Technical Support Document (TSD)

for the Transport Rule

Docket ID No. EPA-HQ-OAR-2009-0491

 Power Sector Variability

U.S. Environmental Protection Agency

Office of Air and Radiation

July 2010

Power Sector Variability

This Technical Support Document (TSD) provides information in support of
section IV.F, “Emission Reduction Requirements Including
Variability”, in the preamble to the proposed Transport Rule.  This
TSD is organized as follows:

1. Introduction

2. Estimating year-to-year variability in emissions

3. Estimating variability over a multi-year time period

4. Results of an analysis done using the air quality assessment tool

1. Introduction.

Section IV in the preamble of the proposed Transport Rule discusses
EPA’s approach to define “significant contribution” and
“interference with maintenance” with respect to the 1997 8-hour
ozone and annual fine particle (PM2.5) National Ambient Air Quality
Standards (NAAQS) and the 2006 24-hour PM2.5 NAAQS.  As discussed in
preamble section IV, the EPA has identified the emissions that must be
prohibited by each state to eliminate the state’s significant
contribution and interference with maintenance.  To facilitate
implementation of the requirement that these emissions be eliminated,
the EPA also developed SO2, annual NOX, and ozone season NOX state
emissions budgets based on its projections of state-by-state power
sector emissions in an average year after the elimination of the
prohibited emissions. 

However, because of the unavoidable variability in baseline emissions
– resulting from the inherent variability in power system operations
– state-level emissions may vary somewhat from year to year after all
significant contribution and interference with maintenance that EPA has
identified in the Transport Rule proposal are eliminated.  This
variability in emissions occurs even when the emission rates of the
units within the state do not change.  For this reason, as discussed in
preamble section IV.F, the EPA has determined that it is appropriate to
develop variability limits for each state budget.  These limits are used
to identify the range of emissions that EPA believes may occur in each
state following the elimination of all significant contribution and
interference with maintenance.  This TSD describes the analyses that EPA
performed to estimate the inherent variability in emissions from the
power sector and to determine variability limits based on that inherent
variability.

Preamble section IV.D discusses EPA’s proposed approach to quantify
for each upwind state the emissions that significantly contribute to
nonattainment or interfere with maintenance downwind for the existing
ozone and PM2.5 NAAQS.   Preamble section IV.E discusses the development
of state emissions budgets for SO2, annual NOX, and ozone season NOX. 
Preamble section IV.F discusses the inherent variability in electric
power system operations and proposes variability limits on emissions for
each state covered by the proposed Transport Rule.  As explained in
section IV.F, the EPA proposes to calculate variability limits for each
state and to use those variability limits in conjunction with the state
budgets (which are based on expected average conditions) to provide
limited emissions flexibility.  The Agency believes that because
baseline emissions are variable, emissions after the elimination of all
significant contribution and interference with maintenance are also
variable and thus it is appropriate to take this variability into
account.

As discussed in preamble section IV.F, the EPA proposes to use two
variability limits:  First, a “1-year” limit, based on the
year-to-year variability in emissions relative to the proposed budgets. 
Second, a “3-year” limit, based on the variability in a three
(consecutive) year average relative to the proposed budgets.  The EPA
determined 1-year variability limits that would apply to a state’s
emissions annually (or seasonally, for the ozone season) and 3-year
variability limits that would apply to a state’s annual (or ozone
season) emissions on a 3-year rolling average basis.  Preamble section
IV.F discusses EPA’s rationale for implementing 1-year and 3-year
variability limits.  Section IV.F also describes EPA’s proposed
approach to calculating the proposed 1- and 3-year limits and an
alternative calculation approach.  This TSD describes these approaches
in more detail.

Preamble section V.D describes the proposed remedy and two alternative
remedies.  In that section, EPA describes how the remedies would use
variability limits in the implementation of assurance provisions
designed to ensure that the necessary emissions reductions occur within
each covered state.  As discussed in preamble section V.D, the EPA
proposes to apply assurance provisions and variability limits starting
in 2014 and, as further discussed in the preamble, is also taking
comment on whether to apply them starting in 2012.

