; • probability distributions from available
databases; and
• the standard deviation of the data set
(the standard error, the square root of the • the pedigree matrix approach, based on
variance); data quality indicators (DQIs).
• the standard deviation of the mean (the While this guidance focuses on quantifying
standard error of the mean); parameter uncertainty from measurements,
activity data, and emission factors, the pedigree
• the probability distribution of the data; and matrix approach may also be applicable as
• covariance of the input quantity with other further elaborated in the following sections.
input quantities used in the inventory
calculations.
These calculation methods are discussed
in the context of estimating uncertainty in
Section 5.0.

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. The quantification of uncertainty
intervals for GHG emissions will depend both
on the uncertainty of measurements, its repre-
sentativeness, and the assumed distributions
of other key parameters that are an integral
part of emission estimation. General rules for
evaluating and expressing uncertainty may be
followed at various levels in many fields (ISO,
2005; ISO, 2008; IPCC, 2000).

12
 ADDRESSING UNCERTAINTY IN OIL & NATURAL GAS INDUSTRY GREENHOUSE GAS INVENTORIES

2. ASSESSMENT OF
UNCERTAINTY RELEVANT TO
GREENHOUSE GAS EMISSION SECTION FOCUS
QUANTIFICATION
A myriad of sources contribute to the uncertainty
This section discusses the major
of an emission inventory. Whether at the national,
sources affecting the uncertainty of
entity, or facility level, the ability to quantify
GHG inventories. It moves from general
emissions and understand their associated
concepts to issues that are germane to
uncertainty hinges on two main factors:
GHG inventories in the petroleum and
• readily available data for emission natural gas industry. It also describes
quantification; and factors that could introduce errors into
• knowledge of input parameters for the emission measurements process and
statistical calculation of uncertainty. contribute to the range of uncertainties
of estimated emissions.
The overall range of uncertainty associated
with an entity GHG inventory is primarily The subsections address:
determined by the uncertainty associated
with the largest (“key”) sources of emissions.
• Overview of emissions inventory
uncertainty;
These key emission sources differ by industry
sector and characteristics of company • Sources of Measurement Uncertainty;
operations1. Although very large confidence • Emission Calculation Approaches;
intervals may be associated with the data • Inventory Steps and Data
used to characterize some small sources, the Aggregation; and
overall impact on the range of uncertainty
at the entity, or installation level, may often
• Emissions Inventory and Uncertainty
Assessment in the Petroleum and
be very small. In turn, the confidence interval
Natural Gas Industry.
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 Uncertainties might be associated with one

uncertainty or more factors such as: sampling, measuring,
incomplete reference data, or inconclusive
Inventory uncertainties are largely dependent expert judgment. Uncertainties associated with
on the quality of available data. For inventories of emissions from known sources
determining the uncertainty ranges one has to can be broadly categorized into scientific un-
evaluate three error categories: certainty and estimation uncertainty. Scientific
a. Spurious errors: May be due to incomplete, uncertainty is a function of understanding the
unclear, or faulty definitions of emission science of the actual emission and/or removal
sources that result from human error or process. Estimation uncertainty--the main sub-
machine malfunction. ject of this document--can be further classified
into model (mathematical equation) uncertain-
a. Systematic errors: May be due to the ty and parameter uncertainty. Model uncertain-
methods (or models) used to quantify ty refers to the uncertainty associated with the
emissions for the process under mathematical equations (i.e., models) used to
consideration. characterize the relationships between various
a. Random errors: May be due to natural parameters and emission processes. Parameter
variability of the process that produces the uncertainty refers to the uncertainty associated
emissions. with quantifying the parameters used for the
calculations (e.g., activity data, emission factors
or other parameters).

1 Emissions inventory examples are provided in Chapter 8 of the API Compendium of Greenhouse Gas Emission Methodologies
for the Oil and Natural Gas Industry (API, 2009) and the relative uncertainty of the various sources is addressed in Appendices G
and H of this document.
13
IPIECA • API

Uncertainties due to models (or equations) Uncertainty estimates are tools that enable
depend on the proper application of the inventory preparer to assess the major
quantification methods for the respective contributing factors to the emissions inventory
source categories. These errors typically can and target the most significant ones, i.e., those
be eliminated as far as possible in advance, exhibiting the largest range of uncertainty for
when planning the compilation of an emissions more research and refinement. Table 2-1 provides
inventory, and are often addressed as part of an overview of selected methods recommended
emission inventory assurance processes (API/ by the U.S. Environmental Protection Agency
IPIECA/OGP, 2011). (EPA) for qualitative and quantitative estimation
of the ranges of uncertainty for emissions. The
Parameter uncertainties can be evaluated
uncertainties associated with natural variability
through statistical analysis, measurement
inherent to the emission process and its
equipment precision determinations, and
underlying data can be assessed by the statistical
a variety of expert judgment techniques.
analysis methods introduced in Section 5 and
Quantifying parameter uncertainties,
discussed further in Appendix F.
and then estimating source category
uncertainties based on these parameter If measured parameters uncertainties are
uncertainties, is the focus of this document. unknown, a pedigree matrix approach can
Hence, adhering to appropriate sampling, be used to calculate uncertainties. In this
measurement and estimation procedures – approach, which is a method for structuring
with applicable quality control and quality the expert judgment, qualitative data quality
assurance measures – are all part of a quality assessment results are used to relate data
improvement management system and can quality indicators to uncertainty ranges for
help minimize uncertainties. individual parameters (WRI/WBCSD a, 2011).

Table 2-1. Overview of methods used to estimate emissions uncertaintya

Level Of
Methodology Description Of Method
Effort

Qualitative Discussion
-- Sources of uncertainty are listed and discussed.
Low
-- 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 emission
Low
Quality Ratings factor or parameter.
Data Attribute Rating -- Numerical values representing relative uncertainty are assigned through objective
Medium
System (DARS) methods.
-- Experts estimate emission distribution parameters (i.e. mean, standard deviation,
and distribution type).
Expert Estimation
-- Simple analytical and graphical techniques are then used to estimate confidence
limits from the assumed distributional data. Medium
Method
-- In the Delphi method, expert judgment is used to estimate uncertainty directly.
-- In the Pedigree Matrix method, experts are used to set up the appropriate matrix for
estimating data quality
-- Emission parameter means and standard deviations are estimated using expert
Propagation of Errors judgment, measurements or other methods.
Medium
Method -- Standard statistical techniques of error propagation typically based upon Taylor’s
series expansions are then used to estimate the composite uncertainty.
-- Monte Carlo, Latin hypercube, bootstrap (resampling), and other numerical methods
are used to estimate directly the central value and confidence intervals of individual
Direct Simulation emission estimates.
High
Method -- 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 indirect field measurements of emissions are used to compute emissions
and emissions uncertainty directly.
Direct or Indirect
Measurement
-- Methods include direct measurement such as stack sampling and indirect
High
measurement such as tracer studies.
(Validation) 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,” EPA, Research Triangle Park, NC, July 1996
14
 ADDRESSING UNCERTAINTY IN OIL & NATURAL GAS INDUSTRY GREENHOUSE GAS INVENTORIES

Expert judgment on methodology selection In the pedigree matrix approach, data

and choice of input data is the basis of all sources may be assessed according to five
inventory development. When developing data quality indicators, which comprise
an emissions inventory, the goal of expert precision and completeness, along with three
judgment is to ensure: representativeness considerations (temporal,
geographical, and technological). Each of these
• choice of the proper methodology;
indicators is assigned one of four data quality
• selection of applicable emission factors criteria (very good, good, fair, and poor). Exhibit
and their uncertainty from ranges; 2-1 describes the five data quality indicators
• identification of the use of relevant that are useful for assessing the uncertainty of
activity data; emission inventories.
• formulation of the most appropriate way Further discussion on the use of the Pedigree
to apply a given methodology; and Matrix approach and an example application
• determination of the appropriate mix of for petroleum and natural gas greenhouse gas
technologies represented. inventories are provided in Sections 5.1.3 and
5.3.2, respectively.

EXHIBIT 2-1: DATA QUALITY INDICATORS

• Precision – relates to the repeatability of measurements as defined by the error
estimate or the spread of the measurements cluster.
• Completeness – relates to the statistical properties of the data: how representative
is the sample, does the sample include a sufficient number of data points and is
the period adequate to even out normal fluctuations. Completeness relates to the
properties of the sample itself.
• Temporal Representativeness – relates to the time correlation between the year of
data collection and the inventory year. This is an essential consideration for industry
segments with rapid technology development, which might cause the emissions and
the production efficiency to be totally changed.
• Geographical Representativeness – relates to the spatial match between the area
where data are collected and the area included in the inventory.
• Technological Representativeness – relates to the correlation of the data used in
the inventory to the data of the specific entity, processes, and materials that lead to
emissions.

2.2 Sources of measurement spatial or profile uncertainty when using

uncertainty distinct measurements to represent similar
processes at different times and locations.
The measurement process is comprised
of different steps and each can introduce 2.3 Emission calculation approaches
uncertainty into the final results, where each
of these steps may be associated with either Emissions information typically is obtained
activities or emission factors. either through direct on-site emission
measurements, or by using engineering
The sources of uncertainty summarized in emission techniques that are based
Table 2-2 range from methods choice to on appropriate equations, models, or
physical constraints of the measurement emission factors that describe the physical
process itself. Uncertainty may also be process. Emission estimates are used
introduced due to errors in the processing for facility permitting, development of
of collected data such as operating control strategies, compliance review,
conditions or other constants used for the and demonstration of attainment of
calculation. Similarly, they may be due to environmental goals.

15
IPIECA • API

Table 2-2. Summary of typical sources of measurement uncertainty

Sources of
Description
Uncertainty
Some common sources of measurement system uncertainties include:
-- Improper placement of monitoring device or extraction of unrepresentative samples.
Measurement -- Environmental effects on measurement instruments, such as heat transfer effects on a temperature probe, or
methods pressure considerations for flow measurements.
-- Drift of an instrument between successive calibrations.
-- Electrical interference with electronic components.
-- Variation between the calibration and usage conditions.
Measurement instruments are typically calibrated before they are used in the field or in a plant:
-- Instrument calibration needs to be checked periodically to detect instrument drift and reduce measurement
Calibration uncertainty.
-- The calibration process ought to be traceable to a known reference standard.
-- Allowance should be made for adjustments of the measurement instrument if a bias is detected during the
calibration process.
Uncertainties in data acquisition systems depend on system design:
-- For manual data collection, and more specifically data entry, human error can be a factor.
Data acquisition -- 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.
Uncertainty in data processing may be attributable to multiple sources:
-- Model equations for simulating process emissions may be oversimplified and not adhere to physical realities.
Data Processing -- Coefficients of regression for instrument calibration indicate scatter of the measurement data.
-- Use of non-uniform sets of conditions and improper unit conversions, introduce bias to the calculated values.
-- Improper uploading of electronic data may introduce bias.
Quality control techniques and robust data management practices can minimize the effects of many sources of
uncertainty:

Data Quality -- Comparing known input values with their measured or computed results can provide an estimate of the data
acquisition uncertainty.
-- When it is not possible to do this in practice, it is advisable to evaluate potential individual and aggregated
errors when assessing uncertainty.

The general equation for quantifying emissions is:

Emissions = Activity Rate × Emission (conversion) Factor (Equation 2-1)
When calculating emissions and their associated uncertainties, it is important to note that the
overall uncertainty is based both on the variability associated with the activity data (process flow,
throughput, usage or equipment count) and on the accuracy of the emission factors used.
In practice, for estimating GHG emissions from the oil and natural gas industry this means:

• For CO2 (combustion)

Emissions = Volumetric × Carbon Composition × Combustion (Equation 2-2)
Gas Flow Efficiency

– or –

Emissions = Fuel Energy × Carbon per Heating × Combustion (Equation 2-3)

Consumption Value Unit Efficiency

• For CH4 (non-combustion)

Emissions = Component or Event Count × Emissions (Conversion) Factor (Equation 2-4)

16
 ADDRESSING UNCERTAINTY IN OIL & NATURAL GAS INDUSTRY GREENHOUSE GAS INVENTORIES

The four basic approaches for estimating emissions be used broadly across many industry source
and their potential error sources are summarized in categories and operations. The published
Table 2-3. Each contributor to uncertainty should emission factors are typically accompanied by
be assessed independently and then aggregated a description of the group of processes and the
in the final analysis, as discussed below. conditions they represent. Authoritative factors
are generally published by the EPA in AP-42
In the absence of direct emission measurements,
(EPA, 1995 and further updates), or by the EU
the use of emission factors has traditionally
EMEP/CORINAIR Emission Inventory Guidebook
been the simple, low cost method of choice for
(EEA, 2013). The IPCC has also launched a new
estimating emissions. Over the years, the EPA
Emissions Factors Database (IPCC, EFDB) for use
and other emission factor repository databases
with GHG emission calculations.
have provided average emission factors that can
Table 2-3. Common emission estimation approaches and their error sources
Emissions Estimation
Description
Approach
An emission factor relates the rate of emission of a specific compound to an activity
rate associated with its release.
The errors associated with the use of emissions factors are attributed to:
-- Variations in operating conditions during the collection of emission and activity
data that were used in the development of the emission factors.
Emissions Factors -- Variability in emissions that arises from differences in operating conditions
among different facilities where the factors are used.
-- Difference between the actual composition of the stream to which the
emission factor is applied and the default composition on which the emission
factor is based.
-- Uncertainty that is due to measurement errors, systematic errors, and random
sampling errors.
The 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 applicable monitors at the source.
Uncertainty associated with CEMS generally includes:
Continuous -- Stack flow measurements.
Emissions Monitoring -- Concentration measurements.
System (CEMS) -- Measurements of stack temperature and pressure required to correct emissions
to standard conditions.
The error associated with these determinations varies for different compounds.
CEMS will not necessarily produce emission estimates with lower uncertainties than
alternative methods especially since CEMS are not available for monitoring all GHG
emissions. CEMS may not be practical for a large number of emission sources.
This technique involves either extracting a sample or placing a monitor at a
source, followed by analysis to characterize the emitted species. The measurement
campaign is typically limited to a specified number of hours, and the average
emission rates calculated are used 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.
Source Testing The errors associated with these measurements are due to:
-- Sporadic testing of limited duration that is used and extrapolated to
characterize emissions over a longer period, i.e., a year, not allowing for a
robust estimate of variability.
-- Testing methods used are either improperly calibrated or do not have Version
2, September 2014 2-7 sufficient sensitivity to enable characterization of the
full gamut of emission rates.
This technique is based on the material balance or total quantity of material used,
e.g., fuel flow into a combustion device and its carbon speciation.
Material Balance The errors associated with this method include:
-- Improper calibration and measurement of applicable flow devices.
-- Inaccurate determination of material stream composition.

17
IPIECA • API

2.4 Inventory steps and data total emissions for each of the GHG species,
aggregation along with the global warming potential-
weighted sum of these emissions (also known
When developing GHG emission inventories, as the CO2e emission).
emission estimates are obtained from many
intermediate and independent results, each The overall uncertainty range for each GHG
of which is calculated from a separate set of species, and CO2e, should also be reported with
data that is characterized by a different range the total emissions in the format of:
of uncertainties. The compilation of an entity- Emissions = Average Value ± %
wide GHG emissions inventory typically follows (at the 95% confidence limit).
a sequence of steps:
(Equation 2-5)
a. Establishing boundaries – Where the
organizational and operational boundaries
are defined (for a first-time inventory), 2.5 Emissions inventory and
or examined (for recurring cycles), this uncertainty assessment in the
step will be largely dictated by local petroleum and natural gas industry
requirements or corporate policies. It
might involve facility-by-facility assessment Inventorying of GHG emissions by entities is
prior to aggregation, or it could use other a ‘bottom-up’ summation of emissions from
pertinent entity indicators and information. individual sources (or emissions from the total
consumption of different fuel types) at a report-
b. Collecting and inputting data – Where ing unit to create an inventory for that report-
the activity data are collected and archived ing unit. Emissions from disparate reporting
based on the boundaries established units may be aggregated to create an entity,
above. The data are then incorporated into or corporate, inventory. Reporting units may
appropriate calculation tools for emission be defined by the reporting entity to represent
calculations. The level of ‘granularity’ of logical groupings of activities and assets, or
the data collected and the details of the could be mandated by governments as part
calculation methods are dictated by local of GHG reporting regulations. Developing
requirements with industry guidance (API these GHG inventories entails both the proper
Compendium) as a resource to provide accounting of activity levels for operations that
relevant technical details. may lead to GHG emissions as well as the con-
c. Validating data compiled – Where various version of these activity levels to quantitative
techniques are used to compare the new GHG emissions using proper measurement and
data with earlier versions (if available) to calculation methods.
identify potential large errors. These errors
could include: either large changes or The petroleum and natural gas industry en-
unchanged activity data for given facilities; compasses a wide variety of activities, rang-
operations that are not accounted for; lack ing from the exploration and production of
of supporting data for measurements or petroleum and natural gas to the delivery of
emission factors used; erroneous units or products to consumers. As part of defining the
unit conversions among others. scope of GHG inventories, industry guidelines
(IPIECA/API/OGP, 2011) are available to assist
d. Assessing data uncertainty – Where the companies in determining which emissions
confidence intervals associated with the related to their activities should be included
data available for each of the emission within the organizational boundaries they
sources are characterized independently, have established. In all cases, mandatory GHG
as discussed later in this document. The reporting requirements would take precedence
uncertainty information could be based over industry guidelines, as may be the case in
on documentation of data repositories different jurisdictions.
(API, 2009), expert judgment, or on
measurements conducted during the For a complex and dynamic sector such as
inventory year. the oil and natural gas industry, a variety of
e. Finalizing the inventory – Where the methods are applicable to quantifying GHG
quality-checked and validated data are emissions, ranging from simple activity mea-
aggregated for reporting based on company surements multiplied by applicable emission
policy or local requirements. The preferable factors to more sophisticated quantification
way of reporting the results is in terms of the algorithms. Advanced engineering estimation

18
 ADDRESSING UNCERTAINTY IN OIL & NATURAL GAS INDUSTRY GREENHOUSE GAS INVENTORIES

methods consist of an integrated approach Figure 2-1. Hierarchy of emission estimation approaches
that relies on the use of emissions and physical
activity factors in conjunction with generic Types of Approaches Hierarchy
process simulation models, source-specific
models, and species profiles databases. Each of
these estimation methods will have a different Published emission factors
level of uncertainty. For example, combustion
emissions are dependent on the amount of fuel
Equipment manufacturer
consumed. This can be estimated based on the
emission factors
hours of equipment operation, equipment rating, Improved accuracy
and thermal efficiency. The uncertainty associated
with this approach differs from the uncertainty Engineering calculations Additional data requirements
associated with directly measuring fuel use. Higher cost
Monitoring over a range
Selection of an appropriate emission estima-
of conditions and deriving
tion method must consider:
emission factors
• The objectives of the emission inventory;
• The contribution of the emission source Periodic monitoring of
to the overall accuracy of the emission
emissions or parameters for
inventory; and
calculating emissions
• The costs and practicality of the emission
estimation method.
Continuous emissions* or
Figure 2-1 provides a general hierarchy of accu-
parameters monitoring
racy associated with common emission estima-
tion approaches. Accuracy is generally improved *Continuous emissions monitoring may not be directly
as emission estimation progresses from default applicable to certain greenhouse gases or to all emission
emission factors to more direct estimation or sources.
measurement methods. For example, measuring from key emission sources associated with
CH4 and CO2 emissions from every dehydrator the natural gas storage and transmission
in a natural gas production field may produce sector of the industry (INGAA, 2005) and
a highly accurate emission rate, but at great the American Gas Association (AGA), which
expense for a source that has minimal contribu- has also published specific guidelines for
tion to the GHG inventory of the field and can estimating GHG emissions from gas distribution
be estimated using an engineering modeling operations (AGA, 2008).
approach, such as GLYCalc™ (GTI, 2000).
Industry guidelines also served as major input
to the formulation of regulations governing
The third edition of the API Compendium (API, the mandatory reporting of GHG, such as the
2009) provides an extensive compilation and U.S. GHG Reporting Program (EPA, 2009), the
tabulation of methods used by companies Western Climate Initiative (WCI, 2009), and the
in all segments of the petroleum and natural California Air Resources Board (CARB, 2012).
gas industry for consistent calculation of GHG
emissions. The API Compendium provides The categories of sources that contribute to
a wide range of emission factors and other overall GHG emissions as classified in the API
emission estimation methods that are directly Compendium (API, 2009) are:
applicable to all sectors of industry operations. • Stationary combustion sources – linked to
Summary tables and decision trees are emissions resulting from the combustion of
provided to illustrate the variety of available fuels in boilers, furnaces, burners, heaters,
emission estimation options and the associated and stationary turbines and engines. This
considerations. It also lists the uncertainty category may also include the combustion
ranges (at the 95% confidence level) associated of waste gases-or emergency releases-in
with multiple case study examples featured in incinerators and flares.
Section 8 of the 2009 API Compendium.
• Mobile combustion sources – linked to
Related industry guidelines include those emissions resulting from combustion
of the Interstate Natural Gas Association of of fuels in ships, barges, trains, trucks,
America (INGAA), which provides supplemental automobiles, aircrafts and other off-road
guidance for estimating GHG emissions devices such as drilling rigs.

