Accuracy of Refinery
Emission Factors
Presented by
Scott Evans
Clean Air Engineering
What They Aren’t
They are not estimates of emissions from specific
emission units
They are not to be used for NSR/PSD modeling
They are not for establishing or complying with permit
limits
Although they are routinely used
for all these purposes
What They Aren’t
AP-42 Introduction
What They Are
They are averages of emission data from
multiple sources with varying degrees of
quality that may or may not share important
characteristics with any specific emission
unit
What They Are
AP-42 Introduction
An Emission Factor Is…
James H. Southerland, NC Department of Environment and Natural Resources
a number which passes as an accurate tool for approximation of
emissions,
is prolifically developed from information that is extremely
incomprehensible,
calculated with micron level precision,
involving extremely vague assumptions,
based on debatable data from inconclusive tests and incomplete
experiments,
using instruments of problematic accuracy,
by persons of doubtful reliability and rather dubious mentality.
Problems
Many factors developed in the 1970’s with primitive (by
today’s standards) sampling technology.
Many early factors are “fixed value” factors that do not
take relevant activity and operating parameters into
account.
e.g. AP-42 factor for loading shallow draft barge
w/crude
The greater the variation in actual emissions, the
greater the potential uncertainty in an emissions
estimate that does not account for this variation.
Quality Ratings
Data Attribute Ranking System (DARS)
Alphabetic scale: A, B, C, D, E, U
Qualitative not Quantitative
Bears NO RELATIONSHIP to accuracy or reliability
Primary basis is number of data points
Measurements can be very accurate, but represent only
emissions occurring for some specific combination of
conditions at the time of measurement.
Emission Factor Uncertainty
Current practice is to ignore uncertainty
Since uncertainty in emission factors is unknown,
uncertainty in inventories is unknown
How are sound regulatory and management decisions
made under these circumstances?
EPA under pressure (Nat’l Academy of Sciences, IPCC)
to include uncertainty in emission factor development
Emission Factor Uncertainty
Recent study by EPA shows that for any given unit,
actual emissions can range from 20% to 400% of
published emission factor.
Variability is factor dependent but is unknown for
most factors.
Picard study has quantified factor variability for many
sources in the oil and gas industry
Sources of Variability
Test method used
Test equipment used
Operating pressures and temperatures
Types of equipment in specific services
Equipment size
Maintenance practices
EPA’s Problem
Source:
EPA Solutions
Creation of Emission Factor and Policy Applications
Group (EFPAC) in 2003
Have performed some studies but no action yet
Funding remains a problem
EFPAC is actively looking for partners
API for TANKS and Turbines/Gas-fired Combustion
LDAR and Method 21
Screening method to determine concentration surrounding
leaking components
Promulgated in 1981?
Lax quality specifications – direct calibration precision
±10%! EPA now requires ±2% for direct calibrations
Allows use of generic Response Factors
Results affected by ambient conditions, response factor
uncertainty, operator experience
Results go into correlation equations of dubious value
Bottom line: results ±300% (Lott 1999)
DIAL Studies
Differential Absorption Light Detection and Ranging
UV and IR
Alberta, Texas, Sweden, UK
Studies show poor correlation between DIAL and
standard emission factors
Questions as to whether studies were conducted over a
long enough time period to represent annual emissions
Refinery VOC Comparison
Source: Alberta Research Council, Chambers, 2006
Refinery Benzene Comparison
Source: Alberta Research Council, Chambers, 2006
Tanks
TANKS software calculations based on AP-42 Section
7.1
From AP-42 Section 7.1 Background Report
Flares
Most emission factors for flares are at 98% based on
1980’s studies (Pohl, et. al.)
These studies done at high flow conditions on
unassisted flares.
Recent consent decree testing of flare combustion
efficiency with Passive FTIR technology (steam assisted
flares)
Preliminary results indicate that over-steaming may
result in combustion efficiencies significantly lower
than 98% particularly in low flow (i.e. high turndown)
situations.
Cooling Tower PM
AP-42 factor based on a fatally flawed chromate study
from 1980’s
AP-42 factor is about 0.02% of circulating water.
Typical drift rates measured by CTI Test Code ATC-
140 average around 0.001%.
Current state-of-the-art systems are guaranteed at
0.0005% of the circulating water.
Review Draft of “Emission Estimation Protocol for
Petroleum Refineries” relies on AP-42 only for PM.
The Future of Emission
Factors
Include quantitative uncertainty estimates
More site-specific factor development
Industry continues as main driver in EF development
Substantial revisions based on improved measurement
technology (e.g. DIAL and IR cameras)
EPA and other agencies pushing for direct
measurement rather than reliance on emissions
Recommendations
Don’t use emission factors blindly. Understand the
source of the factor and its limitations. Use AP-42
background documents when available.
Develop site-specific factors when possible
Focus on data quality improvement on the larger
sources.
Use data quality scoring to evaluate overall
improvements in data quality.
Sources of Emission Factors
EPA AP-42
“Emission Estimation Protocol for Petroleum Refineries”
CARB “California Implementation Guidelines for
Estimating Mass Emissions of Fugitive Hydrocarbon
Leaks at Petroleum Facilities”
API 4615 “Emission factors for Oil and Gas Production
Operations” and others
Many GHG factor sources (IPCC, WRI/WBCSD, etc)
Questions?
Scott Evans
sevans@cleanair.com