Preamble section IV.F presents proposed 1- and 3-year variability limits
for each state and alternative 1- and 3-year limits calculated using the
alternative approach.  Table IV.F-1 in the preamble presents proposed
and alternative 1- and 3-year variability limits on SO2 emissions for
each state.  Table IV.F-2 presents proposed and alternative 1- and
3-year variability limits on annual NOX emissions.  Table IV.F-3
presents proposed and alternative 1- and 3-year variability limits on
ozone season emissions.

For the alternative where the limits would apply starting in 2012
instead of 2014, Table IV.F-4 in the preamble presents proposed and
alternative 1- and 2-year variability limits.  Preamble section IV.F
explains that, for this alternative, EPA considered both 3-year average
and 2-year average variability limits and determined the 2-year limits
are preferable.

Preamble section IV.F describes EPA’s proposed approach to determine
1-year variability limits.  As discussed in the preamble, the approach
would determine 1-year limits based on the expected annual (or ozone
season) variation in power sector emissions derived from historical
variation in annual (or ozone season) power sector heat input in
combination with projected controlled emissions rates.  The preamble
discusses two approaches to determine 1-year limits based on expected
variation in power sector emissions.  Section 2 in this TSD, titled
“Estimating year-to-year variability in annual emissions”, describes
in greater detail the method EPA used to estimate expected variation in
year-to-year power sector emissions and the proposed and alternative
approaches to determine 1-year limits based on that expected variation.

As discussed in preamble section IV.F, after determining 1-year
variability limits, EPA used statistical methods to estimate multi-year
(3-year and 2-year) average variability limits for covered states based
on each state’s 1-year variability.  Section 3 in this TSD, titled
“Estimating variability over a multi-year time period”, describes
the approach in greater detail than that provided in the preamble.  As
discussed in the preamble and in section 3 in this TSD, the average
variability of a multi-year-average is the average variability of a
single year divided by the square root of the number of years in the
multi-year average.  Thus, the variability of a 3-year average is equal
to the annual variability divided by the square root of three.

As discussed above and in preamble section IV.F, for the alternative
where the limits would apply starting in 2012 instead of 2014, the EPA
determined 1- and 2-year variability limits.  For this alternative EPA
also considered 3-year variability limits instead of 2-year limits. 
Section 3 in this TSD compares the 3- and 2-year limits and discusses
why EPA determined that the 2-year limits are preferable.  Like the
3-year average variability discussed above, the variability of a 2-year
average is equal to the annual variability divided by the square root of
two (see section 3, below).

Section 4 in this TSD presents the results of an analysis using the air
quality assessment tool (AQAT).

2. Estimating year-to-year variability in emissions.    

This section describes the method that the Agency used to estimate the
year-to-year  (“1-year”) variability in annual SO2, annual NOX and
ozone season NOX emissions.  The method uses variation in heat input as
a proxy for emissions.  For an electric generating unit (EGU) fleet
equipped with a constant set of control technologies and consistently
using specific fuel types, the variability in heat input would be
directly related to variability in emissions.  This section in the TSD
provides information on:

The historical data set EPA established on a state-by-state basis of
yearly heat input values applicable to each of the pollutants regulated
in the Transport Rule (annual SO2, annual NOx, and ozone season NOx).
These historical heat input values are used to estimate the inherent
variability in emissions due to power system operation.

The method EPA developed (as well as an alternative method) to estimate
the year-to-year variability in the heat input values.  The year-to-year
variability in heat input is estimated on a state-by-state basis.

The approach EPA used to link heat input variability with projected
pollutant emission rates to estimate the inherent variation in pollutant
emissions.

The method EPA developed to identify a single set of variability
parameters that could be applied to all states in the program.

(a) Establishing a historical data set for use in estimating inherent
variability in emissions.

 The objective of this section is to describe the inherent year-to-year
(1-year) variability in emissions by characterizing the year-to-year
variance in total annual heat input for each state in the Transport
Rule.  EPA is concerned with variation in total emissions from
year-to-year (or the variation in total emissions from one ozone season
to the next), not variation from day to day or month to month within a
given year (or ozone season).  Thus, EPA used total yearly heat input
values equaling the sum of heat input from all units operating in each
state during a particular year.