19
IPIECA • API

• Process Vents – linked to the emissions composition of these self-generated fuels

that result from physical or chemical vary with the nature of the producing
processing of materials within the formations, while for refining it depends on the
petroleum industry. The materials composition of the feed slate and the diversity
may include both gaseous and liquid of refinery process units that determine the
hydrocarbon streams that contribute to products manufactured. On the other hand,
both CO2 and CH4 emissions. for natural gas transmission and distribution
• Equipment Leaks – linked to leakage from operations, gas quality and its composition are
piping components such as from seals, expected to adhere to contract requirements
gaskets, valves, packing rods, emergency and would vary only within a narrow
relief devices and similar items. specifications range. Hence, the use of average
fuel compositions data has to be evaluated
It is important to note that under global
when compiling an emission inventory since
practices adopted by the IPCC for national
it might result in wide uncertainty ranges for
GHG inventories, the aggregate of all non-
some segments and operations, while it might
combustion emissions are referred to as
be perfectly acceptable for others.
“fugitive emissions.” Under those practices,
even flares are included within the “fugitive A different set of parameters is important for
emissions” category. Confusion might arise understanding emissions associated with
since under the U.S. Clean Air Act the term process vents and fugitive emissions due to
“fugitive emissions” refers to equipment leaks equipment leaks. For many of the large process
only, while the U.S. GHG Inventory adheres to units in refineries and natural gas processing
the IPCC format and its broader definition of plants, numerical models (equations) are
“fugitive emissions”. available for estimating these emissions. For
high-pressure pipelines transmitting natural
Industry operations vary widely among
gas over long distances, the main GHG
operating segments due to the nature of the
emissions are due to reciprocating compressor
operations, their geographical locations and
engines and turbines, venting due to gas blow-
local practices. Operations in some of the
down, and fugitive emissions associated with
industry segments are highly centralized in
leaking piping components. For low-pressure
large and complex facilities while in others
gas distribution, most of the GHG emissions
they extend over large geographical areas,
result from compressors and leaks from gas
with some of the operations not contained in
distribution mains and associated equipment.
traditional “facilities”. Additionally, company
operations tend to encompass many Quantifying emissions, and their associated
jurisdictions, which add to the complexity uncertainty ranges, for venting and equipment
of compiling an emission inventory, even leaks in the exploration and production
for a given corporate entity. Data availability segment poses a real challenge. These
may be different among industry segments emissions vary widely in magnitude depending
and operating regions due to an operational on the characteristics of the producing
considerations and local requirements. formation but they are generally attributable
to workover and completions, gas well liquids
The uncertainty associated with quantifying
unloading, venting from pneumatic controllers,
CO2 emissions from combustion would be
flashing from condensate and oil storage tanks
primarily attributable to variation in the
and more.
composition of combusted fuels and their
respective consumption rates (or total Operators in the USA, as well as in other
volumes). For quantification of combustion jurisdictions, maintain required records for
emissions, quality data are typically available reporting (and archiving) inadvertent gas-
for industry facilities in all segments, though venting incidents. USA reports of “lost and
significant effort may be required to collect unaccounted for gas” from natural gas pipelines
data for smaller operating installations that are account for both vented and fugitive emissions,
geographically dispersed. as well as metering errors. Therefore, it will be
necessary to disaggregate the data in order
Since a large fraction of industry operations
to derive a separate average emission factor
rely on self-generated fuels, it is the knowledge
for venting incidents only. When it comes to
of the carbon content – or heating values – of
fugitive emissions from equipment leaks, the
such fuels that is at the root of determining
uncertainty associated with current practices is
their associated CO2 emissions. For the
significant. The most reliable emission factors
exploration and production segment, the

20
 ADDRESSING UNCERTAINTY IN OIL & NATURAL GAS INDUSTRY GREENHOUSE GAS INVENTORIES

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 gas
emitted 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 from
combustion – 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 frequency of venting and its duration
under normal operations, along with the
methane content of the emitted gas, is just
now starting to be measured or estimated
routinely in some jurisdictions (e.g.,
GHGRP, 2010). For inadvertent venting and
methane releases, there is no uniformity
of record-keeping requirements 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. Uncertainty is also
introduced by 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.

21
IPIECA • API

3. OVERVIEW OF
SECTION FOCUS
MEASUREMENT PRACTICES FOR
ESTIMATING GREENHOUSE GAS
EMISSIONS This section provides an overview
of measurement practices focusing
In defining aggregated uncertainty of on gas flow measurements and the
measurement ensembles used for developing determination of carbon content or
emission inventories, the uncertainty for each the heating values of combusted
measurement stream must be assessed in a fuels. The section recognizes industry-
way that is applicable to that measurement recommended practices and standards
method and its implementation in practice. including those that traditionally applied
Random errors could be a major factor in to “custody transfer” of fuel products and
the uncertainty of an individual observation; new ones that have been more recently
however, their contribution to the overall developed for use in an emissions
emission inventory diminishes as more estimation context.
measurements are obtained during the
reporting period. This section goes on to link these
practices to the acquisition of GHG
Note: Random errors tend to average out emissions and related activity data,
during the year, whereas systematic errors (or representing a move from reliance on
measurement bias) become more important and available data or engineering judgment.
tend to accumulate rather than diminish over The subsections address:
longer periods of time such as a year.
• Flow Measurement Practices;
In fact, determining the true value of any
measured variable is not practical due to the
• Flow Measurements for GHG
Inventories;
limitations of measurement equipment and
procedures, and the possibility of human error. • Sampling and Analysis for
Hence, industry measurement procedures and Quantifying GHG Emissions;
standards have been developed to emphasize • Carbon Content Measurement
practices that lead to collecting better quality Practices;
data, especially for critical measurements. • Heat Content Determination;
Industry uses standards from several different
standards-developing organizations, resulting
• Venting and Fugitive Emissions
Measurements; and
in equivalent measurements, based on the
scope of the standard, company preference, • Laboratory Management System.
and type of devices used. Consensus industry
standards, such as those developed by ANSI,
API, ASTM, ISO and other standard-setting of specific measurement standards (and their
organizations, have rigor in their development respective editions) from several standards-
process, and measurement standards are setting organizations, which could be used
reviewed at least every five years to ensure to support the calculation of GHG emissions.
that standards are in step with technological It is up to the user of these measurement
changes and advancements. Most, if not all, standards to reference the specific standards/
of the measurement standards are developed editions used for a given measurement, and
for measurements associated with ‘custody to incorporate the updated measurement
transfer’ and to define quantities that are procedures, as applicable.
essential for robust financial transactions. Custody transfer measurements are typically
API publishes one of the more comprehensive are expected to meet a set of performance
sets of custody transfer measurement specifications. For other measurements,
standards, but it is neither complete nor the such as those performed to support the
only widely recognized source for such industry development of a GHG emissions inventory,
practices. API’s MPMS (Manual of Petroleum data quality objectives should be established
Measurement Standards) includes over 140 prior to initiating any data collection to
titles, and API publishes approximately eight ensure that the uncertainty ranges of the
new or revised measurement standards each measured quantities are consistent with the
year. Appendix B presents a comprehensive list intended use of the data. Throughout this
section, we provide references and describe

22
 ADDRESSING UNCERTAINTY IN OIL & NATURAL GAS INDUSTRY GREENHOUSE GAS INVENTORIES

a select subset of industry standards that 3.1 Flow measurement practices

are most typically used for the respective
measurements discussed. Even so, the full Continuous handling of very large liquid and
list of measurement methods provided gas flow volumes is a characteristic of all the
in Appendix B could be used to provide sectors of the petroleum and natural gas
equivalent measurements to meet company industry. Therefore, an in-depth understand-
practices and available instrumentation. ing of flow measurement is essential both for
internal process control and for transferring
For many petroleum and natural gas industry “custody” of intermediate streams or finished
installations, CO2 emissions from combustion products. The measurement accuracy of
and flaring are the largest contributors to “custody meters” is historically quite high, and
overall GHG emissions. Therefore, this section practices follow rigorous industry standards.
focuses on measurements and methods
typically used for quantification of these CO2 Industry has been instrumental in develop-
emissions. The subsections below provide ing international voluntary standards such
details on flow measurement practices, and as ISO 5168:2005 (ISO, 2005) establishing
their associated uncertainties, as well as general principles and describing procedures
methods for the measurement of carbon for evaluating the uncertainty of measuring
content and heating values of combusted fuels. fluid flow rate or quantity. Annex A of ISO
For some other petroleum and natural 5168 provides a step-by-step procedure for
gas industry sectors, such as exploration, calculating and reporting these measurement
production, processing, transmission and uncertainties. Similarly, Chapter 14 of the API
distribution operations, emissions from other Manual of Petroleum Measurement Standards
GHGs (such as methane) can contribute (MPMS), contains detailed procedures and
significantly to a facility’s GHG emissions. practices for all aspects of natural gas (and
Available emission factors exhibit large similar gases) fluids measurement and calcu-
uncertainties and might not be representative lation of their associated uncertainties, at the
of current operating practices. Therefore, point where oil or gas enters the marketplace
for inventories where CH4 constitutes a (“custody transfer”) (API a, 2006). Those same
larger fraction of the emissions, the overall practices are not as rigorously applied to in-
uncertainty would be expected to be ternal accounting and process control during
substantially higher. normal operations.

Industry is now more fully internalizing the 3.1.1 Measurements by Orifice Meters
potential impact of measurement errors
and bias through the full chain of emission Orifice meters are by far the most prevalent
calculations including measurement flow meter type used in the petroleum
equipment, software calculations, and natural gas industry, and are used for
simulation models, and the limitations of metering products during “custody transfer”
reliance on existing emission factors on the as well as for process control and internal
uncertainty range of resultant GHG emissions accounting. These flow meters are also used
inventories. Considerations of the need for to account for fuel volumes when estimating
more representative measurement data, CO2 emissions. Flow meters are designed
and the assessment of equipment design for long-term reliability and ruggedness
and age, are gaining more prominence in under a variety of component mixtures and
the process of assembling high certainty conditions that are essential for consistent
emission inventories. This section focuses on fluid blending and processing. Although for
measurements that are pertinent to improved this type of meter, temperature and pressure
characterization of calibrations can be done while the units
are operating; they generally have limited
combusted fuels and the quantities used, access to direct orifice plate inspections and
but might not be directly applicable to maintenance outside of planned shutdown
measurement of leakages and fugitive (‘turnaround’) cycles.
emissions. The measurement practices
highlighted here represent an initial step in Recommended practices for the installation,
what could end up being a multi-year effort to calibration, and calculation of flows for
improve measurements of GHGs and quantify these custody meters are provided in
the activities that cause their emissions. Section 3 of Chapter 14 of API’s MPMS (API,

23
IPIECA • API

2013). This standard was developed through • Part 1: General equations and uncertainty
a collaborative effort by members of API, guidelines.
AGA, and the Processing Gas Association, • Part 2:Specifications and installation
with contributions from the Canadian Gas requirements;
Association, American Chemistry Council, the
European Union, Norway, Japan, and others.
• Part 3: Natural gas applications.
It is designed to ensure global consistency for • Part 4: Background, development,
petroleum and natural gas transactions and is implementation procedures, and
recognized by the Version 2, September 2014 3-3 subroutine documentation.The standard
recognizes that many factors contribute to
American National Standards Institute (ANSI) as overall measurement uncertainty associated
an American National Standard. The four-part with many metering applications, as
standard for square-edged, concentric orifice summarized in Exibit 3-1
meters consists of:

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 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
• A 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 typically are 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.

24
 ADDRESSING UNCERTAINTY IN OIL & NATURAL GAS INDUSTRY GREENHOUSE GAS INVENTORIES

All these factors should be assessed when from molecular weights approaching that of
estimating the overall range of uncertainty for hydrogen to molecular weights of C5+.
flow measurements using thin plate, concentric,
As with other flow measurements, the accurate
square-edged metering systems.
determination of flow to flares is dependent on
In the reapproved 2012 version of the standard, many parameters such as the ability to predict –
several changes were incorporated to reduce the or measure – the quantity of gas flared, mixture
uncertainty attributable to installation effects composition, pressure, temperature, and/
and to improve the rigor of the flow calculation or density. For example, gas-to-oil ratio (GOR)
routines. The revised standard recognizes the measurements can be used to estimate the
lead time necessary for upgrading existing volume of gas flared for isolated oil production
installations, and leaves this lead time to the operations, or engineering estimates based on
discretion of facility operators and their data purge rates or operating logs can be used to
quality targets for flow measurement data. estimate flare volumes for refineries or gas plants.
However, it should be recognized that if orifice The accuracy of measurements associated with
meter installations are not upgraded to conform highly variable flare gas mixtures will depend
to the new recommendations, measurement largely on the meter technology type and the
bias error may occur. This bias might be due ability of the flare flow measurement system
to improper upper and lower distances from (FFMS) to achieve the targeted response time
bends and points of flow turbulence that and analytical accuracy levels. Exhibit 3-2
might lead to inadequate flow conditioning below lists the basic steps needed to conduct
prior to measurement. Additionally, even a simplified analysis for determining the
without changing equipment installations, the uncertainty ranges of a given flow measurement
standard recommends adopting new calculation system. The actual approaches for the required
procedures and techniques (explained in calculations are provided in Section 5, and
Part 1 and 3 of the standard) that represent Appendix F provides examples demonstrating
significant improvements over the previously how to apply these methods.
adopted approach. It is important to note
that the expected uncertainty ranges for flow
measurements quoted in Part 1 of the reaffirmed EXHIBIT 3-2: STEPS FOR SIMPLIFIED
standard may differ from those obtained in UNCERTAINTY ANALYSIS
practice when the equipment installation differs.
a) Determination of the equation that defines the
meter output
3.1.2. Measurement of flow to flares • Governing equations are those applicable for the
meter technology type used.
The measurement of flow to flares is distinctly
different than other flow measurements. b) Determination of the combined sensitivity
Flares are designed as safety relief systems coefficient
and typically are capable of handling highly • Numerical values of uncertainty are associated with
variable flow rates of widely varying gas pressure, temperature, and composition.
compositions. Therefore, some of the practices • Meter accuracy is estimated from calibration data,
that are generally applicable to custody transfer pipe size, or other installation effects.
or process control flows have to be modified • Sensitivity coefficients are obtained by dividing the
when addressing flows to flares. API published calculated percent uncertainty for an input variable
a measurement standard addressing gas or by the percent change in that input variable.
vapour flare flow measurements, which also c) Derive the combined uncertainty range (extended
includes cautionary details about the effects standard deviation)
of fouling (due to entrained liquid droplets, • Combined uncertainty is calculated by summing
aerosol mists, or other contamination) on the the square of the errors for pressure, temperature,
measurement (API MPMS, July 2007). composition, meter calibration, and installation
effects.
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

25
IPIECA • API

Variability in flare composition may also Each component in the measurement system
be a significant factor in determining the will exhibit variations that contribute to
measurement uncertainty of an FFMS. variations in the process result. Variability
Knowledge of flare composition may have a may be due to procedure(s), operator(s),
major effect on the calculation of the actual environment, instrument(s) and maintenance.
volume, standard volume, or mass measured by Consequently, repeated results from the same
the meters. For example: measurement process may vary, and there is
always the potential that many small variations
• For a differential pressure meter: the
will affect the final results.
output is a function of the square root of
the flare gas density. Therefore, it is advisable to analyze the overall
• For thermal flow meters: knowledge uncertainty of the flow measurement process
of the actual volume (or standard to affirm its adequacy for the intended
volume) requires consideration of application. As indicated in Exhibit 3-2, a few
the compositional effect on thermal key steps should be followed to determine
conductivity and dynamic viscosity. the combined uncertainty associated with
individual flow measurements. An expanded
• For ultrasonic flow meters: sound speed discussion of factors to be considered
is a function of gas composition.
when evaluating the uncertainty of flow
Converting volume flow to mass flow requires measurements used for GHG emission
knowledge of gas composition in order calculations is included in Section 3.2.
to derive gas density. When an analyzer is
incorporated in the measurement system to
correct for composition, care must be taken to 3.1.4 Uncertainty specifications for “custody
ensure that the response time of the system is transfer” measurements
short compared to the upset flow event during
flaring to ensure representative measurement “Custody transfer” measurements are defined
during actual flaring. as measurements that provide quantity and
quality information, which can be used as
An example of how all of these elements would the basis for a change in ownership and/or
be combined into the overall measurement a change in responsibility for materials. In
uncertainty is provided in Table 3-1. most petroleum and natural gas producing
jurisdictions around the world, national
regulations and directives have emerged
3.1.3 Flow measurements uncertainty to specify requirements for the expected
analysis accuracy and uncertainty ranges associated
The result of a measurement process is the with “custody transfer” and other critical
determined quantity of the parameters measurements. For such precise metering
measured. For flow measurements the applications, the flow meters and adjacent
quantities may include pressure, temperature piping used in the measurement system
and volumes, (i.e., multiple components) are expected to meet the requirements of
contribute to the accuracy, or uncertainty, of the relevant, preferably the most stringent,
the final result. specifications of the API and ISO standards
that are cited in many national regulations.
Table 3-1. Example of flare flow measurement systems combined
uncertainty a One such example of measurement
requirements promulgated for petroleum
Combined
Variable Sensitivity and Error (S x U95)2 and natural gas operations is Directive
(S x U95) 017 of the Alberta Energy Regulator (AER,
2013), which spells out the measurement
Pressure 2.0% 4.00
requirements for upstream oil and gas
Temperature 0.1% 0.01 operations in Alberta, Canada:
Flare Composition 2.0% 4.00
• The standards are stated as “maximum
Meter error (calibration 1.4% 1.96 uncertainty of monthly volume” and/or
Installation effects 0.5% 0.25 “single point measurement uncertainty,”
Sum of squares 10.22 as listed in Table 3-2 below for petroleum
(oil) Systems and Natural Gas Systems
Square root of sum of square 3.2%
measurements, respectively.
a Table 4 from API MPMS Section 14.10 (API, 2007)

26
 ADDRESSING UNCERTAINTY IN OIL & NATURAL GAS INDUSTRY GREENHOUSE GAS INVENTORIES

Table 3-2. Summary of Alberta ERCB accuracy requirementsa

Maximum Single Point

Uncertainty of Measurement
Monthly Volume Uncertainty
PETROLEUM SYSTEMS MEASUREMENTS
(i) Total battery oil (delivery point measurement)
Delivery point measures >100 m3/d N/A 0.5%
Delivery point measures < 100 m /d 3
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 m /d but < 16.9 103 m /d
3 3
10% 3%
< 0.50 103 m /d
3
20% 10%
(iv) Well oil (proration battery)
Class 1 (high), > 30 m3/d 5% 2%
Class 2 (medium), > 6 m /d but < 30 m /d
3 3
10% 2%
Class 3 (low), > 2 m /d but < 6 m /d
3 3
20% 2%
Class 4 (stripper), < 2 m /d
3
40% 2%
(v) Well gas (proration battery)
> 16.9 103 m3/d 5% 3%
> 0.50 103 m /d but < 16.9 103 m /d
3 3
10% 3%
< 0.50 103 m /d
3
20% 10%
GAS SYSTEMS MEASUREMENTS
(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 m /d 3
N/A 1%
(iii) Plant inlet or total battery/group gas 5% 3%
(iv) Plant inlet or total battery/group condensate N/A 2%
(recombined)
(v) Fuel gas
> 0.50 103 m3/d 5% 3%
< 0.50 103 m /d
3
20% 10%
(vi) Flare gas 20% 5%
(vii) Acid gas before compression N/A 10%
Acid gas after compression 3%
(viii) Dilution gas 5% 3%
(ix) Well gas (well site separation)
> 16.9 103 m3/d 5% 3%
< 16.9 103 m /d 10% 3%
(x) Well gas (proration battery) 15% 3%
(xi) Well condensate (recombined) N/A 2%

Note: Extracted from Section 1.8.1 and 1.8.2, respectively, ERCB Directive 017, September 2012
a

27
IPIECA • API

• The uncertainties are to be applied as the manufacturers’ recommendations for the

“plus/minus” (e.g., ±5%), and only mea- installation, calibration, and maintenance of
surements at the delivery or sales points these flow meters. The error levels cited refer
are required to meet the highest accuracy primarily to random errors that are observed
standards since they would have a direct under ‘ideal’ laboratory conditions and that
impact on royalty determination. decrease with repeated measurements. They
The directive makes it clear that other do not properly account for systematic errors
measurement points that play a role in the (or bias) where the errors are due to improper
overall control and accounting process would installations, inadequate calibrations, or device
be subject to less stringent accuracy standards. drift, as discussed above.
These less stringent accuracy standards are However, these manufacturers’ specified
designed to accommodate physical limitations measurement errors might not be attainable
and/or the overall economics of achieving due to the practical operational limitations
very stringent accuracy standards for each of the facility. In most petroleum and natural
volumetric measurement.specifications of the gas industry facilities, detailed inspections,
API and ISO standards that are cited in many maintenance, and recalibration of process
national regulations. control flow meters are possible only once
every few years when process units are shut
down for scheduled turnaround.
3.2 Flow measurements for GHG
inventories
Measuring gas or liquid flow rates, or their 3.2.2 Fuel gas measurement
total volume for calculation of GHG emissions, API Technical Report 2571 (API, 2011) provides
requires that the flow meters used should a performance-based methodology for the
be fit for purpose. They should be properly measurement and reporting of fuel gas
installed and calibrated, to ensure that they consumption. Specifically, considerations are
are capable of providing data that are within provided for measurement device selection,
the uncertainty ranges required by the installation, maintenance, calibration and
governing climate program. Consideration documentation to achieve the targeted
should be given to differences between the performance in terms of availability and
manufacturers’ specifications of flow meters’ uncertainty. If the performance of any installed
expected measurement errors and those that measurement device is determined not to
are attained when using the flow meters in be in compliance with an acceptable level
the field. It is common practice to test flow of uncertainty, the measurement device, its
meters in a laboratory setting under controlled installation, or maintenance practices, etc.,
conditions prior to field installations. However, can be upgraded. Techniques are described to
these laboratory bench tests typically do not assess the uncertainty contribution of individual
simulate “real world” variations in fluid flow components of fuel gas measurement systems
and other possible fluctuations, and drift of and the overall facility fuel gas measurement
the entire measurement system. For any given uncertainty.
operating facility, only a very limited number of
“custody transfer” meters are equipped for field Following the guidance and calculation
calibration under real operating conditions. procedures of this TR 2571 provides for achieving
cost-effective fuel gas measurements of
appropriate quality. In most cases, the rigorous
3.2.1 Flow meter types requirements of industry standards intended
for custody transfer quality measurements
As an example, Table 3-3 provides a listing can be reduced and still achieve the desired
of different meter types, their applicable measurement uncertainty.
fluid medium, and a brief description of
their operating principles. The table also TR 2571 addresses the most common fuel gas
lists manufacturers’ specified instrument measurement devices in use and contains:
errors, as provided in a survey conducted • A brief description of the working
by the EU-ETS Support Group (ETSG, 2007). principles of different types of fuel gas
The information provided is an indication of meters and their influence parameters;
potential error ranges and not the expected • Installation recommendations;
uncertainty ranges obtained in practice.
Appendix B provides additional details about • A uniform method to ascertain the
measurement uncertainty;
28
 ADDRESSING UNCERTAINTY IN OIL & NATURAL GAS INDUSTRY GREENHOUSE GAS INVENTORIES

Table 3-3. Compilation of specifications for common flow metersa

Manufacturers’
Meter Type Medium Technical Description
Reported Errorsb
Rotary meter Gas The rotary flow meter is a type of positive displacement (PD) 0-20% of the
flow meter that is widely used for utility measurements of gas measurement
(Expected life
flow. range: 3%
span: 25 years)
Rotary flow meters have one or more rotors that are used to 20-100% of the
trap the fluid. With each rotation of the rotors, a specific amount measurement
of fluid is captured. Flow rate is proportional to the rotational range: 1.5%
velocity of the rotors. Rotary meters are used for industrial
applications.
Turbine flow Gas Turbine flow meters have a rotor that spins in proportion to flow 0-20% of the
meter rate. Many of those used for gas flow are called axial meters. measurement
Axial turbine meters have a rotor that revolves around the axis of range: 3%
(Expected life
flow. Axial meters differ according to the number of blades and
span: 25 years) 20-100% of the
the shape of the rotors. Turbine meters are used as billing meters
measurement
to measure the amount of gas used at commercial buildings and
range: 1.5%
industrial plants.
Bellows Gas The bellow gas meter performs volumetric measurement via its 0-20% of the
meter bellows. The measurements are based on the principle that the measurement
flexible bellows are periodically filled and emptied. range: 6%
(Expected life
span: 25 years) A major problem with the bellows system is the residue in the 20-100% of the
pipe. The internal mechanisms fail to perform their tasks due range: 4%
to such residue, causing the meter to dysfunction and fail to
perform a sound measurement.
Orifice meter Gas and Orifice meters belong to the category of differential pressure 30-100% of the
Liquid flow meters that consist of a differential pressure transmitter, measurement
(Expected life
together with a primary element, such as the orifice plates. range: 1.5%
span: 30 years)
The orifice plates place a constriction in the flow 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.
Venturi meter Gas and Venturi meters are another example of differential pressure flow 20-100% of the
Liquid meters, as described under orifice meters above. measurement
(Expected life
range: 1.5%
span: 30 years) In this case, the primary element is a Venturi flow nozzle. Venturi
meters are especially suited to high-speed flows. They are also
used for custody transfer of natural gas.
Ultrasonic Gas and There are two main types of ultrasonic flow meters: transit 1-100% of the
meter Liquid time and Doppler. The transit time meter has both a sender measurement
and a receiver. It sends two ultrasonic signals across a pipe range: 0.5%
(Expected life
at an angle: one with the flow, and one against the flow. The
span: 15 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.

29
IPIECA • API

Table 3-3. Compilation of specifications for common flow metersa (continued)

Manufacturers’
Meter Type Medium Technical Description Reported
Errorsb
Coriolis Gas and Coriolis flow meters contain one or more vibrating 1-100% of the
meter Liquid tubes. These tubes are usually bent, although maximum
straight-tube meters are also available. The fluid to measurement
(Expected
be measured passes through the vibrating tubes. It range: 1%
life span: 10
accelerates as it flows toward the maximum vibration
years)
point, and slows down as it leaves that point. This
causes the tubes to twist. The amount of twisting is
directly proportional to mass flow. Position sensors
detect tube positions.
Vortex Gas and Vortex flow meters are one of the few types of 10-100% of the
meter Liquid meters, besides differential pressure, that can measurement
accurately measure the flow of liquid, steam, and range: 2%
(Expected
gas. Vortex flow meters operate on the von Karman
life span: 10
principle of fluid behavior, where the presence of
years)
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
with purpose of converting a volume of gas measured at and -10 – 40°C:
Electronic one set of conditions to a volume of gas expressed 0.5%
Volume at another set of conditions. The device incorporates
Conversion integral (internal or external) temperature and/
Instrument or pressure measurement transducers. It may
(EVCI) be directly mounted onto a single meter (with
mechanical drive or magnetic drive coupling) or
(Expected
connected to a remotely located meter from which
life span: 10
it is fed volumetric pulses. The device may perform
years)
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.

• A recommended method to determine the application; proper installation of the equipment;

frequency of maintenance; the ability to inspect, verify, or calibrate the
• Performance verification or calibration of various measurement system components; and
the meter and secondary instruments; and the frequency of those maintenance activities.
The performance of the meter may also depend
• Other relevant and necessary information. on the piping configuration and compensation
Fuel gas can be measured by different for variability of operating pressure, temperature,
types of flow meters, and the selection of a and fluid composition.
meter typically depends on several factors
such as operating conditions and their An example application of the TR 2571 method
variability, desired (or required) accuracy, is provided in Section 5.3.3.
cost of installation, life expectancy, and 3.2.3. Integrated measurement systems
other applicable factors.
As indicated above, the uncertainty of flow
The uncertainty of measurement depends on measurements depends not only on the
the measurement equipment selected for the hardware selected but also on how hardware

30
 ADDRESSING UNCERTAINTY IN OIL & NATURAL GAS INDUSTRY GREENHOUSE GAS INVENTORIES

performs under field conditions. The integrity

and accuracy of the overall measurement
system also depends on the calibration
methods and calibration equipment used and
the calculation protocols embedded in the
operating software, along with a combination
of factors as indicated below:
• Design, installation, and operation of the
flow measuring device;
• Choice of measurement equipment (charts,
transmitters, smart transmitter, analog/
digital converters, data loggers, etc.;
• The means of data transmission (analog,
pneumatic, digital, manual);
• The calculation procedure and means
for computation (chart integration, flow
computer, laptop, etc.;
• The frequency of calibration checks and
meter calibrations;
• The effects on the operation/calibration of
ambient temperature, fluid temperature
and pressure, response time, atmospheric
pressure and the like; and
• Proper unit conversion from ambient
conditions to designated standard
conditions.
The majority of flowmeters used in the
petroleum and natural gas industry typically
are either mechanical or pressure-based
flows, many of which incorporate electronic
devices that can correct for varying pressure
and temperature (i.e., density) conditions, and
non-linearity, and for the characteristics of
the fluids being measured. When electronic
transducer-based systems are used for pressure
measurements, ASTM developed a standard
practice guide for static calibration that is
designed to ensure reliable conversion from
system output to pressure readings for both
laboratory and field use (ASTM D5270-95).
It is important to recognize that site-specific
parameters may influence the measurements.
Since the principle of operation and differing
influence parameters have varying degrees
of influence by meter type, it is important to
identify and define the significant influence
factors for a meter when determining the total
or combined measurement uncertainty.
Exhibit 3-3 below provides a checklist of factors
to consider when evaluating the uncertainty
of flow measurements used for GHG emissions
quantification. Additional analyses are required
to convert this individual measurement
uncertainty to an assessment of the uncertainty
range of annual measurements.

31
IPIECA • API

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 usually are applicable to determine the amount of
gas volume though they may be applied to the measurement of liquids materials, such as
in fuel pumps.
• The actual amount of gas flow has to be corrected for pressure and temperature to the
specified standard conditions to avoid major systematic errors.
d) Determination of total uncertainties
• Individual uncertainties determined in a), b), and c) above

3.3 Sampling and analysis for analytical measurements based on these

quantifying GHG emissions samples. Analyses of gas samples are used
for multiple purposes and are applied to a
The emission measurement process comprises variety of calculations including determination
either direct measurement at the source level, of heating values, gas density and viscosity,
or collection of samples and their subsequent hydrocarbon dew point, and compressibility.
analysis in the laboratory to determine These analyses are essential for obtaining
concentrations and ultimately the conversion information about the gas composition,
to mass emissions using the quantified flow including contaminants in the gas stream.
measurements discussed above. Sampling and Calculations based on these analyses are
analysis are part of the same measurement essential to optimization of process conditions,
process and their combined contribution determination of adherence to contractual
to emissions estimation uncertainty is specifications, or estimation of GHG emissions
obtained by their combined variances, as when such a stream is combusted.
detailed in Section 5. Emission measurement
uncertainties for the processes of sampling The API MPMS provides specific details for
and analysis depend on random errors, collecting and handling natural gas samples for
measurement precision, and systematic errors critical measurements such as custody transfer
or bias of the measurement process. (API, 2006). Exhibit 3-4 provides guidance on
general considerations of inaccuracies that
might be introduced in the measurement
3.3.1 Gaseous samples collection and system when collecting samples that are
handling used for carbon content and/or heating value
determinations for GHG emissions. At the same
Proper collection and handling of natural time, designing a sampling and analysis system
gas samples could have a major impact on always should take into account regulatory
the accuracy and representativeness of the requirements and contractual obligations.

32
 ADDRESSING UNCERTAINTY IN OIL & NATURAL GAS INDUSTRY GREENHOUSE GAS INVENTORIES

EXHIBIT 3-4: KEY FACTORS AFFECTING 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, typically are 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

3.3.2 Quantifying sampling precision same time. The determination of sampling bias,
or the difference between the mean of several
Quantifying sampling precision requires that
measurements and the true value, is more
primary samples be collected according to
difficult. Biases could arise from several causes
a defined protocol, but randomized in some
such as sample fouling, inappropriate handling,
way for each sample (in either space or time).
or unrepresentative sampling. The “true” value
For example, a pre-selected percentage of
for the concentration of unknown species in a
the total number of samples can be collected
sample is never known since it is impossible to
in duplicate and the repeatability of the
construct a true reference laboratory standard
measurement determined from these duplicate
for an unknown mixture of gases.
samples. Additionally, duplicate gaseous
samples can also be analyzed in duplicate Sampling uncertainty is a relatively new
and thus a full record of both sampling and concept whose importance is slowly
analysis system variations can be obtained. beginning to be recognized. Recent research
In this case, the sampling component of the has shown that sampling uncertainty is
variance represents the natural variability of often far greater than analytical uncertainty.
the sampling target as well as any errors in the Therefore, combining sampling and analytical
sample collection and preparation. uncertainties to provide an estimate of
measurement uncertainty is an important
When in situ measurement techniques are
component of quantifying the overall
used (e.g., infrared gas analyzer), both the
uncertainty of GHG estimations that rely on
sampling and analysis are addressed at the
sampling gaseous fuels of varying composition.

33
IPIECA • API

3.4 Carbon content measurement by the same operator with the same
practices apparatus under constant operating
conditions on identical test materials
Different types of gas chromatography (GC)
systems are normally used to analyze the
• Reproducibility is the difference between
two results obtained by different operators
carbon content of gaseous streams. The GC
in different laboratories on identical test
systems might be laboratory based or set up
materials.
as an online device for automated collection
of samples and their analysis. The systems When using a natural gas measurement
typically are set up to analyze the individual method for refinery fuel gas, care should be
components in the sampled gas and provide taken to ensure that the range of compositions
detailed reports of properties including of individual components is within the bounds
composition, calorific value, and density. specified by the method. Table 3-4 lists selected
commonly used laboratory measurement
The results of the determination of individual methods that are frequently used for the
or groups of carbon-containing species are determination of fuel carbon content.
then used to assess the total emissions of CO2 Additional details about these methods and
upon combustion of such a fuel. Several key further discussions of their main features are
considerations include: provided in Appendix D.
• If the method is capable of determining
CO2 content with the rest of the carbon
containing species, no further correction 3.4.2 On-line measurements
of the carbon content data is required to On-line determination of fluid stream
properly account for all CO2 emissions. compositions is quite challenging due to
• If the method is set up to provide possible variations in these compositions. This
information only on hydrocarbon species, is especially notable for self-generated fuel gas
the CO2 content should be obtained by an such as refinery fuel gas or other processing
independent measurement and added to plant gas. Conversely, for commercial products
the fuel carbon content data. such as natural gas, liquid fuels, coal and coke, or
for the analysis of associated gas in exploration
• If the method is capable of a quantitative and production operations, the challenges are
determination of CH4 content, these data more related to the ability to analyze multiple
can be used separately for calculating streams rapidly and ascertain that they all are
evaporative and processing leaks along within a desired property range.
with venting losses.
Instrumentation in this field has been
• All carbon content measurement data developed to provide a measurement of stream
should be used in conjunction with the components in order to achieve optimum
applicable fuel flow measurements when control and assure product quality. The
calculating emissions. configurations of such analyzers are customized
to accommodate typical site parameters and
operating practices. Most such analyzers
3.4.1 Laboratory-based measurements
are designed with ASTM and ISO standards
Several ASTM and ISO methods are available in mind, and their calibration routines are
for determining the composition and carbon designed to provide both reported data and
content in the natural gas range as well as for its associated uncertainty. As mentioned
liquid and solid fuels. The methods usually previously, the two primary applications are for
document measurement precision in terms refinery gas analyzers and natural gas analyzers,
of repeatability and reproducibility of the and these are discussed briefly below.
measurement when comparing multiple
sets of measurements. Although the terms • Refinery Gas Analyzer: Refinery gas
“repeatability” and “reproducibility” are samples are delivered to the sample inlet
applicable to all measurement situations (see of the GC after passing through a sample
Appendix A for terminology and definitions), conditioning system that selectively
they are used here in the context in which they removes any liquid fractions and
are defined in ASTM standards: particulate matter from the sample. This
ensures that only the gas phase sample
• Repeatability is defined as the difference is delivered to the analyzer. An internal
between two successive results obtained vacuum pump draws this conditioned

34
 ADDRESSING UNCERTAINTY IN OIL & NATURAL GAS INDUSTRY GREENHOUSE GAS INVENTORIES

Table 3-4. Summary of selected carbon content measurement methods

Method Title Brief Description Measurement Precisiona

ASTM 1945-03 Analysis of Natural -- Covers the determination of -- Repeatability ranges from
Gas by Gas the chemical composition of 0.01- 0.10 mole%
Chromatography natural gases and similar gaseous -- Reproducibility ranges from
mixtures within specified ranges 0.02-0.12 mole%Version 2,
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 from
mixtures 0.1 - 1.0 mole%
-- Applicable to mixtures containing:
hydrogen, oxygen, nitrogen,
carbon monoxide, carbon dioxide,
methane, ethane, and ethylene
ASTM Refinery Gas -- Used for determining the -- Quantification from 0.1-
UOP539-97 Analysis by Gas composition of refinery gas 99.9 mole% for a single
Chromatography 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:
(Six parts)
Determination method for the quantitative -- 2% for species < 1.0mole%;
of Composition determination of the content -- 0.8% for species 1-50mole%
with Defined of hydrogen, helium, oxygen, -- Relative reproducibility:
Uncertainty by Gas nitrogen, carbon dioxide and C1 -- 4% for species < 1.0mole%
Chromatography to C8 hydrocarbons in natural gas -- 1.6% for species 1-50mole%
samples
-- Uses either two or three packed or
capillary columns combinations
ASTM Standard Test -- Applicable for the quantitative -- NOT applicable for
D2650-04 Method for Chemical analysis of gases containing constituents < 0.1 mole%.
Composition of specific combinations of the -- Developed on a specific
Gases by Mass following components: hydrogen; type of measurement
Spectrometry hydrocarbons (up to 6-carbons system
per molecule); carbon monoxide; -- Users have to modify for
carbon dioxide; mercaptans (1-2 their instrument.
carbons per molecule); hydrogen
sulfide; and air (N2, O2, and Ar)
ASTM Standard Test -- Applicable to samples such as -- NOT recommended for
D5291 - 10 Methods for crude oils, fuel oils, additives, and the analysis of volatile
Instrumental residues for carbon and hydrogen materials such as gasoline,
Determination of and nitrogen analysis gasoline-oxygenate blends,
Carbon, Hydrogen, -- Tested in the concentration range or gasoline type aviation
and Nitrogen in of at least 75 - 87 wt% carbon, at turbine fuels
Petroleum Products least 9 - 16 wt% hydrogen, and 0.1
and Lubricants - 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.