EPA estimated the expected variation in power sector emissions for a
yearly time period based on the “standard deviation” of yearly power
sector heat input (HI) assessed over a 7-year time frame (2002 through
2008).  As described in section IV.F.1.a of the preamble, EPA chose to
examine historical variability in heat input rather than emissions
because emissions have changed over time due to controllable factors
such as fuel switching and installing new emissions controls.   EPA is
interested in describing inherent variation in emissions due to factors
such as variation in power demand, timing of maintenance activities, and
unexpected shutdown of units.  These factors are strongly correlated
with heat input and variation in electric generation. 

EPA chose the time period 2002 through 2008 for the analysis of
variation in heat input because it represents a time period where there
was substantial reporting of heat input and emissions data across many
states and EGU source types for units in the Acid Rain Program (ARP). 
When the analysis began in 2009, the last complete year of data
available was for 2008.  A starting year of 2002 was selected for this
analysis because in prior years (2000 and 2001) there had been large,
uneven changes in annual heat input from fossil units for some states
due, in part, to increased electricity demand and changes in composition
of the power sector fleet. (Note, for instance, that all EGUs in the US
affected by the ARP came under Phase 2 of the program starting in 2000).
 Consequently, incorporating data from years prior to 2002 in the
analysis would lead to the inclusion of a single year where there is
large change in heat input that is not representative of typical
year-to-year variability, thereby leading to overestimates of
variability.  Since the objective is to estimate the inherent
variability in heat input during time-periods of relative constancy in
the fleet composition, including time-periods where there is uneven
growth in electrical demand in the analysis would skew the variability
estimates.  Consequently, EPA chose 2002 for the start of the analysis
to minimize that effect while still balancing the need to estimate the
variance in heat input over as long a time period as possible.  

For each year of the 7-year time period, EPA estimated total power
sector heat input on a state-by-state basis using the sum of historical
heat input for all units within each of three general categories of
EGUs:  (1) coal-fired units; (2) combined cycle turbines; and (3) a
combination of oil- and gas-fired boilers and simple-cycle combustion
turbines.  Total annual heat input values (in million mmbtu) for the
coal-fired EGU source category and for the total of all EGU source
categories can be found in Table 1 and Table 2, respectively. 
Ozone-season heat input values for the total of all three EGU categories
can be found in Table 3.  For each state, EPA assessed the inherent
year-to-year variability in annual and ozone season NOx emissions using
the total yearly heat inputs summed across all three general EGU source
types (Tables 2 and 3, respectively), while the Agency assessed
year-to-year variability in annual SO2 emissions using the yearly heat
inputs from just the coal-fired EGUs (Table 1). 

EGUs built prior to 2002 that reported heat input and emissions data
between 2002 and 2008 and new EGUs were included for this analysis. 
EGUs that were built prior to 2002 but that did not start reporting
until after 2002 were excluded from the assessment. 

Table 1: Heat Input* (million mmbtu) from Coal-Fired Units for Each
Year.