35
IPIECA • API

sample into each injector, which then injects (API, 3rd edition, August 2009). The carbon
the mixture onto each of the columns for content of a mixture is the weighted
analysis. A complete analysis of hydrogen, average of the individual component carbon
saturated and olefinic hydrocarbons (C1-C5, contents and can be calculated from the
and C6+ grouped peaks), plus fixed gases compositional analysis of the mixture (see
(O2, N2, CO, and CO2) typically is performed. API Compendium Table 3-6). Hydrocarbon
Precise “retention times” information and gas chromatographic analysis is generally
component areas translate into accurate reported in mole fraction or mole percent,
component identification and quantification and hydrocarbon liquid chromatographic
of the relative magnitude of individual analysis is generally reported in mass fraction
components present in refinery gas. or mass percent; therefore, calculations
• Natural Gas Analyzer: These types of for both types are included in API’s TR
analyzers are applicable to natural gas 2572. Molecular weights of hydrocarbon
samples from wellhead to pipeline- components can be found in the Table of
quality gas. Samples are introduced Physical Properties for Hydrocarbons and Other
using sample cylinders, Tedlar bags, or Compounds of Interest to the Natural Gas
by direct connection to the pipeline or Industry (GPA, 2145).
wellhead sampling points. Usually, two Carbon content can also be estimated from
chromatographic modules are used other process or fuel supply properties
to quickly separate and measure the such as heating values that are correlated
individual components in natural gas. to the fuel type and carbon content. The
The analyzer separates and measures API Compendium (3rd edition, August 2009)
the permanent gases and hydrocarbons summarizes a number of these correlations in
present via an optimized, dual-channel its Tables 3-8 and 3-9, which provide carbon
portable gas chromatograph. Wellhead content and heating value information
samples may often contain significant for petroleum products and natural gas
amounts of H2S. Many instruments heating value ranges, respectively. Use of
take this into account and there are no these correlation factors works well for
interferences, which mean that H2S can be compositions with limited variability or for
measured from 50 parts per million (ppm) fuel supplies where the heating value is
to 30-mole%. continuously monitored or controlled.
For process or fuel supplies with widely
3.4.3 Carbon content calculation varying compositions or that are subject to
periodic upset conditions, the frequency,
API Technical Report (API, 2013) provides
duration and carbon content of this
guidance and a methodology for determining
variability needs to be characterized to
the carbon content of hydrocarbon-based
accurately calculate the average carbon
petroleum and petrochemical products, and
content. Section 3.6 below provides further
the uncertainty of the average carbon content
details on heat content determination.
as calculated from multiple samples taken
during a sampling period. This method is API TR 2572 provides references and
intended to make use of industry-accepted supplemental information on applicable
mixture property data and test methods with industry practices based on published
no new or modified test methods introduced resources, existing industry standards,
in this document. The method is applicable to industry-accepted physical constants or
carbon-content-based reporting or trading for properties of hydrocarbons for measurement,
all gaseous and liquid hydrocarbons. sampling, sampling frequency, and analysis of
hydrocarbon samples. It provides guidance on
It is a companion to API TR 2571 on Fuel
estimating carbon content sampling period
Gas Measurement (see Section 3.2.2), which
uncertainty based on multiple periodic samples
provides guidance on measuring the volume/
taken at intervals that are independent of the
mass of process fuel gas or feedstock.
process or fuel supply operation.
Equations for the calculation of carbon
To determine the number of samples required
content for a pure component and a product
to meet a certain average carbon content
analysis typically are based on the definitions
uncertainty, one can use Equation 3-1 below:
found in Section 4.3 of the API Compendium

36
 ADDRESSING UNCERTAINTY IN OIL & NATURAL GAS INDUSTRY GREENHOUSE GAS INVENTORIES

2
 σ 
 k 95 × ×100 
CCaverage
Number of Sampleas = 
 TargetPercentCCuncertainty 
 
 
(Equation 3-1)
Where:
Number of samples = the number of samples to be collected
k95 = the 95 % confidence coverage factor (typically assumed to
be 3)
σ = the carbon content standard deviation of the samples
analyzed
CC = carbon content
Target Percent = the target percent uncertainty for the reporting period
CCUncertaint

This equation could be reversed to derive the level 3.5 Heat content determination
of uncertainty given a specific sampling frequency.
The heating value or calorific value of a
According to the discussion in Annex B of API substance is the amount of heat released during
TR 2572, the uncertainty of reporting period the combustion of a specified amount of the
fuel carbon content (i.e., its 95 % confidence substance. The calorific value is a characteristic
interval) is dependent on the number of for each substance and is measured in units
multiple fuel samples used. Increasing the of energy per unit of the substance, such as:
number of samples reduces the uncertainty of kcal/kg, kJ/kg, J/mole, Btu/m³. The heat of
the reporting period average carbon content combustion for fuels is expressed as:
by the square root of the number of samples.
• HHV: higher heating value (or gross calorific
For example, if the average annual carbon value or gross energy or upper heating value)
content is determined from monthly fuel is determined by bringing all the products
sampling and analysis (12 spot samples) and of combustion back to the original pre-
the standard deviation of these 12 samples combustion temperature, and in particular
is 3.9%, the 95% confidence interval for condensing any water vapor produced.
annually reported carbon content will be: • LHV: lower heating value (or net calorific
± (3×3.9% / √¯12), or ± 3.4 %. However, if value) is determined by subtracting the
quarterly samples are used (four samples heat of vaporization of the water vapor
per year with the same sampling standard from the higher heating value and treating
deviation of 3.9%), the 95% confidence any water formed as a vapor. The energy
interval for the annually reported carbon required to vaporize the water therefore is
content would be + 5.9%. not realized as heat.

37
IPIECA • API

In the petroleum and natural gas industry, the The ASTM approach recognizes that the
two most prevalent modes of determining calorific value of a mixture, such as in refinery
heating values of gaseous fuels are either fuel gas systems, is a function of the mol%
by measuring it directly, which can be composition of the individual components of
accomplished either by stoichiometric that mixture. For refinery fuel gas, this mixture
combustion or by calorimetric techniques, or could contain both carbon containing species,
using computational methods that are based on which upon combustion will contribute
standardized calculation procedures using gas directly to CO2 emissions, as well as non-carbon
sample composition data. A brief summary of containing species. Hydrogen, for example, is
these two types of practices is provided below. an important contributor to refinery fuel gas
heating values but it does not contribute to CO2
emissions when combusted.
3.5.1 Direct measurements
To implement this practice, the user would first
The heating value indicates the amount determine the molar composition of the gas in
of energy that can be obtained as heat by accordance with any applicable ASTM or GPA
burning a unit of gas. The heating values of a method. For a precise calculation, at least 98% of
gas depend not only upon the temperature the sample constituents should be determined
and pressure, but also upon the degree of as individual components, with no more than
saturation with water vapor. a total of 2% in terms of groups of components
(e.g. butanes, pentanes, hexanes, and so forth).
As mentioned above, general practices for
determining fuel gas heating values rely on An ideal combustion reaction for fossil fuels
either calorimetric techniques or stoichiometric (that may contain hydrogen) in the ideal gas
combustion practices. Table 3-5 provides a state can be generally represented as in the
listing and a brief description of some selected reaction depicted in Eq. 3-2:
methods for heating value determination for
gaseous, liquid, and solid fossil fuels. Further  b c b 
discussion of the main features of these
Ca H b + cH 2 +  a+ +  O 2 = aCO 2 +  + c  H 2 O
 4 2 2 
methods can be found in Appendix C.
(Equation 3-2)
The ideal net heating value is the heating value
3.5.2 Computational methods
that is observed when all the water remains in
Heating value may be determined from gas the ideal gas state, while the ideal gross heating
compositional analysis in accordance with a value is observed when all the water formed by
standard practice established by the ASTM for the reaction condenses to liquid. The difference
calculating heating values for natural gas and between them is the enthalpy of vaporization of
similar mixtures from compositional analysis. the water formed during the combustion process.
ASTM D3588-98 (Reapproved 2003), is the Therefore, the ideal gross heating value for a
standard recommended practice for calculating mixture can be expressed as shown in Eq. 3-3:
heating values, compressibility factors, and relative
densities of gaseous fuels (ASTM D3588-98). This n
practice covers procedures for calculating these
quantities at base conditions (14.696 psia and
∑x
j =1
j × M j × H mj
60°F or15.6°C) for natural gas mixtures from Hm = n
compositional analysis. It applies to all common
types of utility gaseous fuels, i.e., dry natural gas,
∑x
j −1
j ×M j
reformed gas, oil gas (both high and low Btu), (Equation 3-3)
propane-air, coke oven gas, and other gaseous Where:
fractions for which suitable methods of analysis
are designated.
xj= the mole fraction of Component j;
Mj = the molar mass of Component j;
The ideal gas heating value and ideal gas
relative density are calculated from the molar n = the total number of components; and
composition and the respective ideal gas Hmj = the ideal gross heating value per unit mass
values for the components; these values for Component j (at 60°F or 15.6°C), as
are then adjusted by means of a calculated tabulated in ASTM D3588 (ASTM, 2003).
compressibility factor. Values of Hm are independent of pressure,
but they vary with temperature.

38
 ADDRESSING UNCERTAINTY IN OIL & NATURAL GAS INDUSTRY GREENHOUSE GAS INVENTORIES

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

Table 3-5. Summary of selected heating value measurement methods

Method Title Brief Description Measurement Precision (*)
ASTM Test Method for -- Used for the determination of -- Repeatability: 0.76 Btu/
D4891-89 Heating Value of heating value of natural gases scf, 95% confidence: 2.1
(Reapproved Gases in Natural and similar gaseous mixtures Btu/scf.
2006) Gas Range by within a specified composition -- Reproducibility: 1.67
Stoichiometric range Btu/scf, 95% confidence:
Combustion -- Provides an accurate and 5.1 Btu/scf.
reliable procedure for regulatory -- Average bias: within
compliance, custody transfer, 0.1% from reference
and process control value
ASTM Standard Test -- Used for the determination of -- Weekly standardization
D1826-94 Method for the total calorific (heating) value with methane
(Reapproved Calorific (Heating) of fuel gas produced or sold in -- Errors < 0.5%
2003) Value of Gases the natural gas range from 900- within a week after
in Natural 1200 Btu/scf standardization
Gas Range by -- Provides a reliable method for -- Higher errors
Continuous measurement on a continuous expected for longer
Recording basis with a recording standardization periods
Calorimeter calorimeter
ASTM Standard Practice -- Used to determine the heating -- No generic precision
D7314-08 for Determination value of gaseous fuels with at- data apply here since
of the Heating line and in-line instruments this is a practice and not
Value of Gaseous -- Suitable for periodic operation a method
Fuels using on a continuous basis -- The installation
Calorimetry and -- Suitable for monitoring systems and operation of
On-line/At-line for tracking properties when particular systems
Sampling using or producing gaseous vary with process type,
fuels in industrial processes performance, and
regulatory requirements
ASTM Standard Test -- Covers determination of the heat -- Strict adherence to all
D4809-13 Method for Heat of combustion of hydrocarbon details of the procedure
of Combustion fuels is essential
of Liquid -- Can be used for a wide range -- The error contributed
Hydrocarbon of volatile and nonvolatile by each individual
Fuels by Bomb materials where slightly greater measurement that
Calorimeter differences in precision can be affects the precision shall
(Precision tolerated be < 0.04 %, insofar as
Method) -- Under normal conditions, the 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 Standard Test -- Pertains to the determination of −
D5865-13 Method for Gross the gross calorific value of coal
Calorific Value of and coke by either an isoperibol
Coal and Coke 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.

39
IPIECA • API

3.6 Venting and fugitive emissions of instruments, drip pots, and scrubbers.
measurements Continuous applications may include
disposal of associated gas, treater off-gas
Vented emissions are releases to the and tank vapors at oil production facilities,
atmosphere that are due to process or casing gas at heavy oil wells, process waste,
equipment design, or operational practices. or byproduct streams. Depending on local
Vented emissions may come from a variety requirements, some of the vented streams
of emission points or non-fired stacks, or may be either captured for on-site use/sale
sometimes from flares as is commonly the or flared, depending on the economic value
approach used for categorizing emissions in of the stream and the availability of local
GHG emission inventory compilations that infrastructure.
adhere to the IPCC 2006 national inventory
guidance. In all cases, these are part of the These emission sources tend to be very specific
process design and are intended for pressure to the type of operation and detailed methods
relief purposes. Fugitive emissions are due to for quantifying these emissions are provided
equipment leaks from piping components and in Section 5.0 and 6.0 of the API Compendium
a variety of seals found in typical petroleum (3rd edition, August 2009).
and natural gas installations. Table 3-6 provides a summary of the vented
Vents may either be intermittent or continuous. and fugitive emission source types and
Intermittent applications may include the their measurement requirements under the
disposal of waste volumes from emergency EPA’s GHG Reporting Program (GHGRP). The
pressure relief episodes, operator initiated or emissions sources are categorized around the
instrumented depressurization events, plant subparts of the EPA regulation: Petroleum
or system upsets, well servicing and testing, and Natural Gas Systems (Subpart W) and
pigging events, and routine blowdown Petroleum Refining (Subpart Y).

Table 3-6. GHGRP measurement methods for vented and fugitive emission sources
Emission
Source Type GHGRP Measurement Method
Category
Petroleum and Natural Gas Systems
Vented -- Pneumatic device and pump vents, Engineering Estimate using
Emissions -- Gas well liquids unloading monitored process operating
-- Gas well completions parameters and either simulation
-- Gas well workovers models or emission factors
-- Storage tanks
-- Dehydrators
-- Acid gas removal
-- Compressor wet seal vents Direct Measurement where
-- Compressor rod packing vents meters depending on source
-- Transmission condensate tanks measured
Fugitive -- Compressor station leaks Leak Detection Surveys in
Emissions -- Processing plant equipment leaks conjunction with emission
-- Fugitives and non-compressor plant/station fugitives factors for identified “leaking”
components
-- Valves, connectors and pump seals from: Equipment Count in conjunction
• Offshore petroleum and natural gas production with applicable population
• Onshore petroleum and natural gas production emission factors
• Local distribution company mains,
• Service lines, vaults, and other inaccessible sources
Vented -- Fluid Catalytic Cracking Engineering Estimate using
Emissions -- Catalyst regeneration monitored process operating
-- Hydrogen production parameters and either simulation
-- Acid gas removal models or emission factors
-- Sulphur recovery
-- Cokers

40
 ADDRESSING UNCERTAINTY IN OIL & NATURAL GAS INDUSTRY GREENHOUSE GAS INVENTORIES

3.6.1 Vented emission measurements • Calibrated Bagging: a calibrated bag is

used to enclose an emissions source to
The term “GHG emission measurement” may
completely capture all the leaking gas.
refer to direct measurement of emissions, the
The volume of the bag and time required
use of engineering equations or models that
to fill it is used to determine the mass
rely on some measured operating parameters,
emission rates. This is a suitable technique
or known activities with default emission
for emission sources that are within a
factors. The accuracy of these estimation
safe temperature range and can be safely
methods varies and should weigh the
accessed. The speed of measurement is
considerations outlined in Section 2.5.
highly dependent on the emissions rate
The sections below focus on three types of and the results are susceptible to human
technologies that can be used as appropriate to errors that are introduced when enclosing
measure or quantify the magnitude of vented the emission source and taking the
emissions. measurement data.
• High Volume Sampler: a high volume Calibrated bags can be used only where
sampler consists of a simple fixed-rate the emissions are at near-atmospheric
induced flow sampling system to capture pressures and the entire emissions volume
and measure the volume of emissions. can be captured for measurement. Using
The emissions and the air surrounding these bags on high pressure vent stacks
the emissions source are drawn into the can be dangerous. The mass emission rate
instrument using a sampling hose. The is usually estimated as the average of three
instrument measures the flow rate of the measurements of the time required to
captured volume of air and emissions fill the bag followed by chemical analysis
mixture, and the concentration of of the bag content to determine mass
hydrocarbons in the mixture. A separate emission rates of individual species. Section
sample of the ambient air is taken by the 3.5 presents further information that is
instrument to correct for the background specific to carbon content measurements.
concentration of hydrocarbons in the
ambient air. High volume samplers are
limited in the range of flows and emission 3.6.2 Fugitive emission measurements
rates that they can capture; therefore, Similar to vented emissions (Section 3.6.1),
large component emissions and many fugitive GHG emissions from equipment
vent emissions are above the high-volume leaks may be quantified either by direct
sampler capacity and therefore require the measurement of leaking components or by
use of other measurement instruments. a count of the myriad of components and
• Meters: several types of meters measure associated default emission factors. The
liquids and gas flow rates, and can be accuracy of the emission estimate increases
used for measurement of emissions from with more precise knowledge of the number
sources where the volume of emissions of components of each type and service, and
is high, as in vent stacks. Typical meters with characterization of their leaking status.
include: rotameters, turbine meters, Although the use of default emission factors
hot-wire anemometers, pitot tube flow is expedient in some instances - especially for
meters, and vane anemometers. Further remote locations - it may lead to overestimating
details on flow measurement practices are the contribution of emissions from leaking
provided in Section 3.1 and the uncertainty process components due to the conservative
associated with flow measurements for GHG nature of the emission factors commonly used.
inventories is addressed in Section 3.2.
When site-specific measurements are
To ensure accurate measurements when undertaken, leak detection has been
using metering, all emissions from a single traditionally performed by Toxic Vapor
source are channeled directly through Analyzer (TVA) and the Organic Vapor Analyzer
the meter. An appropriately sized meter (OVA). New gas imaging technologies are now
should be used to prevent the flow from available using remote infrared detection of
exceeding the full range of the meter, the leaking gases. One of these methods could
while having sufficient momentum for be used in the field depending on factors such
the meter to register continuously in the as size and complexity of facilities and local
course of measurement. requirements.

41
IPIECA • API

• Organic Vapor Analyzers: OVA and in the absence of precipitation and

instruments are equipped with flame high wind. EPA’s Alternative Work Practice
ionization detectors that are used to (AWP) provides further instructions for
detect the presence of hydrocarbons and calibration and operation including the
measure their concentrations at the leak ability to detect a minimum flow rate (see
interface. It consists of a probe that is EPA, 2008: 40 CFR, Part 65, Section 7) before
moved close to and around the potential each use.
emissions locations where emissions are
detected when the instrument records
positive readings on its monitoring scale. 3.7 Laboratory management system
However, these concentrations are not
a true measure of mass emission rate, An additional consideration for uncertainty
but represent “screening” values. The of GHG measurements is the credibility
component screened is said to be leaking and technical veracity of the laboratory
if the measured concentration exceeds a performing the requested tests, as specified by
defined concentration threshold that is different programs. For example, in the EU-
defining a “leak.” ETS program the Monitoring and Reporting
Guidelines (MRG) require the demonstration
This method is suitable for use for all of laboratories management systems (EU-ETS
equipment leak detection that is safely MRGs, 2007; Section 13.5). Any laboratory used
accessible at close range. For each to determine an emission factor, calorific value,
potential emissions source, all joints, carbon content, or composition data should
connections, and other potential paths to be accredited according to ISO 17025:2005. If
the atmosphere should be monitored for non-accredited laboratories are used, the EU
emissions. Further guidance on operation regulations provide specific requirements for
and calibration of these instruments the validation and testing of such laboratories.
is provided by EPA (EPA, method 21
guidelines). ISO/IEC 17025:2005 (ISO/IEC 17025, 2005) is
not intended to be used for overall laboratory
• Remote Infrared Gas Imaging Detectors: certification, and it does not address compliance
these instruments are based on the with regulatory and safety requirements.
ability of hydrocarbon compounds It merely emphasizes the need for a well-
to absorb infrared radiation. The developed and communicated laboratory
instruments use this property to scan a management system that addresses areas such
given facility area and provide distinct as: quality, administrative procedures, and
imaging of the gas that is leaking from technical systems that govern the operations
piping components. of the laboratory. The ISO standard highlights
To be able to detect the visually the need for laboratories to ensure that their
displayed image of the leaking personnel are aware of the relevance and
components, it is essential to have an importance of their activities to the achievement
appropriate background. Therefore to of the management system’s objectives.
identify all emission sources, the operator
inspects the emissions source from The standard specifies the general requirements
multiple angles or locations until the for demonstrating competence to carry out
entire source has been viewed without specific tests and calibrations, including field
visual obstructions. sampling. It covers testing and calibration
procedures that are performed using standard
The minimum detectable quantity of methods, non-standard methods, and
equipment leaks using an IR instrument laboratory-developed methods. This standard
depends on a number of factors may be applied to all organizations performing
including manufacturer, viewing tests and/or calibrations, including both internal
distance, wind speed, gas composition, companies’ laboratories as well as external
ambient temperature, gas temperature, contract laboratories.
and type of background behind the
leaking components. For best survey ISO/IEC 17025:2005 is applicable to all laboratories
results, equipment leak detection regardless of the number of personnel or the
should be performed under favorable extent of the scope of testing and/or calibration
conditions, such as during daylight hours, activities. When a laboratory does not undertake

42
 ADDRESSING UNCERTAINTY IN OIL & NATURAL GAS INDUSTRY GREENHOUSE GAS INVENTORIES

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.

43
IPIECA • API

4. OVERVIEW OF REPORTING PROGRAMS’ UNCERTAINTY

REQUIREMENTS AND TOOLS
The development of GHG emissions inventories
and their reporting have evolved over the past
two decades starting with IPCC rules designed
originally for national reporting under the SECTION FOCUS
UNFCCC (IPCC, 1996 and 2006 guidelines).
Emerging from those rules were guidelines
and international standards more applicable to This section provides an overview of the
corporate reporting such as the GHG Protocol uncertainty requirements from various
Corporate Standard (WRI/WBCSD, 2004) and GHG reporting programs applicable to
the International Standards Organization (ISO) the petroleum and natural gas sector,
and its family of 14064 standards (ISO, 2006). and compares elements of programs
Since the development of the World Resources and their uncertainty targets. Where
Institute (WRI) GHG Protocol Corporate available, calculation tools available
Standard, WRI and the World Business Council to assist reporters in quantifying the
for Sustainable Development (WBCSD) have uncertainty of GHG emission estimates
collaborated in an initiative to assist developing for these programs are also introduced.
countries and rapidly industrializing countries
The subsections address:
(such as Brazil, China, India, Mexico and the
Philippines) to use the GHG Protocol as a • Global uncertainty approaches
foundation for their GHG programs2. applicable to the petroleum and
natural gas sector;
The sections below provide an overview
of select GHG reporting programs that are • Uncertainty targets for mandatory
pertinent to global petroleum and natural programs; and
gas operations. An emphasis is placed on • Tools for calculation uncertainty.
summarizing various program approaches for
addressing uncertainty and the availability of
calculation tools.

4.1 Global approaches applicable to For example, the EU ETS now operates in 30
petroleum and natural gas sector countries and covers CO2 emissions from some
11,000 installations. In Japan, in 2009, over
Global trends indicate that a growing number 11,000 enterprises reported their CO2 emissions
of companies assess and address the potential under the mandatory GHG Accounting and
threats and opportunities of climate change Reporting system, accounting for about half of
for their business. At the same time, there is the total emissions of Japan nationwide. In the
also an increasing demand from governments U.S. around 8,000 entities reported data in 2012
and other stakeholders for corporate climate under the GHG Reporting Program, covering an
change-related information, therefore, estimated 85-90% of total U.S. GHG emissions.
encouraging or mandating companies to
disclose GHG emissions data. This has given Table 4-1 provides an inter-country comparison
rise to a myriad of sector-specific guidance for of mandatory reporting programs in selected
voluntary reporting of GHG emissions, such countries with significant petroleum and
as for the petroleum and natural gas industry natural gas industry operations. As indicated in
(IPIECA, 2011), along with increased requests the table, these programs have varying policies
from investors for data disclosure (CDP, 2010). for implementing their quality assurance,
At the same time, a wide range of regulatory verification, and uncertainty assessment
schemes are being developed that mandate requirements but they all share the common
GHG emissions reporting by installations and goal of ensuring the reporting of high quality
business entities at the national and regional data to be used as the basis of policy decisions.
levels (OECD, 2012).

2 http://www.ghgprotocol.org/programs-and-registries

44
 ADDRESSING UNCERTAINTY IN OIL & NATURAL GAS INDUSTRY GREENHOUSE GAS INVENTORIES

Table 4-1. Inter-Country comparison of select mandatory GHG reporting programs

Program
Countries
Elements
Australia Canada EU France UK USA
Program National Environment European Union Bilan démissions Greenhouse Greenhouse
Name Greenhouse Canada GHG Emissions Trading de GES Gas Emissions Gas Reporting
and Energy Emissions System Reporting Program
Reporting Reporting (GHGRP)
(EU-ETS)
(NGER) Program
(GHGRP)
Inception 2009 2004 2005 2011 2009 (Mandatory 2009
Date April 2013)
Legal National Canadian EU Regulation “Grenelle II” Climate Change FY2008
Authority Greenhouse Environmental 601/2012 Act of 2008 Consolidated
n°2010-788,
and Energy Protection Act and pending Appropriations
(21-JUN-2012),
Reporting (CEPA, 1999) – 12 July 2010 regulations Act (H.R. 2764;
Monitoring and
(NGER) Act of Section 46 (Article 75) PL 110– 161),
Reporting of
2007
GHG emissions
pursuant to
Directives
2009/29/EC and
2003/87/EC
Geographical Australia Canada EU- 28?, Norway, France Global emissions U.S. entities
Coverage Lichtenstein and of UK companies
Switzerland
Gases 6 Kyoto GHGs 6 Kyoto GHGs Primarily CO2, 6 Kyoto GHGs 6 Kyoto GHGs 6 Kyoto GHGs +
Reported with provision to NF3
include CH4 and
N2O in the future
Reporting Corporate Facility level Installation - Companies over All London Stock Facilities
Boundaries Control 500 employees; Exchange- listed
Fossil fuels or
- Sub-national firms
industrial GHGs
government
Consistent with suppliers
over 50,000
GHG Protocol
inhabitants; Vehicles
- Public bodies and engines
over 250 manufacturers
employees
Reporting 25 KtCOe 50 KtCO2e 25 KtCO2e per No No 25 KtCOe per
threshold (facility) (facility) year (excluding year
emissions from
50 KtCOe
biomass);
(company)
Combustion
Production /
of fuels in
consumption
installations with
> 200TJ
a total rated
thermal input
exceeding 20 MW

45
 IPIECA • API

Table 4-1. Inter-Country comparison of select mandatory GHG reporting programs (continued)

Program
Countries
Elements
Australia Canada EU France UK USA
consumption > 200TJ Combustion
of fuels in
installations with a
total rated thermal
input exceeding
20 MW
Scope of - Direct operational - Direct operational - Direct regular - Direct operational Direct Direct
emissions emissions emissions operational emissions operational operational
- Indirect emissions emissions - Indirect emissions emissions emissions
from energy - Abnormal events from energy
Indirect Potential
consumption including start-up consumption
emissions from emissions from
- Other indirect and shut-down
energy distribution of
emissions (optional) - Emergency
fossil fuels and
situations
industrial gases
- Exception of
emissions from
mobile machinery
for transportation
purposes.
Calculation - 2009 Technical - Gov’t of Canada - Commission - Bilan methods DEFRA Guidance - Detailed
methods Guidelines Technical Guide Regulation (EU) démissions de GES on how to calculation
- Australian EFs Based on IPCC No. 601/2012: - Inspired by ISO measure and methods for
provided Guidelines and Good - Subsection 2 – 14064-1, 2006 report GHG each of the
Practice Guidance Activity Data - Consistent with emissions 41 subparts
- Sector-specific - Subsection 3 – GHG protocol addressing
(September 2009)
manuals Calculation Factors - EFs from ADEME sectors of the
- National EFs and - Subsection 4 – database economy that
GWPS Specific calculation are required to
- Conversion factors factors report
and Electricity Industry
intensity consensus
standards for
measurements
Reporting - Annual report to - Annual report to - Annual verified - Every three years Annual report Annual
Ministry of Climate Environment Canada report with financial reporting
Change and Energy - Electronic Data - Information report through the
Efficiency Reporting (EDR) reported in EPA’s electronic
- Online System system on the GHG rounded-off tonnes GHG reporting
Comprehensive reporting website of CO2e tool (eGGRT)
Activity Reporting
(OSCAR)
Quality - Internal - Information should - Member States - Internal Not specified - QA/QC checks
Assurance be “verifiable” are required specified in
to ensure that each of the rule
emissions are subparts
monitored in - Guidance for
accordance with substitution of
the monitoring missing data
and reporting
guidelines.

46
 ADDRESSING UNCERTAINTY IN OIL & NATURAL GAS INDUSTRY GREENHOUSE GAS INVENTORIES

Table 4-1. Inter-Country comparison of select mandatory GHG reporting programs (continued)

Program
Countries
Elements
Australia Canada EU France UK USA
Verification - 3rd party - No specific - Member States are required - Not required - No specific - Self-
verification not requirements to ensure that the reports by law requirement certification
required for a 3rd party submitted by operators are for 3rd party by facilities’
verification. verified verification of registered
GHG data designated
representatives
- Verification
checks by EPA
Uncertainty - Required - No specified - Maximum Permissible - Not specified - Not - Specifications
assessment - Uncertainty ranges assessment Uncertainty for each tier of beyond GHG addressed for calibrations
provided with EFs required activity data (Anx II, Table1) protocol in General
guidance Provisions
(Subpart A)

Citations http://www.cleane- www.ec.gc.ca/ http://eur-lex.europa.eu/Lex- www.develop- www.defra. www.epa.gov/

nergyregulator.gov. ges-ghg UriServ/LexUriServ.do?uri=O- pement-du- gov.uk/en- climatechange/
au/National-Green- J:L:2012:181:0030:0104:EN:PDF rable.gouv. vironment/ emissions/ghg-
houseand-Ener- fr/IMG/pd- economy/ data/index.html
gy-Reporting/Pages/ f/09003PLAN- business-effi-
default.aspx CLIMAT.pdf ) ciency/report-
ing

Since not all governments have reporting by major investors. Table 4-2 provides a high
programs, a few non-governmental initiatives, level comparison of the scope and coverage
require or encourage enterprises to quantify of mandatory reporting in the USA. under the
and report their GHG emissions. These voluntary GHGRP as compared to the voluntary disclosure
initiatives are part of environmental and other of GHG emissions under the CDP, which
non-financial disclosure requirements, with brings into the voluntary CDP reporting the
the most notable one being that of the Carbon measurements and QA/QC checks associated
Disclosure Project, which is backed globally with mandatory GHG reporting in the USA.

Table 4-2. Comparison of reporting scope between EPA’s GHGRP and CDP
EPA Greenhouse Gas
CDP Comment
Reporting Program
Reporting - Monitor and report GHG emissions - Monitor and report GHG emissions - EPA’s GHGRP supplements and
requirements - Include climate change related complements other US government
risks, opportunities, strategies & programs that may include CDP subject
performance) matter
Reporting - Facilities over 25,000 tonnes CO2e - At the corporate level (including - Information about GHG emissions from
responsibilities the facilities they own, manage or facilities is a subset of the information
- Suppliers report at the corporate
operate) requested by CDP
level.
Scope of - 41 industry sectors - Same as EPA GHGRP, including - CDP includes all of the activities covered
reporting combustion and fugitive emissions by the EPA GHG RP and a few more
- Examples: stationary fuel
combustion, electricity generation, - Mobile source emissions
industrial and chemical processing,
- Petroleum and Natural Gas
Suppliers of fuels, products and
industrial gases.
Reporting - Annually (calendar year) - Annually (latest or more recent - For CDP reporting could coincide with
frequency 12 month period for which data is financial reporting
reported)
47
IPIECA • API

4.2 Uncertainty targets for excerpt from the updated EU regulation

mandatory GHG programs specifying the rules for monitoring and
reporting (EC, 2012). The regulation (Annex
With the emergence of new mandatory II, table 1) sets up tiers for maximum
reporting regulations and emission reduction permissible uncertainty for activity data
compliance obligations, new requirements tiers. Facility owners or operators may
are being promulgated for the accuracy of apply a fallback monitoring methodology
fuel flow measurements when such flows if they can demonstrate to the satisfaction
are used to quantify GHG emissions. A few of the regulatory authority that they can
prominent examples are provided below: attain overall uncertainty thresholds for
a. California Air Resources Board (CARB) the annual level of GHG for the whole
Mandatory GHG Reporting Regulations. installation that do not exceed 7,5 % ,
Flow measurement uncertainties are 5.0% and 2.5% for category A, B, and C
expected to be ±5% (CARB, 2012). The installations,3 respectively.
California requirements specifically apply The EU-ETS requirements are applicable
to all GHGs emitted from petroleum to a limited set of petroleum and natural
refineries, hydrogen plants, and gas industry installations and track only
cogeneration plants. CO2 emissions from fuel combustion and
b. EU-ETS. The European Union Emissions flaring. The requirement to quantify these
Trading System (EU-ETS) specifies a sources within such tight uncertainty
tiered approach for emission calculations ranges is a reflection of the fact that these
together with required uncertainty are the sources for which appropriate
ranges (EU-ETS MRG, 2007). It sets up a emission calculation methods are available,
matrix of uncertainty requirements for while they are also the largest emission
different facility sizes and measurement sources and the key contributors to most
approaches used. Table 4-3 provides an installations3 GHG emissions.

Table 4-3. EU regulation maximum permissible uncertainty for activity data tiersa
Parameter to which uncertainty
Tier 1 Tier 2 Tier 3 Tier 4
is applied
Combustion of fuels and fuels used as process input
Commercial standard Amount of fuel [metric tons] or
7.5% 5.0% 2.5% 1.5%
fuels [Nm3]
Other gaseous and liquid Amount of fuel [metric tons] or
7.5% 5.0% 2.5% 1.5%
fuels [Nm3]
Solid fuels Amount of fuel [metric tons] 7.5% 5.0% 2.5% 1.5%
Flaring Amount of flare gas [Nm3] 17.5% 12.5% 7.5%
Petroleum Refining
Hydrocarbon feed [metric tons]
(mineral oil refining)
Uncertainty requirements apply
Catalytic cracker, catalyst separately for each emission source
regeneration and flexi- 10.0% 7.5% 5.0% 2.5%
cokers (Total uncertainty of all emissions
from a given source)
Hydrogen production 7.5% 2.5%
a
Excerpt from Table 1 of Annex II of the June 21, 2012 EU-ETS regulation.

3 For the EU-ETS: Category A installation, where average verified annual CO emissions are equal to or less than 50,000 tonnes
2
of CO2e; Category B installation, where the average verified annual CO2 emissions are more than 50,000 tonnes of CO2e and
equal to or less than 500,000 tonnes of CO2e; Category C installation, where the average verified annual CO2 emissions are
more than 500,000 tonnes of CO2e.

48
 ADDRESSING UNCERTAINTY IN OIL & NATURAL GAS INDUSTRY GREENHOUSE GAS INVENTORIES

c. EPA GHGRP. The regulations governing date at which data collection is required
the requirements for the GHG Reporting to begin using a measurement device. All
Program (GHGRP) in the USA. are included measurement devices must be calibrated
in many subparts to Part 98 in Title according to a method specified either in a
40 of the Code of Federal Regulations relevant subpart of the regulations in Part
(40CFR98). The rules are specified under 98 or the manufacturer’s recommended
general provisions along with sector- procedures, or an appropriate industry
specific subparts that invoke the general consensus standard.
provisions and add more requirements that
Highlights of the calibration and
are pertinent to specific industry sectors.
maintenance requirements are provided
General Provision §98.3 specifies that
in Table 4-4 below for all equipment and
reporters must follow the procedures for
measurements used in association with
emission calculation, monitoring, quality
applicable subparts to provide data for the
assurance, missing data, recordkeeping,
GHG emission calculations. Subsequent
and reporting that are specified in each
recalibrations of the flow meters and
relevant subpart.
other measurement devices must also be
In accordance with these provisions, flow performed, as shown in Table 4-4. If no
meters and other devices that measure recalibration frequency is specified in the
data used to calculate GHG emissions applicable subpart, the recalibrations must be
(e.g., GC analyzers used to determine performed at the frequency recommended
composition data and flow meters used by the manufacturer or as specified by an
for calculations) must be calibrated by the industry consensus standard practice.

Table 4-4. Highlights of calibration requirements under the EPA GHGRP

Calibration Error
Activity Applicability Requirements Methods
(CE)
Initial - Meters that must - Flow meters used for - Calibration methods as - CE < 5% at each
Calibration be calibrated prior to reporting must meet in a relevant regulation measurement point
commencing reporting: applicable accuracy subpart For differential pressure
- Flow meters that measure specification. - Alternatively: (delta-P), total pressure,
liquid and gaseous fuel feed - Exception: flow meters - Manufacturer’s and temperature
rates used to provide engineering recommended transmitters, CE < 2%
- Meters that measure estimates. procedures at each measurement
process stream flow rates - Calibrations should include point
- Appropriate industry
- Meters that measure measurement points within consensus standard - Alternatively, the sum
feedstock rates the normal meter operating of the calculated CE
- The calibration
range values for the three
- All other measurement method(s) used shall
- For orifice, nozzle, and transmitters < 6% at
devices (e.g., weighing be documented in the
venturi flow meters: In-situ each calibration level
devices) written Monitoring Plan
calibration of the differential
pressure (delta-P), total
pressure, and temperature
transmitters
Maintenance - Meters used to measure - In general, a primary - For continuously - Data from the
and data used for calculation of element inspection (PEI) shall operated units and flow meter shall be
Recalibration GHG emissions be performed at least once processes with considered invalid if:
- Requirements do not every three years infrequent outages: - Initial calibration or a
apply to any units where - For petroleum refineries, - The initial recalibration fail to meet
the rule allows for the use flow meter shall be calibration may be the required accuracy
of “company records” to recalibrated: postponed until specification
quantify fuel usage or other - At the minimum the next scheduled - Invalidation begins
parameters frequency specified by the maintenance outage with the first hour of the
manufacturer
- Fuel billing meters - Subsequent failed calibration
- Every two years
are exempted from the recalibrations may - Invalidation continues
- At the interval specified
maintenance requirements also be postponed until a successful
by the industry consensus
if no common ownership - Postponements should calibration is completed
standard practice used
with the fuel supplier be documented in the
facility Monitoring Plan

49
IPIECA • API

4.3 Tools for calculating uncertainty variables required for the Monte Carlo
simulation software.
A few reporting programs have developed
tools to assist reporters in quantifying GHG Monte Carlo simulation has an advantage
emission uncertainty. Alternatively, general over uncertainty propagation in that one
statistical methods or commercially available can specify multivariate distributions
tools can also be used to evaluate uncertainty. to account for correlations between
A brief description of some common methods different sources of uncertainty. Ideally,
are provided below. the distributions should be derived from
data and knowledge of the underlying
process. It is helpful in many instances
4.3.1 Numerical methods for calculating to first graph the data, and then use the
uncertainty shape of the graph to determine the
underlying distributions. The IPCC Good
a. Monte Carlo Simulation
Practices document recommends choosing
Monte Carlo simulation is a complex, one of the following distributions: normal,
model-based method for iteratively lognormal, uniform, or triangular.4
evaluating uncertainty associated with
It is then necessary to statistically test the
individual parameters. It applies to any
hypothesis that the data follow a certain
probability distribution and may be the
distribution. The test will vary based on the
preferred approach where more complex
hypothesized distribution. For example, the
equations are assessed, or where more
Shapiro-Wilks test is often used to test if
than two correlated variables exist. It is one
the data are normal or lognormal. Options
of many methods for analyzing uncertainty
to test for other distributions include
propagation where the goal is to determine
Empirical Distribution Functions.
how random variation, lack of knowledge,
or error affects the sensitivity, performance, Selecting the proper probability
or reliability of the resulting emission distribution function for the model
inventory. parameters may be difficult in light of
inadequate data or a lack of understanding
The principle of the Monte Carlo
of the underlying physical processes. If
analysis is to generate random inputs
there are not enough data to assume
from probability distributions of the
normality by the Central Limit Theorem
respective variables and to calculate
(more than 30 data points), there are
the corresponding emission values. This
most likely not enough data to determine
procedure is repeated many times using
the underlying distribution of the data.
computer software, such as @RISK or
Consequently, such analyses are often
Crystal Ball. The uncertainty model relies
forced to rely on subject matter expert
on repeated random sampling of all
opinion to determine distributions rather
inputs and simultaneous recalculation
than on observed data.
of emissions (outputs) to measure
variation over the course of numerous Once distributions are determined for
model iterations. The data generated all of the data sources, the Monte Carlo
from the Monte Carlo simulation can be simulation will proceed by randomly
represented as probability distributions sampling each of the distributions that
(or histograms) or converted to error bars describe the data used for estimating
and confidence intervals. emissions. As many as 10,000 replicate
samples are typically taken, with the
Monte Carlo simulation can also be
total emissions being estimated for each
computationally intensive, with 10,000
replicate. These repeated determinations
simulations being the norm. A significant
of emissions are used to generate a
difficulty/barrier is the need to build a
distribution of the total emissions
spreadsheet-based simulation model
with its mean being the estimate of
for uncertainty analysis that reflects the
total emissions, and its uncertainty
complexity of the emissions estimation
determined by its variance.
model and allows generation of the

4 The IPCC Good Practices document discusses how to perform Monte Carlo simulation in Section 6.2 (IPCC, 2000). It discusses
choice of distribution in Section A1.2.5.