State	2002	2003	2004	2005	2006	2007	2008	Average HI	Max. Difference from
Average

AL	779.0	822.6	784.5	828.3	813.0	818.8	768.9	802.1	26.1

CT	18.1	28.3	30.9	30.2	31.5	26.1	30.5	27.9	3.5

DE	42.6	45.4	50.8	52.1	50.1	58.3	53.7	50.4	7.8

FL	700.6	735.9	659.3	697.0	711.4	707.3	660.8	696.0	39.9

GA	781.2	787.4	808.8	871.6	872.5	910.6	861.1	841.9	68.7

IA	393.1	389.6	386.7	384.5	378.9	415.9	431.2	397.1	34.1

IL	930.9	961.9	1,027.1	1,002.1	1,002.1	1,030.5	1,032.9	998.2	34.7

IN	1,231.4	1,236.3	1,266.0	1,281.2	1,269.9	1,256.9	1,243.6	1,255.0	26.1

KS	427.5	427.5	411.9	405.2	377.6	405.4	373.5	404.1	23.4

KY	946.5	931.1	938.1	970.3	987.0	977.9	978.5	961.3	25.7

LA	120.9	122.1	128.1	134.6	128.7	116.1	118.5	124.1	10.4

MA	108.9	103.9	101.7	115.2	107.2	114.8	101.7	107.6	7.5

MD	276.9	278.0	269.9	273.7	274.7	277.1	256.0	272.3	5.7

MI	700.1	706.9	709.7	721.1	699.7	736.0	714.4	712.5	23.4

MN	389.2	411.0	386.7	380.1	370.3	362.9	349.8	378.6	32.5

MO	724.5	769.6	770.8	783.6	783.5	762.0	732.8	761.0	22.6

NC	693.3	689.1	696.5	719.7	699.1	726.4	713.2	705.3	21.1

NE	227.0	238.3	231.2	238.3	234.0	222.4	233.8	232.1	6.2

NJ	80.0	75.3	79.6	92.2	81.9	77.8	62.8	78.5	13.7

NY	234.1	235.9	226.7	213.4	211.2	211.0	181.0	216.2	19.8

OH	1,290.2	1,316.3	1,247.3	1,315.1	1,294.0	1,319.7	1,290.0	1,296.1	23.6

PA	1,026.9	1,050.1	1,078.6	1,090.2	1,090.5	1,105.2	1,065.1	1,072.4	32.8

SC	381.5	384.7	401.4	407.7	410.4	422.6	417.1	403.7	19.0

TN	624.7	574.9	589.4	586.3	606.9	617.2	578.0	596.8	27.9

VA	340.5	312.5	296.7	298.4	288.0	298.9	264.2	299.9	40.6

WI	457.1	475.9	482.9	473.2	451.2	447.5	450.8	462.6	20.3

WV	898.8	898.5	851.0	861.2	852.4	883.3	851.6	871.0	27.8

*Source:  EPA, March 2010. All relevant ARP units in the Transport Rule
region.  These data are available at http://www.epa.gov/airmarkets/
through Data and Maps.Table 2: Total Heat Input* (million mmbtu) from
All Three EGU Categories for Units Greater Than or Equal to 25 MW for
Each Year.