50
 ADDRESSING UNCERTAINTY IN OIL & NATURAL GAS INDUSTRY GREENHOUSE GAS INVENTORIES

b. “Taylor” series expansion associated with their entity’s direct emissions

with their annual GHG emission reports.
The Taylor series expansion is an analytical
The figures reported are used to assist the
method used to combine the uncertainty
government in assessing uncertainty in
associated with individual parameters
the national greenhouse inventory that is
from a single scenario. It is a representation
prepared as part of Australia’s commitments
of a function as an infinite sum of terms
under the United Nations Framework
that are calculated from the values of the
Convention on Climate Change.
function’s derivatives at a single point. As a
result, it provides a means of predicting the Technical guidance sets out how
value of a function at one point in terms uncertainty is to be calculated, requiring
of the function value and its derivatives at reporters to identify uncertainty for each
another point. source of emission and each facility in
order to arrive at uncertainty for the entire
This analysis involves the mathematical
entity (Australian Government, 2009).
evaluation of the model equation. The
The NGER guidance provides default
statistical moments (mean, variance,
parameter and aggregated uncertainties,
skewness, etc.) for the model are calculated
per the examples in Table 4-5.
by expanding the model equation in a
Taylor series about the means. The result
of the expansion process is an equation
for the overall model statistical moments, Table 4-5. Estimated uncertainty levels based on Australia
which is a function of the variable National Greenhouse Gas and Energy Reporting (NGER)
moments and the partial derivatives of the
model equation. Online calculators exist for Fuels Combusted or Operations Uncertainty
applying the Taylor series expansion. Performed Level (%)a

Drawbacks associated with this type of FUELS

COMBUSTION
analysis include:
CO2 Diesel oil; kerosene; fuel oil; heating oil 2
• The numeric expansion may only be
Crude oil (including crude oil 3
good for functions that are linear or
condensates); Liquid Petroleum Gas;
nearly linear (unless many terms are jet fuel
used in the expansion).
Natural gas if distributed in a pipeline; 4
• The method is strictly based upon Coal Bed Methane or Coal Mine Waste
calculations involving a parameter’s Gas (if captured for combustion);
statistical moments and does not compressed natural gas; unprocessed
directly incorporate the parameter’s natural gas; Town Gas; Liquefied Natural
probability distribution (e.g., normal, Gas; gasoline
lognormal, gamma). Naphtha; black coal (other than used to 5
• The expansion results only return an produce coke)
estimate of the statistical moments Coking coal 7
and not a distribution. Other natural gas liquids 9
• Higher order estimates may be nec- Ethane 10
essary to adequately address highly
Petroleum coke; refinery coke; 17
skewed distributions. ethanol (for use as a fuel in an internal
• The model may not be defined explic- combustion engine); biodiesels
itly as a function of the input variables. Refinery gas and liquids; sludge biogas 18
As such, a Taylor expansion about the (if captured for combustion)
mean is not possible for these cases. Coke oven gas 19
CH4 and N2O All Fuels 50
4.3.2 Examples of program-based tools FUGITIVE Oil and gas exploration, production, 50
EMISSIONS processing, transmission
a. NGER Uncertainty Calculator
(including
The Australian National Greenhouse Gas Flaring)
and Energy Reporting (NGER) requires that
reporting entities provide a statistical a Note: The uncertainty estimates are from the 2006
estimate of the level of uncertainty IPCC Guidelines, volume 2.

51
IPIECA • API

The example uncertainty levels extracted b. GHG protocol calculation tool

from the NGER technical guidelines are those
The GHG Protocol (WRI, 2011) has
that are most relevant to the oil and natural
developed a spreadsheet-based tool
gas industry based on common fuel use and
that automates the aggregation steps
operations. They are primarily applicable to
associated with a basic uncertainty
simple calculation methods that use generic
assessment for GHG inventory data.
default emission factors. The data indicate that
The GHG Protocol tool and guidance
for such generic factors the default uncertainty
document are based on the IPCC
percentages vary from fuel to fuel and by type
Guidelines for National GHG Inventories.
of operation as a result of the assumed variable
The tool aggregates uncertainty assuming
range of compositions and activity levels for
a normal distribution of the variables using
each. Commercial liquid fossil fuels generally
a first-order error propagation method.
have a limited range of compositions, and
Use of this approach requires that the
therefore have a low associated uncertainty.
following assumptions are valid for the
Other self-generated fuels have higher default
data set:
uncertainties due to variability in compositions.
For higher order emission estimation methods,
• The errors in each parameters must
be normally distributed.
which rely all or in part on site-specific
measurements, the NGER guidelines recognize • There must be no biases (i.e., the
that the uncertainty associated with those estimated value is the mean value);
estimates should reflect the uncertainty • The estimated parameters must be
associated with the measured data that is the uncorrelated.
basis of the estimate, and not necessarily with • The individual uncertainties in each
the default assumptions listed in Table 4-5. parameter must be less than 60% of
The Clean Energy Regulator (CER) developed an the mean.
uncertainty calculator to assist NGER reporters If these assumptions are not valid for the
that use higher order methods for estimating data set, the Monte Carlo method can be
and reporting their emissions to this national used.
program (CER, 2012). The calculations and
factors incorporated in the calculators are in
accordance with the NEGR 2008 regulations and
the methodologies of Chapter 8 of the Technical
Guidelines (Australian Government, 2009).
NGER requires reporting the level of
uncertainty attributed to sampling and analysis
variation at the 95% confidence level, and to
report an assessment of the overall entity-wide
uncertainty for its direct emissions (Scope 1).
Reporters are not required to use the calculator,
and can use any method for calculating
uncertainty that complies with the regulatory
requirements. The tool’s calculations are based
on the following assumptions:
• The estimated parameters are
uncorrelated.
• Errors in each parameter are normally
distributed.
• No biases exist.
• Individual uncertainties for each parameter
must be less than 60% of the mean.

52
 ADDRESSING UNCERTAINTY IN OIL & NATURAL GAS INDUSTRY GREENHOUSE GAS INVENTORIES

5. UNCERTAINTY CALCULATION
Section Focus
OVERVIEW
Uncertainty is used to characterize the
dispersion of values that could be reasonably This section provides an overview
attributed to a measured quantity. Uncertainty of general uncertainty calculation
may be expressed as a qualitative ranking, such principles and more detailed guidance
as the letter ratings assigned in EPA’s AP-42 on the calculation methods that are
publication series (USA the EPA, 1currently applicable for defining single source
has995 with Supplements through 2000), uncertainties and for aggregating
or as a quantified value. For the purpose of them at the facility (or entity) level.
quantifying the uncertainty of a GHG inventory, Uncertainty aggregation equations and
this section first addresses measurement an example are provided. Methods for
uncertainty, then discusses uncertainty reducing uncertainty are also addressed.
associated with emission factors, and finally
The subsections address:
addresses the propagation of uncertainty.
A step-by-step tutorial with examples of • Measurement uncertainty
the uncertainty calculations is provided in • Uncertainty propagation
Appendix F. The aggregation of uncertainty
estimates for two example facility-based
• Uncertainty aggregation
GHG emission inventories are provided in • Reducing uncertainty
Appendices G and H.

5.1 Measurement uncertainty

method may produce results that are not very
At the most basic level, a GHG inventory precise (having highly dispersed observations),
is comprised of estimated emissions from but result in a good estimate on average, as
individual emission sources. For a given shown in Figure 5-1. The variability or precision
emission source, an emission estimate associated with such an estimate will decrease
generally consists of an emission factor and as more data points are collected.
some measure of the activity that results in
Bias, on the other hand, refers to how
the emission (referred to as the activity factor;
accurately a method estimates the true value.
see also Sections 2.3 and 2.4). Emissions from
An example of bias would be a meter that
multiple sources are then aggregated to
is not calibrated correctly and consistently
produce the inventory. The quantification of
overestimates the measurement. Ideally,
uncertainty should be applied at the emission
the data measurement scheme should be
source level (or grouping of similar emission
designed in a way to minimize bias. For
sources) and then propagated to the total
example, if the measuring device is well
inventory (as discussed in Section 2.5).
maintained and calibrated to minimize drift,
Activity factors generally are a measured there may be no bias in the measurement.
quantity, such as a count of equipment
Figure 5-1 shows an example of the error of
or measure of fuel consumed. Emission
an unbiased measurement over time, where
factors may be either based on site-specific
the errors are centered around zero. When
measurements or based on published values
these measurements are aggregated, some of
that were derived from averaging a variety of
the positive errors are offset by some of the
measurements. Where measurements are used
negative errors, resulting in a lower estimate of
for either activity factors or emission factors,
uncertainty for the aggregate measurement.
two components of uncertainty need to be
considered: precision and bias. Figure 5-2 shows an example of the error in a
biased measurement over time. Similar to the
unbiased case in Figure 5-1, some of the high
5.1.1 Precision and bias errors are offset with some of the low errors so
Precision, as defined by the API’s MPMS (see the uncertainty in the precision of the estimate is
Exhibit 1-1), is “the degree to which data lower for the aggregate. However, in Figure 5-2,
within a set cluster together”, which relates to the errors are not centered on zero as in Figure
the dispersion of the measurement values. A 5-1; they are centered on five. This means that
there is a bias in the estimate. Unlike precision, the
53
IPIECA • API

Figure 5-1. Measurement error over time of an unbiased estimate 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.

5.1.2 Confidence intervals

Uncertainty is commonly expressed in terms
of confidence intervals, where the confidence
Figure 5-2. Measurement error over time of a biased estimate intervals establish the lower and upper
tolerances associated with an estimated
number. Expressed as an absolute value, the
confidence interval is computed as:

s
±t ×
n (Equation 5-1)
Where:
s = standard deviation;
n = sample size; and
t = t-value for “n-1” degrees of
freedom
And
uncertainty due to the bias generally does not
decrease as more data points are collected. 1 n
Thus, precision and bias should be considered s= ∑
n − 1 i=1
( xi − x ) 2
(Equation 5-2)
separately, if possible.
Where:
If bias can be quantified, it should be corrected
and thus eliminated from consideration in
quantifying uncertainty. For example, if a
xi = the ith observation in the data
set and
fuel stream has two types of measurement
devices, data can be collected from both x = the mean of the data set.
devices to check for agreement. A bias would
n
be indicated if the measurements differed.
However, quantifying and correcting for ∑x i

bias might not always be practical since x= i =1

it often requires application of frequent

n (Equation 5-3)
calibration routines and implementation of
quality control procedures that would allow Tables for the Student’s t-distribution can be
instrument adjustments and/or corrections found in most basic statistics references. Most
under prescribed conditions. It will also require spreadsheet software programs have a function
proper quantification routines that account that will calculate the necessary t-value. This is the
for drift or other causes of bias between preferred method since the software generally
calibrations on an ongoing basis. retains more significant digits for the t-value than
a look-up table would display.

54
 ADDRESSING UNCERTAINTY IN OIL & NATURAL GAS INDUSTRY GREENHOUSE GAS INVENTORIES

5.1.3 Implementation of Pedigree Matrix Table 5-1. Example of Pedigree Matrix for
approach determining uncertainty scaling factors
based on data quality ratingsa
The Pedigree Matrix could be a useful tool when
measured single parameter uncertainties are Very
unknown. This approach may be used to elicit Indicator score Good Fair Poor
good
a structured expert’s judgment for determining
Precision 1.00 1.10 1.20 1.50
single parameter uncertainty values, which can
then be propagated using techniques such as Completeness 1.00 1.05 1.10 1.20
series expansion or Monte Carlo simulations, as Temporal
discussed below. 1.00 1.10 1.20 1.50
Representativeness
In the pedigree matrix approach, qualitative data Geographical
1.00 1.02 1.05 1.10
quality assessment is used to select appropriate Representativeness
uncertainty scaling factors for single parameters. Technological
Assessment of data quality should be conducted 1.00 1.20 1.50 2.00
Representativeness
separately for activity data and emission factors, a
Weidema and Wesnaes, 1996
which will be merged in the ensuing uncertainty
propagation analysis.
In this proposed adaptation, which is based on and represent the contribution to the square of
guidance provided by the GHG Protocol (WRI/ the geometric standard deviation.
WBCSD a, 2011), an uncertainty scaling factor is
Data sources used when compiling the
selected by assigning a data quality score (very
emissions inventories need to be assessed. The
good, good, fair and poor) to each of five data
total uncertainty, expressed as a 95% confidence
quality indicators (precision, completeness,
interval, SDg95 (the square of the geometric
temporal geographical and technological
standard deviation), is calculated using the
representativeness). The ranges of recommended
formula shown below (WRI/WBCSD a, 2011):
uncertainty scaling factors are shown in Table 5-1

SDg95 ≅ σ 2 = exp (ln U 1) 2 + (ln U 2) 2 + (ln U 3) 2 + (ln U 4) 2 + (ln U 5) 2 + (ln Ub) 2

(Equation 5-4)
Where:

U1 = uncertainty factor of precision

U2 = uncertainty factor of completeness
U3 = uncertainty factor of temporal representativeness
U4 = uncertainty factor of geographic representativeness
U5 = uncertainty factor of other technological representativeness
Ub = basic uncertainty factor

In addition to the uncertainty factors for the

data quality categories, one has to include
basic uncertainty factors for the kind of input
considered. Examples of applicable basic
uncertainty factors that are pertinent to
addressing GHG emissions uncertainty are
provided in Table 5-2. An example application
of this approach is provided in Section 5.3.2.

55
IPIECA • API

Table 5.2. Basic Pedigree Matrix uncertainty factors for selected pollutants a
Indicator score Very good Good
CO2 1.05 1.05
CH4 1.50 Not available
N0x N20 1.50 1.40
NMVOC total 1.50 1.50
NH3 1.50 1.20
Individual Hydrocarbons 1.50 2.00
a
Overview and Methodology, Data v2.0 (2007). Ecoinvent Report No. 1, Dübendorf, December 2007.
http://www.ecoinvent.org/fileadmin/documents/en/01_OverviewAndMethodology.pdf

5.2 Overview of uncertainty Through the propagation of uncertainty for all

propagation emissions in the inventory, the impact of small
emission sources with large uncertainties can be
Uncertainty propagation involves mathematically evaluated relative to the entire inventory. This
combining individual sources of uncertainty to evaluation can be used to identify and prioritize
establish an estimate of the overall uncertainty. emission sources that require more data to
Specific uncertainty propagation techniques are reduce the overall uncertainty of the inventory.
discussed in Section 5.2.1.
The second assumption is based on the
The following three assumptions are important normality of the distribution of the underlying
when applying the uncertainty propagation source data (i.e., symmetrical around the mean).
technique for overall uncertainty assessment According to the Central Limit Theorem, for a
(IPCC, 2000: Annex 1, section A1.4.3.1): large enough sample size (n>30), we can relax
the normality assumption but still assume that
1. The uncertainties are relatively small, which
the sampling distribution of the sample means is
is defined as the standard deviation divided
normally distributed (Casella and Berger, 1990).
by the mean value being less than 0.3.
Hence, if the calculated uncertainty is based
2. The uncertainties have Gaussian (normal) on statistical sampling of the population, one
distributions. would need to obtain more samples to approach
normality. Alternatively, we might consider data
3. The uncertainty values (i.e., the errors or transformation, i.e., mathematically transforming
uncertainties associated with the measured the data to a different scale and using that
data or reported values) are mutually transformed ‘normal’ distribution to derive the
independent. 95% confidence interval. For example, in the case
In many cases, the first assumption may be where the data distribution is skewed and the
difficult to meet. For example CH4 and N2O uncertainty is >100% of the mean (i.e., where
emissions often have very sparse data and the lower limit would be less than zero), the data
large associated uncertainties. Conducting could be transformed to a lognormal distribution.
a Monte Carlo simulation (discussed further This approach, however, requires the confidence
in Section 4.2.3) is an option if the standard interval to be transformed back to the original
deviation divided by the mean is greater scale to express the uncertainty in the original
than 0.3. However, Monte Carlo simulations units, which can introduce error. As a result, there
require a significant level of detail for the is a trend away from using transformational
description data to characterize the probability approaches due to issues in transforming the data
distributions. Without such information, the back to their original scale.
potential error introduced from incorrectly The third assumption states that there
specifying the distributions for a Monte Carlo is no significant covariance between the
simulation could outweigh the potential uncertainties that are to be combined, which
error that might be associated with applying is equivalent to saying that the errors or
an uncertainty propagation technique for uncertainties are independent or that there
sources with large uncertainties. Therefore, this is no correlation between the uncertainty
document suggests that the first assumption terms. The uncertainties in two quantities
can be relaxed for emission estimates with a would be considered independent if they were
small overall contribution to the GHG inventory. estimated by entirely separate processes and

56
 ADDRESSING UNCERTAINTY IN OIL & NATURAL GAS INDUSTRY GREENHOUSE GAS INVENTORIES

there were no common source of uncertainty. is calculated as the “square root of the
The uncertainties in two quantities would be sum of the squares” using the absolute
dependent if they had a common source of uncertainties, as shown in Equation 5-5.
uncertainty (Williamson et al, 1996).
Covariance between two uncertainty terms U ( abs ) X +Y +...=
N
U X2 + U Y2 + ... + U N2
can be addressed through an additional term
in the uncertainty propagation equations (Equation 5-5)
(discussed further in Section 4.2.2). However, Where:
the IPCC Good Practices document suggests
avoiding the need for the covariance term U(abs) = the absolute uncertainty.
in the equation by “…stratifying the data or
The absolute uncertainty values are
combining the categories where the covariance
used in the equations, and the resulting
occurs” (IPCC, 2001).
aggregated uncertainty (UX+Y+…+N) is also
on an absolute basis. Note that where a
constant is also included in the emission
5.2.1 Propagation equations
estimation calculation, the absolute
Four general equations for propagating uncertainty should include the constant.
uncertainty are used in this document and the This is demonstrated in the example
API Compendium for compiling the uncertainty provided in Appendix F.
associated with a GHG inventory. A general
For two uncertainty parameters that are
introduction to the equations is presented
related to each other, the equation becomes:
here, with example applications provided in
Appendix F.
Consider two quantities that can be measured: U ( abs )CorrelatedX +Y = U X2 + U Y2 + 2r (U X × U Y )
X and Y. The uncertainty for these values can
be expressed on an absolute basis as ±Ux and (Equation 5-6)
±Uy respectively, where U is calculated through Where:
statistical analysis (as represented by Equation
5-1), determined through the Monte Carlo r = the correlation coefficient between UX,
technique, or assigned by expert judgment. UY, (discussed further in Sections 5.2.3 and
Uncertainty may also be expressed on a relative Appendix F).
basis, generally reported as percentage: However, the IPCC Good Practices
guidance states, “Once the summation
U   UY  exceeds two terms and the covariance
±100  X  % or ± 100   % respectively
 X   Y  occurs, the use of the Monte Carlo
approach is preferable where data
Depending on the uncertainty propagation resources are available” (IPCC, 2000).
equation, the absolute or relative uncertainty
value may be required. In addition, selection of the
b. Uncertainty propagation for a product
propagation equation also depends on whether
(or quotient)
the uncertainties associated with the individual
parameters are independent or correlated. The equation for propagating uncertainties
from the product or quotient of two
a. Uncertainty propagation for a sum (or
or more measured and independent
difference)
quantities is similar to Equation 5-5.
Two potential equations are used for However, in this case the relative
computing the total uncertainty from the uncertainties are used, as shown in
addition or subtraction of two or more Equation 5-7. When multiplied by 100, the
measured quantities. The selection between resulting combined uncertainty (U(Rel)XxYxN)
the two equations depends on whether the is expressed as a percentage.
uncertainties associated with the measured
2 2 2
quantities, X and Y, are correlated.  U X   UY  UN 
U ( rel ) X ×= U (rel ) X ÷=
Y ÷...÷ N   +  + ... +  
For uncertainties that are mutually
Y ×...× N
 X   Y   N 
independent, or uncorrelated (i.e., the
uncertainty terms are not related to (Equation 5-7)
each other), the aggregated uncertainty

57
IPIECA • API

Equation 5-7 is used to estimate the These guidelines concentrate solely on the
uncertainty of a product or quotient uncertainty propagation method due to the
of two parameters (X and Y) where the potential to introduce further errors in assigning
uncertainties are correlated and positive the probability distributions for the Monte Carlo
values. Here also, relative uncertainty simulations. As stated previously, applying the
values are used in the equation and the uncertainty propagation methods, even where
resulting combined uncertainty is on a the assumptions are not met, is advised in these
relative basis. guidelines, particularly for emission sources
with a small contribution to the overall GHG
2 2
 U X   UY   U X UY  inventory. As data collection methods improve
U ( rel )CorrelatedX ×Y =   +  + 2r  ×  for GHG inventories, the ability to quantify
 X   Y   X Y  uncertainties will also improve.

(Equation 5-8)
5.2.3 Correlation coefficient
The correlation coefficient, r, used in Equations
c. Combining uncertainties
5-5 and 5-7, is a number between -1 and 1
It may be necessary to combine multiple that measures the linear relationship between
uncertainty parameters associated the errors or uncertainties of two measured
with a single measured value, such as parameters. The value of r is zero when the
combining uncertainties for precision and parameters are independent. As stated
bias. For uncertainty parameters that are previously, once the uncertainty propagation
independent, the combined uncertainty is exceeds two terms and covariance occurs, the
calculated using the absolute uncertainties use of the Monte Carlo approach (described
as shown in Equation 5-5. Similarly, for further in Section 4.0) is preferable (IPCC, 2001).
uncertainty parameters that are related to Additional details on calculating the correlation
each other, Equation 5-6 applies. coefficient are provided in Appendix F. A
simplified explanation follows.

5.2.2 Comparison of uncertainty For two terms that might be correlated, the
propagation and Monte Carlo errors or uncertainties are plotted against each
other. For the purpose of this discussion, Ux
Section 6.3.1 of the IPCC Good Practices represents the uncertainties of one variable
document compares the uncertainty plotted along the x-axis, and Uy represents the
propagation method and the Monte Carlo uncertainties of the second variable plotted
simulations (IPCC, 2000). It notes that on the y-axis. The correlation coefficient, r, is
the uncertainty propagation method’s determined by a linear regression of the Ux and
assumption of normality leads to symmetric Uy values.
95% confidence intervals whereas the Monte
Carlo method can take into account the fact If one suspects that the uncertainty parameters
that emissions are bounded below by zero are correlated, but data are not available to
to fit an asymmetric (and thus narrower) plot or calculate the correlation coefficient, the
confidence interval. If the data are skewed and following rule-of-thumb values could be applied
one transforms the data (discussed earlier), using expert judgment (Franzblau, 1958):
one could achieve the asymmetric confidence • r = 0: no correlation, the data are
intervals using uncertainty propagation, as well. independent
Since the Monte Carlo simulations can • r = ±0.2: weak correlation
assume a truncated distribution, the lower • r = ±0.5: medium correlation
confidence limits tend to be closer to the • r = ±0.8: strong correlation
mean than the upper confidence limits. The
IPCC Good Practices document goes on to
• r = ±1: perfect correlation, the data fall on a
straight line.
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.

58
 ADDRESSING UNCERTAINTY IN OIL & NATURAL GAS INDUSTRY GREENHOUSE GAS INVENTORIES

5.3 Uncertainty aggregation multiplication of an emission factor and activity

examples value, and their associated uncertainties,
so they can be aggregated across source
The following sections provide examples for categories. WRI has also developed a simple
aggregating emissions. spreadsheet tool that applies these calculations
(WRI/WBCSD b, 2011).

5.3.1 IPCC aggregation template Table 5-3 exhibits a generic template that
is an adaptation of the IPCC-recommended
IPCC provides an example calculation table approach for calculating aggregated
for estimating the aggregated uncertainty of uncertainties. This template could be tailored
GHG emissions, which is based on the error for specific oil and natural gas inventories by
propagation method for the multiplication expanding or collapsing the emission sources
of an emission factor and activity value (IPCC, and industry segments included based on the
2006, page 3.31). The approach entails the goals of the assessment and data availability.

Table 5-3. Adapted IPCC Template: Approach for Aggregating Uncertainty

A B C D E F G H I

Activity Activity Emission Emis- Emissions Uncertainty of Calculated

Gas
Data (e.g. Activity data Un- GHG Emission Factor Un- sions in in tonnes Emissions (expressed in
Industry
Quantity Data certainty emission Factor certainty tonnes CO2e ± Percent)
Segment
of fuel Units factor Units
used) ±% ±% A×D G × GWP 1 = C2 + F 2
CO2
Onshore
Production

Offshore
Production
Natural
Gas
Processing

CH4
Onshore
Production
Offshore
Production
Natural
Gas
Processing

2
 ln 
ΣH ∑
n
i =1 
× Hn 
 100 
TOTAL l= ×100%
∑ H

59
IPIECA • API

Table 5-4 provides an example using such a emissions from combusting natural gas.
table structure for calculating uncertainty, Equations 5-6 and 5-4 are applied, assuming
similar to the IPCC table and WRI tool. This that the variables are uncorrelated. More
example illustrates the aggregation of detailed examples are provided in Appendices
uncertainty for the combined CO2 and CH4 F and G.

Table 5-4. Example uncertainty calculation and reporting table

A B C D E F G H I
Emission
Source Activity Emission Emis- Emissions Uncertainty of Calculated
Activity
Description: data Factor sions in in tonnes Emissions (expressed in
Data (e.g. Activity GHG Emission
Example for Quantity Data Uncertainty emission Factor Uncertainty tonnes CO2e ± Percent)
Natural Gas of fuel Units factor Units
Combustion used) ±% ±% A×D G × GWP
1 = C2 + F 2
tonnes
MMBtu/
CO2 Emission 3000 +/- 5.0% 0.0732 CO2/ +/- 10.0% 219.6 219.6 11.2%
yr
MMBtu
tonnes
MMBtu/
CO4 Emission 3000 +/- 5.0% 3.01x10-6 CO4/ +/- 15.0% 0.009 0.19 15.8%
yr
MMBtu

2
 ln 
∑ i=1  100
n

Σ
× Hn 
H l= 
×100%
TOTAL ∑ H

219.79 24.59
= ×100% = 11.2%
219.79

Table 5-5. Pedigree Matrix example

Indicator Example Application Indicator Score
Precision The precision for the volume of gas combusted is good since the Good: 1.10
flow meter is calibrated annually. The precision for the composition
of the gas is good since the gas suppliers follow industry
specifications to measure gas composition. Formal verification is not
conducted on either flow or composition measurements.
Completeness The flow and composition measurements are specific to the natural Very good: 1.0
gas that is combusted. The CH4 emission factor associated with the
compressor is based on industry data, but this is considered below.
Temporal representativeness Recent flow measurements and composition data are applied. The Very good: 1.0
CH4 emission factor for the compressor is accounted for below.
Geographical The flow rate and composition data are for the example compressor. Very good: 1.05
representativeness The CH4 emission factor is based on compressors used in theUSA,
but is considered fair.
Technological The flow rate and composition data are for the example compressor. Very good: 1.0
representativeness
CH4 Combustion emissions From Table 5-2. This rating accounts for the use of a default 1.50
combustion emission factor.

60
 ADDRESSING UNCERTAINTY IN OIL & NATURAL GAS INDUSTRY GREENHOUSE GAS INVENTORIES

5.3.2 Pedigree Matrix example fuel combusted is metered, and the meter
is calibrated annually. The natural gas is
Table 5-5 provides an example application of
purchased from a local distribution company.
the Pedigree Matrix approach for determining
the uncertainty associated with CH4 emissions The uncertainty, expressed as a 95% confidence
from the combustion of natural gas in a interval, is calculated by applying Equation 5-4
compressor. For this example, the amount of and using the uncertainty factors from Table 5-5.

( lnU 1)2 + ( lnU 2)2 + ( lnU 3)2 + ( lnU 4)2 + ( lnU 5)2 + (ln Ub )2
SDg95 ≅ σ 2 =
exp
( ln1.1)2 + ( ln1.0)2 + ( ln1.0)2 + ( ln1.05)2 + ( ln1.0)2 + ( ln1.5)2
= exp
SDg95 = 1.52

The 95% probability range for this example 5.3.3 Uncertainty assessment from API
would be expressed as (median/SDg95) to Technical Report 2571
(median* SDg95). Hence, for this example, if the
Table 5-6 provides an illustration for performing
median CH4 emissions are 5,000 tonnes per
the uncertainty assessment in conjunction with the
year the 95% probability range wo uld
equations and processes set forth in the TR 2571
be 3,290 to 7,600 tonnes per year.
document, using average values as an example.
Table 5-6. Example of flow measurement total uncertaintya
Standard Sensitivity Total
Nominal Standard
Component Units Uncertainty Coefficient Uncertainty
Value Uncertainty
(Ux)% (Sx) (Ux×Sx)2
Unit Conversion Factor NA 1.899 0.0 0.0000 1.00 0.000
Discharge Coefficient Dimensionless 0.6008 0.00411 0.6834 1.00 0.467
Expansion Factor Dimensionless 0.9989 0.00014 0.0144 1.00 0.000
Bore Diameter Inches 0.5 0.0004 0.0750 2.04 0.023
Pipe Diameter Inches 2.067 0.0257 1.2425 0.04 0.002
Base Compressibility Dimensionless 0.9979 0.0010 0.1000 1.00 0.010
Base Temperature °Fahrenheit 60 0 0.0000 1.00 0.000
°Rankine 519.67
Flowing Pressure psia 100 0.46 0.4575 0.5 0.052
Universal Gas Constant ft-lbf/lbmol-°R 1545.33 0 0.0000 0.5 0.000
Gravitational Constant lbm-ft/lbf-s2 32.174 0 0.0000 1.00 0.000
Differential Pressure Inches-H2O 10 0.1120 1.1204 0.50 0.314
Base Pressure psia 14.73 0 0.0000 1.00 0.000
Molecular Weight lbm/lb-mol 16.799 0.1260 0.7500 0.50 0.141
Flowing Dimensionless 0.98664 0.007 0.7500 0.50 0.141
Compressibility
Flowing Temperature °Fahrenheit 60 2.598 0.5000 0.50 0.063
°Rankine 519.67
Installation Effect Dimensionless 1 0.02 2.0000 1.00 4.000
Factor
Meter Condition Factor Dimensionless 1 0.02 2.0000 1.00 4.000
Energy Content Btu/ft3 1050 7.88 0.7500 1.00 0.563
Carbon Content b
lbm/ft 3
0.035 0.0002625 0.7500 1.00 0.563
SUMMARY
Volume Flow Rate SCFH 848 Total Volume Uncertainty 3.04
Energy Flow Rate Btu per hour 890,320 Total Energy Uncertainty 3.13
Carbon Flow Rate Tons per hour 0.015 Total Carbon Uncertainty 3.13
a
Extracted from API TR 2571, Table A.3.5
b
Carbon Content is the fraction of carbon in the fluid expressed as percent by weight.
61
IPIECA • API

The example in Table 5-6 assumes a 2 in. line certain percentage level. A company may also
with a 0.5 in. orifice plate using a multivariable independently wish to refine its own uncertainty
transmitter for measuring temperature, limits, if it considers the uncertainties too large.
pressure, and pressure differential (DP). It also
The reader should note that none of the
assumes that the gas composition is measured
strategies mentioned here are aimed at reducing
at the fuel gas drum by an online analyzer;
the actual emissions of GHGs. That is a separate
hence lower molecular weight uncertainty
subject, and while emission reductions are
values are used.
achievable in some cases, they are beyond
The total uncertainty was derived by summing the scope of this report. This section focuses
the individual combined uncertainties (Ux × on reducing the mathematical and statistical
Sx)2 and taking the square root of the sum. uncertainty associated with an existing emission
The calculation for this example is done inventory.
three different ways to derive three different
Once a decision has been made to refine and
uncertainties (volume, energy, and carbon).
reduce the uncertainty associated with a given
• Volume Uncertainty: all components are emissions inventory, some strategic analysis
included with the exception of energy of the major sources contributing to the
content and carbon content; uncertainty is in order. It is important to know
• Energy Uncertainty: all components are which emission sources and which uncertainties
included with the exception of the carbon significantly contribute to the overall inventory
content; uncertainty. It would make little sense to refine
a term with large confidence bounds, but that
• Carbon Uncertainty: all components are
contributed very little to the overall inventory
included with the exception of the energy
uncertainty. It may also be useful to have a target
content.
uncertainty in mind. For example, if the current
total uncertainty is ±50.0 % and the company
5.4 Strategic reduction of wishes to reduce it to ±25.0%, then that target
uncertainty is useful in the analysis. As uncertainties
of individual values in the calculations are
examined, values that have uncertainties that
5.4.1 Periodic updates are already at or below the 25.0% target, are less
likely to be fruitful targets for reduction. The goal
Activities data that comprise a GHG emission is to find the largest contributors to emissions
inventory will likely change from year to year. that have the largest uncertainties.
Emission factor information may also vary
over time. However, aspects that impact the The largest contributors to uncertainty can
calculation of uncertainty such as measurement be determined by multiplying the emission
equipment or techniques, quantification estimate by the maximum error bound for each
methods, and emission factors, are relatively source. This would result in the upper bound
constant year-on-year. Therefore, it may emission estimate for the particular source. The
not be necessary to conduct an uncertainty emission estimates can then be sorted by the
assessment on an annual basis. A frequency of largest contributors. This is demonstrated in
every three to five years may be sufficient, but Table 5-7 on page 59 for the example crude oil
should ultimately be determined based on the production facility presented in Appendix G.
variability of the factors that contribute to the a. Onshore oil field
quantification of uncertainty.
Each facility should examine the major
categories of emissions and emission
5.4.2 Reducing uncertainty of emission uncertainty, and then examine specific
estimates emission sources within the category.
Prioritizing the largest sources of uncertainty
This section addresses the potential need to can be done with this simple approach.
refine the emission inventory to reduce the
uncertainty in the overall emission estimate. Figure 5-3 illustrates emissions for the
There may be several reasons to do this. In facility with each emission category
some specific locations, there may be state or and each gas converted to CO2e, using
national regulations, or guidelines for voluntary each gas’ GWP. It should be noted that
programs that suggest or require an emission although the GWP itself has uncertainty
inventory to have uncertainties lower than a associated with it, this report treats the

62
 ADDRESSING UNCERTAINTY IN OIL & NATURAL GAS INDUSTRY GREENHOUSE GAS INVENTORIES

Table 5-7. Emission uncertainty ranking for onshore oil production example
Maximum
Emissions Maximum
Source Type Source Uncertainty, Ranking
(tonnes CO2e /yr) Uncertainty, %
(tonnes CO2e /yr)
Combustion Boiler/heaters 5,210 8.77 457 9
Sources
Natural gas engines 14,100 15.6 2,200 6
Diesel engines 220 15.5 34
Flares 30,700 21.1 6,478 3
Fleet vehicles 129 19.2 25
Vented Sources Dehydration and Kimray 5,440 76.0 4,134 4
pump vents
Tanks-flashing losses 40,300 88.7 35,746 1
Amine unit 66,700 9.77 6,517 2
Pneumatic devices 3,360 49.2 1,653 7
Chemical injection pumps 2,530 106 2,682 5
Vessel blowdowns 3.65 319 12
Compressor starts 38.7 187 72 10
Compressor blowdowns 17.3 175 30
Well workovers 0.939 294 3
Other non-routine (PRVs) 6.81 319 22
Fugitive Sources Fugitive components 1,100 83.3 916 8
Fleet vehicle refrigeration, 1.30 112 1.5
R-314a
Indirect Emissions Electricity consumed 553 10.2 56

GWP as a selected constant. Therefore, no Figure 5-3. Onshore oil field: summary of co2
uncertainty is associated or propagated equivalent emissions and uncertainties
from the GWP values. Figure 5-3 also
illustrates the bounds of uncertainty (at
95% confidence) for each emission source
category.
Among the major types of emissions
in Figure 5-3, are three very significant
emission categories: vented sources
of CO2; vented sources of CH4; and
combustion sources of CO2. 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. Examination
of the uncertainties, as shown in the bars
on Figure 5-3, or the absolute uncertainty
values in Table 5-7, shows that vented
emissions of CH4 from tank flashing
losses are the most significant source of Therefore, should the company wish to
uncertainty, contributing 35,746 CO2e reduce uncertainty in the GHG emission
tonnes of uncertainty (based on 40,300 inventory from this onshore oil field
±88.7% CO2e tonnes). Carbon dioxide facility, these categories would be the
vented from the amine unit is the next primary targets for uncertainty reduction.
largest source of uncertainty, contributing These appear to be the sources where
6,517 CO2e tonnes of uncertainty. The third uncertainty reduction efforts could be
largest source are emissions from flares, effectively undertaken to refine the
with 6,478 CO2e tonnes of uncertainty. inventory and reduce the uncertainty.

63
IPIECA • API

b. Vented sources boilers/heaters, natural gas engines and

flares are ranked in the top ten highest
Within vented emissions, are 10 sources
emission sources, with flares contributing
listed in Table 5-7. The highest ranking
the largest uncertainty for the combustion
source, tank flashing losses, contributes
sources. Uncertainty associated with flares
a significant part of the vented emissions
was calculated to be 21.1%, or about
(34% of total vented emissions), as well
±6,500 tonnes CO2e/yr.
as the majority of the uncertainty for the
entire facility. That single category source’s By examining the calculations used to
uncertainty is 1,700 tonnes of methane, or estimate flaring emissions (refer to Section
about 35,700 tonnes of CO2e. Therefore, G.4), the following general strategies could
improvement of this estimate could greatly be selected by the operating company to
reduce uncertainty in the overall inventory. reduce uncertainty in flaring:
Examination of the detailed calculation • Refine the measurement of total gas flared
for that category (presented in Section (the activity factor), to reduce the activity
G.7), shows that the largest uncertainty is factor uncertainty from the current value
in the emission factor, which is a general of 15%; and/or
industry-wide emission factor. The • Refine the gas composition measurements
uncertainty in that factor is ±90.4%. This to reduce the uncertainty from 4%.
uncertainty can be reduced simply by
Refining the measurement of total gas
using an improved estimation method to
flared may result from many approaches,
determine tank flashing losses.
such as improving the meter quality (even
As elaborated in the API Compendium, other possibly replacing the meter); improving
emission estimation methodologies can be the quality control of the existing meter
used to estimate tank flashing emissions (such as number of calibrations and
with lower uncertainties. For example, if inspections); or improving the number of
the “EUB Rule of Thumb” approach (API measurements taken and recorded from
Compendium Section 5.4.1, 2009) were the meter that is used to calculate the total
applied instead of the default tank flashing gas flared (this assumes measurements
emission factor for the example onshore were not already continuous). These
production facility, the emissions for this approaches have varying costs, and some
category would be 20,500 ±49.7% tonnes may be cost-prohibitive. The company
CO2e.5 This one revision would change the would have to determine which approach
overall inventory emissions to 151,000 is the most cost effective.
±9.88% tonnes CO2e.
Refining the gas composition data may
The second largest source of uncertainty also come from several methods, such
for this facility is emissions from amine as taking additional routine samples,
units. Here also, a default emission factor installing a continuous gas analyzer, or
was applied with an uncertainty of 119% using a better analysis method. As with
(refer to Section G.8). Emissions from this the gas flow rate measurement, these
source could be estimated more accurately approaches have varying costs, and some
by applying a mass balance approach may be cost prohibitive. The company
or utilizing a process model, such as would have to determine which approach
AMINECalc (API, 1999). Although published is the most cost effective.
uncertainty values are not available for
If the company is effective in reducing
these methods, expert judgment would
the uncertainty in flare gas CO2 emissions,
suggest these methods are more accurate
it might then elect to proceed to the next
than the default emission factor.
largest uncertainty source. The end-user
c. Combustion CO2 will have to recalculate total emissions
for the company or facility each time,
If the company decided to take the next
and determine if the uncertainty goal or
step of emission reductions, it may target
target has been reached.
the next largest category of uncertainty.
As shown in Table 5-7, emissions from

5
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.

64
 ADDRESSING UNCERTAINTY IN OIL & NATURAL GAS INDUSTRY GREENHOUSE GAS INVENTORIES

Alberta Energy Regulator (AER, 2013 American Petroleum Institute (API b,

Directive 017, Measurement Requirements 2006). Manual of Petroleum Measurement
for Upstream Oil and Gas Operations, Systems, Chapter 14.1, “Collecting and
Formerly released by the Energy Resource Handling of Natural Gas Samples for
Conservation Board. Last released May Custody Transfer,” API Publications,
15, 2013. http://www.aer.ca/documents/ Washington DC, 6th Edition, February
directives/Directive017.pdf. Last accessed 2006, Reaffirmed September 2011.
September 12, 2014.
American Petroleum Institute (API, 2007).
American Gas Association (AGA, 2008). Manual of Petroleum Measurement
Greenhouse Gas Emission Estimation Systems, Chapter 14.10, “Measurement
Guidelines for the Natural Gas Distribution of Flow to Flares,” API Publications,
Sector; American Gas Association, Washington DC, 1st Edition, July 2007,
Washington DC, April 2008. Reaffirmed June 2012.
American Petroleum Institute (API, 1985). American Petroleum Institute (API,
Manual of Petroleum Measurement 2009). Compendium of GHG Emissions
Standards, Chapter 13, Statistical Aspects Methodologies for the O&G Industry,
of Measuring and Sampling, Part 1, American Petroleum Institute, Washington
“Statistical Concepts and Procedures DC; Version 3.0, August 2009.
in Measurement,” API Publications,
American Petroleum Institute (API, 2011).
Washington DC, First Edition June 1985,
Fuel Gas Measurement, Technical Report
Reaffirmed February 2011.
2571, 1st Edition, March 2011.
American Petroleum Institute (API,
American Petroleum Institute (API,
1999). Amine Unit Emissions Model
2013). Carbon Content, Sampling, and
AMINECalc Version 1.0, API Publication
Calculation, Technical Report 2572, 1st
4679-1999, January 1999, Errata June
Edition, May 2013.
1999. http://www.techstreet.com/
api/searches/1088133 (Accessed American Petroleum Institute (API, 2013).
September 12, 2014). Manual of Petroleum Measurement
Systems, Chapter 14.3, “Concentric,
American Petroleum Institute (API a, Square-Edged Orifice Meters,” API
2006). Manual of Petroleum Measurement Publications, Washington DC, 4th Edition,
Systems, Chapter 14, “Natural Gas Fluids November, 2013.
Measurement,” Sections 1 to 10, API
Publications, Washington DC, 6th Edition,
February 2006, Reaffirmed September 2011.