State	2002	2003	2004	2005	2006	2007	2008	Average HI	Max. Difference from
Average

AL	901.0	916.0	904.6	936.6	964.1	1,000.0	938.9	937.3	62.7

CT	69.9	67.5	87.5	91.2	101.1	93.9	84.9	85.2	16.0

DE	52.6	53.3	59.1	60.9	55.6	66.4	60.5	58.3	8.0

FL	1,134.5	1,197.8	1,198.8	1,283.1	1,361.0	1,398.3	1,360.2	1,276.2	122.0

GA	841.9	822.4	853.4	948.0	968.5	1,033.0	958.2	917.9	115.1

IA	393.2	390.3	391.2	401.3	396.3	438.0	446.5	408.1	38.4

IL	976.1	982.7	1,046.2	1,053.8	1,038.2	1,086.2	1,059.9	1,034.7	51.5

IN	1,256.8	1,257.5	1,283.6	1,315.1	1,300.4	1,301.2	1,282.2	1,285.3	29.8

KS	429.2	429.4	412.5	406.8	381.2	410.0	378.6	406.8	22.6

KY	962.4	935.1	942.4	988.3	999.0	997.2	987.8	973.2	25.8

LA	193.6	266.6	312.7	332.2	336.2	322.5	326.2	298.6	37.6

MA	159.8	221.4	221.5	244.4	262.4	276.4	242.9	232.7	43.7

MD	288.2	290.6	280.8	287.8	282.6	286.0	263.7	282.8	7.8

MI	726.9	725.1	727.3	760.5	730.7	776.5	743.8	741.5	34.9

MN	399.0	423.1	396.4	401.8	391.7	394.6	371.6	396.9	26.2

MO	752.1	788.7	792.2	813.6	812.5	800.6	770.3	790.0	23.6

NC	718.8	706.3	718.4	747.2	727.1	767.7	749.7	733.6	34.1

NE	228.4	240.5	233.5	245.3	240.7	231.9	240.3	237.2	8.1

NJ	139.3	132.9	151.5	154.9	158.8	172.7	172.8	154.7	18.1

NY	288.3	297.3	309.9	322.9	380.6	393.1	366.6	337.0	56.1

OH	1,309.6	1,330.3	1,258.3	1,340.0	1,313.6	1,354.5	1,310.4	1,316.7	37.9

PA	1,060.2	1,090.0	1,155.1	1,172.4	1,185.4	1,240.9	1,200.0	1,157.7	83.2

SC	418.2	402.7	442.1	454.9	461.5	474.0	464.4	445.4	28.7

TN	635.7	580.5	592.1	592.6	613.2	623.8	582.3	602.9	32.8

VA	356.6	333.6	329.5	339.5	324.0	354.8	312.3	335.7	20.8

WI	475.3	494.1	498.4	528.5	492.3	497.4	490.0	496.6	31.9

WV	900.5	899.9	852.0	863.3	855.8	887.1	853.3	873.1	27.3

*Source:  EPA, March 2010. All relevant ARP units in the Transport Rule
region.  These data are available at http://www.epa.gov/airmarkets/
through Data and Maps.

Table 3: Total Heat Input* (million mmbtu) from All Three EGU
Categories for Units Greater Than or Equal to 25 MW for Ozone Season.

State	2002	2003	2004	2005	2006	2007	2008	Average HI	Max. Difference from
Average

AL	427.4	428.8	423.6	427.2	456.9	469.9	432.5	438.1	31.9

AR	124.0	152.5	149.2	142.0	173.6	170.4	150.5	151.7	21.8

CT	33.3	28.7	43.5	42.8	46.3	44.4	38.4	39.6	6.7

DE	27.9	24.2	25.4	29.0	23.6	30.3	24.9	26.5	3.8

FL	548.6	554.9	563.3	615.1	639.0	656.3	646.0	603.3	53.0

GA	403.8	382.9	407.1	461.8	469.5	489.6	449.9	437.8	51.8

IL	454.9	430.2	449.6	478.0	460.8	474.6	452.6	457.2	20.8

IN	557.2	545.4	540.7	578.4	566.4	559.8	550.5	556.9	21.5

KS	183.2	180.9	176.8	180.6	172.2	177.1	160.7	175.9	7.3

KY	433.3	400.1	401.0	434.4	435.3	436.1	417.8	422.6	13.6

LA	94.7	121.7	144.9	149.9	161.8	150.7	152.2	139.4	22.3

MD	134.8	118.8	121.8	131.5	123.3	126.8	117.1	124.9	9.9

MI	331.0	312.1	304.6	339.9	328.5	344.5	321.1	326.0	18.6

MS	117.8	116.2	129.5	137.3	147.7	155.9	143.2	135.4	20.6

NC	332.9	308.5	314.0	341.5	340.4	351.5	344.8	333.4	18.1

NJ	68.7	59.9	73.9	76.1	79.9	86.1	86.6	75.9	10.8

NY	126.1	121.4	136.3	152.5	174.7	170.5	158.9	148.6	26.1

OH	583.7	562.8	538.4	574.4	562.0	596.7	553.3	567.3	29.3

OK	204.5	223.4	220.0	248.4	244.6	241.1	247.9	232.8	15.6

PA	465.4	468.7	494.6	529.3	530.6	550.8	515.5	507.8	42.9

SC	203.1	178.4	204.8	214.1	215.8	222.7	222.0	208.7	14.0

TN	291.2	244.3	256.2	270.1	273.2	278.9	261.2	267.9	23.3

TX	1,031.9	1,089.9	1,170.0	1,229.1	1,255.4	1,262.8	1,252.7	1,184.6	78.2

VA	160.0	146.9	150.9	154.3	151.5	167.9	144.5	153.7	14.2

WV	379.5	384.5	361.0	375.4	373.1	390.3	366.9	375.8	14.5

*Source:  EPA, March 2010; All relevant ARP units in the Transport Rule
region.  These data are available at http://www.epa.gov/airmarkets/
through Data and Maps.



(b) Estimating the state-by-state variability in heat input using
historical heat input data sets.

This subsection describes the method used by EPA to estimate the
variability in heat input for each state in the Transport Rule region. 
EPA assessed the year-to-year variability over the 7-year time period of
the yearly total heat input values (or ozone season heat input values)
using the “standard deviation” while accounting for overall growth
or decline in heat input over that time period.  This method is
described in detail in this and the following subsections and was
selected for three reasons: First, it accounted for growth or decline in
heat input over time.  Second, a statistical approach (i.e., using the
standard deviation) is less sensitive to data anomalies present in
finite data sets.  Third, a statistical approach also provides options
for different levels of variability (i.e., different confidence levels).
An alternative method was also examined, where the variability was
defined as the difference between the maximum yearly value and the
average heat input values over the time period.  

Both the preferred and alternative methods were applied on a
state-by-state basis.  The majority of this section focuses on the
analysis for the preferred method using the standard deviation, while
accounting for growth.  Where appropriate, differences between the
preferred and alternative method are described.

In the preferred method, for each state, it was important to account for
trends (growth or decline) in heat input over the 7-year time period. 
After accounting for growth in heat input over time, the year-to-year
variation in heat input was assessed, as the differences between the
actual yearly heat input values and the “yearly” average heat input
values estimated according to the trend.  For each state, to account for
trends, a simple least-squares linear regression equation was fit to the
heat input data as a function of time.  This process fits a straight
line to the data points using an equation of the form (y = mx+b).  In
this equation, “y” is the estimated heat input (million mmbtu) for a
particular year “x”, m is the slope of the line (with units of
million mmbtu/year), and b is the “y-intercept” (the heat input
value when the line is extrapolated to x = 0).  The value of r2
describes how well the data fit that line.  Large r2 values are
important for states where there was substantial growth or decline in
heat input over the time period (i.e., for states that have slopes for
the regression line that are substantially different than zero).  For
states with relatively constant heat input values, low r2 values are
strongly correlated with small values for the slope of the regression
line.  For these states, low r2 values are indicators of large
year-to-year variability (relative to small amounts of growth over
time).  For example, for Delaware (Table 6), the slope of the linear
regression is small, and the r2 value is also very small.  The
conclusion is that for Delaware and other states with similar slopes and
r2 values, the year-to-year variation could likely have been adequately
characterized without using the regression equation (which would account
for growth in heat input over time).  

For each state, the slopes of the regression equations, y-intercepts,
and r2 values, as well as the heat input estimated using the regression
equations, can be found in Tables 4, 5, and 6.  Using the regression
equation for each state, yearly heat input values were estimated for
each state for each year (between 2002 and 2008) in Tables 4, 5, and 6. 
Year-to-year variation was assessed for each year or for each ozone
season as the difference between the actual heat input (from Tables 1-3)
and the heat input estimated using the regression equation (Tables 4-6).



Table 4: Heat Input (million mmbtu) from Coal-Fired Units for Each Year
Estimated Using the Regression Equation.

State	Slope of Linear Regression (million mmbtu/year)	Intercept	r2 Value
2002	2003	2004	2005	2006	2007	2008