65
IPIECA • API

6. REFERENCES
AER (2013). Directive 017: Measurement Requirements for Upstream Oil and Gas Operations. Alberta
Energy Regulator. Formerly released by the Energy Resource Conservation Board. Last released 15
May 2013. www.aer.ca/documents/directives/Directive017.pdf (Accessed 12 September 2014.)

AGA (2008). Greenhouse Gas Emission Estimation Guidelines for the Natural Gas Distribution Sector.
American Gas Association, Washington D.C., 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’. First Edition, June 1985;
reaffirmed February 2011. American Petroleum Institute. API Publications, Washington D.C.

API (1999). Amine Unit Air Emissions Model, AMINECalc Version 1.0. API Publication 4679. Software by
American Petroleum Institute, 31 January 1999 (errata June 1999). http://www.techstreet.com/api/
searches/1088133 (Accessed 12 September 2014.)

API (2006a). Manual of Petroleum Measurement Systems. Chapter 14: ‘Natural Gas Fluids
Measurement’, Sections 1–10. 6th Edition, February 2006; reaffirmed September 2011. American
Petroleum Institute. API Publications, Washington D.C.

API (2006b). Manual of Petroleum Measurement Systems. Chapter 14.1: ‘Collecting and Handling of
Natural Gas Samples for Custody Transfer’. 6th Edition, February 2006; reaffirmed September 2011.
American Petroleum Institute. API Publications, Washington D.C.

API (2007). Manual of Petroleum Measurement Systems. Chapter 14.10: ‘Measurement of Flow to
Flares’. 1st Edition, July 2007; reaffirmed June 2012. American Petroleum Institute. API Publications,
Washington D.C.

API (2009). Compendium of Greenhouse Gas Emissions Estimation Methodologies for the Oil and Natural
Gas Industry. Version 3.0, August 2009. American Petroleum Institute, Washington D.C.

API (2011). Fuel Gas Measurement. API Technical Report 2571, 1st Edition, March 2011. American
Petroleum Institute, Washington D.C.

API (2013). Carbon Content, Sampling, and Calculation. API Technical Report 2572, 1st Edition, May
2013. American Petroleum Institute, Washington D.C.

API (2013). Manual of Petroleum Measurement Systems. Chapter 14.3: ‘Concentric, Square-Edged
Orifice Meters.’ 4th Edition, November 2013. American Petroleum Institute. API Publications,
Washington D.C.

API/IPIECA (2007a). Oil and Natural Gas Industry Guidelines for Greenhouse Gas Reduction Projects.
March 2007. IPIECA, London.

API/IPIECA (2007b). Greenhouse Gas Emissions Estimation and Inventories: Addressing Uncertainty and
Accuracy. Report of the IPIECA/API Workshop held in Brussels, Belgium on 16 January 2007. IPIECA,
London, January 2007. http://www.ipieca.org/publication/greenhouse-gas-emissions-estimation-
and-inventories-addressing-uncertainty-and-accuracy (Accessed 12 September 2014.)

API/IPIECA/OGP (2011). Petroleum industry guidelines for reporting greenhouse gas emissions. 2nd
Edition. American Petroleum Institute/IPIECA/International Association of Oil and Gas Producers.
IPIECA, London, November 2011.

ASTM D1826-94 (2010). Standard Test Method for Calorific (Heating) Value of Gases in Natural Gas Range
by Continuous Recording Calorimeter. Originally approved in 1961. Last previous edition approved in
2003 as D1826-94 (2003). ASTM International, West Conshohocken, PA, 2010, www.astm.org

66
 ADDRESSING UNCERTAINTY IN OIL & NATURAL GAS INDUSTRY GREENHOUSE GAS INVENTORIES

ASTM D1945-03 (2010). Standard Test Method for Analysis of Natural Gas by Gas Chromatography.
Current edition approved 10 May 2003. Reapproved 2010. Originally approved in 1962. Last previous
edition approved in 2001 as D1945–96 (2001). ASTM International, West Conshohocken, PA, 2010,
www.astm.org

ASTM D1946-90 (2011). Standard Practice for Analysis of Reformed Gas by Gas Chromatography.
Current edition approved 1 June 2006. Reapproved 2011. Originally approved in 1962. ASTM
International, West Conshohocken, PA, 2011, www.astm.org

ASTM D2650-10 (2010). Standard Test Method for Chemical Composition of Gases by Mass Spectrometry.
Previous edition published 1 November 2004 as D2650-04. ASTM International, West Conshohocken,
PA, 2010, www.astm.org

ASTM D3588-98 (2011). Standard Practice for Calculating Heat Value, Compressibility Factor, and
Relative Density of Gaseous Fuels. Originally approved in 1998. Previous edition approved in 2003 as
D3588-98 (2003). ASTM International, West Conshohocken, PA, 2011, www.astm.org

ASTM D4809-13 (2013). Standard Test Method for Heat of Combustion of Liquid Hydrocarbon Fuels
by Bomb Calorimeter (Precision Method). Previous edition approved in 2009 as D4809-09ae1. ASTM
International, West Conshohocken, PA, 2013, www.astm.org

ASTM D4891-13 (2013). Standard Test Method for Heating Value of Gases in Natural Gas and Flare Gases
Range by Stoichiometric Combustion. Previous edition approved in 2006 as D4891–89 (2006). ASTM
International, West Conshohocken, PA, 2013, www.astm.org

ASTM D5720-95 (2009). Standard Practice for Static Calibration of Electronic Transducer-based Pressure
Measurement Systems for Geotechnical Purposes. Previous edition approved in 2002 as D5720-95
(2002). ASTM International, West Conshohocken, PA, 2009, www.astm.org

ASTM D5291-10 (2010). Standard Test Methods for Instrumental Determination of Carbon, Hydrogen,
and Nitrogen in Petroleum Products and Lubricants. Previous edition approved in 2009 as D5291-09.
ASTM International, West Conshohocken, PA, 2010, www.astm.org

ASTM D5865-13 (2013). Standard Test Method for Gross Calorific Value of Coal and Coke. Last previous
edition approved in 2012 as D5865-12. ASTM International, West Conshohocken, PA, 2013, www.
astm.org

ASTM D7314-10 (2010). Standard Practice for Determination of the Heating Value of Gaseous Fuels using
Calorimetry and On-line/At-line Sampling. Previous edition approved in 2008 as D7314-08. ASTM
International, West Conshohocken, PA, 2010, www.astm.org

ASTM UOP539-12 (2012). Refinery Gas Analysis by Gas Chromatography. Previous edition UOP539-97.
ASTM International, West Conshohocken, PA, 2012, www.astm.org

Australian Government (2009). National Greenhouse and Energy Reporting (Measurement)

Determination 2008. Version 1.1, Chapter 8: ‘Assessment of uncertainty’. Document ID F2009C00576,
Office of Legislative Drafting and Publishing, Attorney-General’s Department, Canberra, Australia,
June 2009. www.comlaw.gov.au/Details/F2009C00576 Last accessed September 12, 2014.
Department of Climate Change,

CARB (2012). Regulation for the Mandatory Reporting of Greenhouse Gas Emissions. Final Regulation.
California Air Resources Board, Sacramento, California, 19 December 2012.

Casella, G., and Berger, R. L. (1990). Statistical Inference. Wadsworth & Brooks/Cole, Belmont, California.

CEPA (1999). Canadian Environmental Protection Act, 1999.

http://www.ec.gc.ca/lcpe-cepa/default.asp?lang=En&n=26A03BFA-1

67
IPIECA • API

CER (2012). Uncertainty Calculator 2013-14 (Microsoft® Excel™ document). Australian Government
Clean Energy Regulator, National Greenhouse and Energy Reporting (NGER) Scheme. September
2014. www.cleanenergyregulator.gov.au/National-Greenhouse-and-Energy-Reporting/Forms-and-
calculators/Pages/default.aspx (Last accessed 12 September 2014.)

Cochran, W. G. (1977). Sampling Techniques. 3rd Edition. John Wiley & Sons, Inc., New York.

Coleman, H. W. and Steele, W. G. (1989). Experimentation and Uncertainty Analysis for Engineers. John
Wiley & Sons, New York.

EC (2012). Guidance Document: The Monitoring and Reporting Regulation – Guidance on Uncertainty
Assessment. MRR (Monitoring and Reporting Regulation) Guidance document No. 4, Final Version of 5
October 2012. European Commission, Directorate-General Climate Action.
http://ec.europa.eu/clima/policies/ets/monitoring/docs/gd4_guidance_uncertainty_en.pdf
(Last accessed 12 September 2014.)

EEA (2013). EMEP/EEA air pollutant emission inventory guidebook: Technical guidance to prepare
national emission inventories. European Environment Agency Technical report No. 12/2013.
Copenhagen, Denmark. www.eea.europa.eu//publications/emep-eea-guidebook-2013 (Last
accessed 12 September 2014.)

EPA (1971). ‘Method 21 Guidelines’. In Standards of Performance for New Stationary Sources – Title 40
of the Code of Federal Regulations (40 CFR), Chapter 1, Subchapter C, Part 60, Appendix A-7: ‘Test
Methods 19 through 25E’. www.ecfr.gov/cgi-bin/text-idx?SID=1d40301728f289bfce4a3216ebdf2db4
&node=pt40.8.60&rgn=div5#ap40.8.60.a

EPA (1995). Technology Transfer Network, Clearinghouse for Inventories & Emissions Factors
(website): ‘Emissions Factors & AP-42, Compilation of Air Pollutant Emission Factors’. United States
Environmental Protection Agency, Research Triangle Park, North Carolina. www.epa.gov/ttn/chief/
ap42/ (Last accessed 12 September 2014.)

EPA (2008). ‘Alternative work practice for equipment leaks’. In Consolidated Federal Air Rule – Title 40 of
the Code of Federal Regulations (40 CFR), Chapter 1, Subchapter C, Part 65, Subpart A, Section 65.7:
‘Monitoring, recordkeeping, and reporting waivers and alternatives, and alternative work practice for
equipment leaks. www.ecfr.gov/cgi-bin/text-idx?SID=e80816c4cda3834bac4e17f73b08cba4&node=
se40.16.65_17&rgn=div8

EPA (2009). Mandatory Greenhouse Gas Reporting – Title 40 of the Code of Federal Regulations (40
CFR), Chapter 1, Subchapter C, Part 98. Published in the Federal Register (www.regulations.gov) on
30 October 2009 under Docket ID No. EPA-HQ-OAR-2008-0508-2278, and became effective on 29
December 2009; implementation of 40 CFR Part 98 is referred to as the Greenhouse Gas Reporting
Program (GHGRP). www.ecfr.gov/cgi-bin/text-idx?SID=dba613069e26d03a053c2b80fe0e07f9&node
=40:21.0.1.1.3&rgn=div5 (Last accessed 12 September 2014).

ETSG (2007). Compendium of notes ‘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.’
Document ref. 2007/589/EC. European Union Emissions Trading System (EU-ETS), Emissions Trading
Technical Support Group (ETSG), Brussels, Belgium, 18 July 2007. http://www.emissions-euets.com/
attachments/232_M&R%20Guidelines%20589_2007.pdf (Last accessed 12 September 2014.)

Franzblau, A. (1958). A Primer of Statistics for Non-Statisticians. Chapter 7, Harcourt, Brace & World,
New York.

GTI (2000). GRI-GLYCalc™ Version 4.0, Software, Ref. GRI-00/0102. Gas Technology Institute, Illinois.
http://sales.gastechnology.org/000102.html (Last accessed 12 September 2014.)

68
 ADDRESSING UNCERTAINTY IN OIL & NATURAL GAS INDUSTRY GREENHOUSE GAS INVENTORIES

INGAA (2005). Greenhouse Gas Emission Estimation Guidelines for Natural Gas Transmission and Storage.
Revision 2.0, 28 September 2005. Interstate Natural Gas Association of America, Washington, D.C.
http://www.ingaa.org/File.aspx?id=5485

IPCC (EFDB). Emissions Factors Database (website), National Greenhouse Gas Inventories Programme
(NGGIP) of the Intergovermental Panel on Climate Change.
www.ipcc-nggip.iges.or.jp/EFDB/main.php (Last accessed 12 September 2014.)

IPCC (1996). 1996 IPCC Guidelines for National Greenhouse Gas Inventories. Intergovernmental Panel on
Climate Change, Geneva. http://www.ipcc-nggip.iges.or.jp/public/gl/invs1.html

IPCC (2000). Good Practices Guidance and Uncertainty Management in National Greenhouse Gas
Inventories. Good practice guidance report accepted by the IPCC Plenary at its 16th session held
in Montreal on 1–8 May 2000; Corrigendum, 15 June 2001. Intergovernmental Panel on Climate
Change, Geneva. www.ipcc-nggip.iges.or.jp/public/gp/english

IPCC (2006). 2006 IPCC Guidelines for National Greenhouse Gas Inventories. Volume 1: ‘General
Guidance and Reporting’, Chapter 3: ‘Uncertainties’. Intergovernmental Panel on Climate Change,
Geneva. http://www.ipcc-nggip.iges.or.jp/public/2006gl/pdf/1_Volume1/V1_3_Ch3_Uncertainties.
pdf (Last accessed 12 September 2014.)

ISO (2002). ISO 6974-6:2002: Natural gas – Determination of composition with defined uncertainty
by gas chromatography. Part 6: ‘Determination of hydrogen, helium, oxygen, nitrogen, carbon
dioxide and C1 to C8 hydrocarbons using three capillary columns.’ First edition, 15 October 2002.
International Organization for Standardization, Geneva, Switzerland.

ISO (2005). ISO 5168:2005: Measurement of fluid flow – Procedures for the evaluation of uncertainties.
International Organization for Standardization, Geneva, Switzerland.

ISO (2006). ISO 14064-1:2006: Greenhouse gases. Part 1: ‘Specification with guidance at the
organization level for quantification and reporting of greenhouse gas emissions and removals.’
International Organization for Standardization, Geneva, Switzerland.

ISO/IEC (2005). ISO/IEC 17025:2005: General requirements for the competence of testing and calibration
laboratories. Current version effective as of 12 May 2005. Revises ISO/IEC 17025:1999. International
Organization for Standardization, Geneva, Switzerland

ISO/IEC (2008). ISO/IEC Guide 98-3:2008: Uncertainty of measurement – Part 3: Guide to the expression
of uncertainty in measurement (GUM:1995). International Organization for Standardization, Geneva,
Switzerland. http://www.iso.org/iso/catalogue_detail.htm?csnumber=50461

Lloyd, S. M and Reis, R. (2007). ‘Characterizing, propagating, and analyzing uncertainty in life-cycle
assessment – a survey of quantitative approaches.’ In Journal of Industrial Ecology, Vol. 11, No. 1,
pp.161–179, January 2007.

Shires, T. M., and Harrison, M. R. (1996). Methane Emissions from the Natural Gas Industry—Volume 6:
Vented and Combustion Source Summary. Final Report, refs. GRI-94/0257.23 and EPA-600/R-96-080f.
Gas Research Institute and US Environmental Protection Agency, June 1996. http://www.epa.gov/
gasstar/documents/emissions_report/6_vented.pdf

Weidema, B. P. and Wesnaes, M.S. (1996). ‘Data quality management for life cycle inventories—an
example of using data quality indicators’. In Journal of Cleaner Production, Vol. 4, No. 3-4, pp. 167–174.

69
IPIECA • API

WCI (2009). Final Essential Requirements for Mandatory Reporting. Western Climate Initiative, 15 July
2009. www.westernclimateinitiative.org/component/remository/Reporting-Committee-Documents/
Final-Essential-Requirements-for-Mandatory-Reporting (Last accessed 12 September 2014).

Williamson, H. J., Hall, M. B. and Harrison, M. R. (1996). Methane Emissions from the Natural Gas
Industry—Volume 4: Statistical Methodology. Final Report, refs. GRI-94/0257.21 and EPA-600/R-96-
080d, Gas Research Institute and US Environmental Protection Agency, June 1996. http://www.epa.
gov/gasstar/documents/emissions_report/4_statisticameth.pdf (Last accessed 12 September 2014.)

WRI/WBCSD (2004). The Greenhouse Gas Protocol: A Corporate Accounting and Reporting Standard.
Revised Edition, March 2004. World Resources Institute and World Business Council for Sustainable
Development. http://www.ghgprotocol.org/standards/corporate-standard (Last accessed 12
September 2014).

WRI/WBCSD (2011a). Greenhouse Gas Protocol: Quantitative Inventory Uncertainty. Guidance for
companies wishing to quantify their inventory uncertainty. World Resources Institute and World
Business Council for Sustainable Development, October 2011. http://www.ghgprotocol.org/files/
ghgp/tools/Quantitative%20Uncertainty%20Guidance.pdf (Last accessed 12 September 2014.)

WRI/WBCSD (2011b). Uncertainty Calculation Tool (ghg-uncertainty.xlsx). World Resources Institute

and World Business Council for Sustainable Development, October 2011, http://www.ghgprotocol.
org/calculation-tools/all-tools (Last accessed 12 September 2014.)

70
ADDRESSING UNCERTAINTY IN OIL & NATURAL GAS INDUSTRY GREENHOUSE GAS INVENTORIES

71
BG Group Mærsk Oil Santos
BP Marathon Oil Saudi Aramco
Chevron Nexen Shell
CNOOC Noble Energy SNH
ConocoPhillips NOC Libya Statoil
EDF OMV Talisman
eni Petrobras Total
ExxonMobil Petronas Tullow Oil
Hess Petrotrin Woodside Energy
Hunt Oil PTT EP

IPIECA also has an active global network of oil and gas industry association members.
Please refer to our website for a full list.
IPIECA is the global oil and gas industry association for environmental and social issues. It develops,
shares and promotes good practices and knowledge to help the industry improve its environmental and
social performance, and is the industry’s principal channel of communication with the United Nations.

Through its member-led working groups and executive leadership, IPIECA brings together the collective
expertise of oil and gas companies and associations. Its unique position within the industry enables its
members to respond effectively to key environmental and social issues.
IPIECA
5th Floor, 209–215 Blackfriars Road, London SE1 8NL, United Kingdom
Telephone: +44 (0)20 7633 2388 Facsimile: +44 (0)20 7633 2389
E-mail:
Members [email protected] Internet: www.ipieca.org

Addax
© IPIECAPetroleum Husky Energy
2014 All rights reserved. Qatargas
Anadarko INPEX RasGas
Bashneft KPC Repsol
BG Group Mærsk Oil Santos
BP Marathon Oil Saudi Aramco
Chevron Nexen Shell
The American Petroleum Institute is the primary trade association in the United States representing
the oil and natural gas industry,
CNOOC and the only oneSNH
Noble Energy representing all segments of the industry.
ConocoPhillips NOC Libya Statoil
Representing one of the most technologically advanced industries in the world, API’s membership
includes more than 400 corporations
EDF OMV involved in all Talisman
aspects of the oil and gas industry, including
exploration and production, refining and marketing, marine and pipeline transportation and service
eni
and supply companies to the Petrobras Total API is headquartered in Washington, D.C.
oil and natural gas industry.
and has offices in 27 state Petronas
ExxonMobil capitals and provides its members
Tullow Oilwith representation on state issues in
33 states. API provides a forum for all segments of the oil and natural gas industry to pursue public
Hess Petrotrin
policy objectives and advance the interests of the Woodside
industry. Energy
API undertakes in-depth scientific,
technical
Hunt Oil and economic research
PTT EP to assist in the development of its positions, and develops
standards and quality certification programmes used throughout the world. As a major research
institute, API supports these public policy positions with scientific, technical and economic research.

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

Telephone: +1 202 682 8000 Internet: www.api.org
IPIECA also has an active global network of oil and gas industry association members.
Please refer to our website for a full list.

IPIECA
5th Floor, 209–215 Blackfriars Road, London SE1 8NL, United Kingdom
Telephone: +44 (0)20 7633 2388 Facsimile: +44 (0)20 7633 2389
E-mail: [email protected] Internet: www.ipieca.org

© IPIECA 2014 All rights reserved.

© IPIECA / API 2015 All rights reserved