AL	-0.34	1,492	0.00	803.2	802.8	802.5	802.1	801.8	801.5	801.1

CT	1.19	-2,365	0.30	24.4	25.6	26.7	27.9	29.1	30.3	31.5

DE	2.08	-4,126	0.75	44.2	46.3	48.3	50.4	52.5	54.6	56.7

FL	-4.44	9,600	0.12	709.4	704.9	700.5	696.0	691.6	687.2	682.7

GA	19.64	-38,527	0.74	783.0	802.6	822.3	841.9	861.5	881.2	900.8

IA	5.69	-11,002	0.41	380.1	385.8	391.4	397.1	402.8	408.5	414.2

IL	14.94	-28,960	0.70	953.4	968.3	983.3	998.2	1,013.2	1,028.1	1,043.0

IN	2.92	-4,592	0.12	1,246.3	1,249.2	1,252.1	1,255.0	1,257.9	1,260.9
1,263.8

KS	-8.59	17,633	0.74	429.9	421.3	412.7	404.1	395.5	386.9	378.3

KY	8.52	-16,113	0.68	935.8	944.3	952.8	961.3	969.9	978.4	986.9

LA	-0.67	1,468	0.05	126.2	125.5	124.8	124.1	123.5	122.8	122.1

MA	0.20	-290	0.01	107.0	107.2	107.4	107.6	107.8	108.0	108.2

MD	-2.13	4,545	0.36	278.7	276.6	274.5	272.3	270.2	268.1	265.9

MI	3.26	-5,825	0.30	702.8	706.0	709.3	712.5	715.8	719.1	722.3

MN	-8.25	16,915	0.80	403.3	395.1	386.8	378.6	370.3	362.1	353.8

MO	0.80	-836	0.01	758.6	759.4	760.2	761.0	761.8	762.6	763.4

NC	4.90	-9,111	0.54	690.6	695.5	700.4	705.3	710.2	715.1	720.0

NE	-0.32	869	0.01	233.1	232.8	232.5	232.1	231.8	231.5	231.2

NJ	-1.58	3,241	0.15	83.2	81.7	80.1	78.5	76.9	75.4	73.8

NY	-8.01	16,282	0.85	240.2	232.2	224.2	216.2	208.2	200.2	192.2

OH	1.88	-2,480	0.03	1,290.4	1,292.3	1,294.2	1,296.1	1,298.0	1,299.8
1,301.7

PA	8.46	-15,880	0.46	1,047.0	1,055.5	1,063.9	1,072.4	1,080.8	1,089.3
1,097.7

SC	6.84	-13,312	0.90	383.1	390.0	396.8	403.7	410.5	417.3	424.2

TN	-1.36	3,314	0.02	600.8	599.5	598.1	596.8	595.4	594.1	592.7

VA	-9.45	19,252	0.77	328.2	318.8	309.3	299.9	290.4	281.0	271.5

WI	-3.84	8,161	0.34	474.2	470.3	466.5	462.6	458.8	455.0	451.1

WV	-6.09	13,082	0.36	889.2	883.2	877.1	871.0	864.9	858.8	852.7





Table 5: Total Heat Input (million mmbtu) from All Three EGU Categories
for Each Year Estimated Using the Regression Equation.

State	Slope of Linear Regression (million mmbtu/year)	Intercept	r2 Value
2002	2003	2004	2005	2006	2007	2008

AL	12.18	-23,490	0.55	900.8	913.0	925.1	937.3	949.5	961.7	973.9

CT	3.98	-7,895	0.48	73.2	77.2	81.2	85.2	89.1	93.1	97.1

DE	1.65	-3,256	0.54	53.4	55.0	56.7	58.3	60.0	61.6	63.3

FL	44.30	-87,543	0.89	1,143.3	1,187.6	1,231.9	1,276.2	1,320.5	1,364.8
1,409.1

GA	31.61	-62,466	0.75	823.1	854.7	886.3	917.9	949.5	981.1	1,012.8

IA	9.30	-18,233	0.72	380.2	389.5	398.8	408.1	417.4	426.7	436.0

IL	16.09	-31,231	0.73	986.5	1,002.5	1,018.6	1,034.7	1,050.8	1,066.9
1,083.0

IN	6.44	-11,635	0.39	1,265.9	1,272.4	1,278.8	1,285.3	1,291.7	1,298.2
1,304.6

KS	-7.92	16,295	0.70	430.6	422.7	414.8	406.8	398.9	391.0	383.1

KY	9.17	-17,418	0.56	945.7	954.8	964.0	973.2	982.3	991.5	1,000.7

LA	19.04	-37,877	0.63	241.4	260.5	279.5	298.6	317.6	336.6	355.7

MA	14.29	-28,415	0.67	189.8	204.1	218.4	232.7	247.0	261.3	275.6

MD	-2.89	6,080	0.47	291.5	288.6	285.7	282.8	279.9	277.0	274.1

MI	5.60	-10,480	0.37	724.7	730.3	735.9	741.5	747.1	752.7	758.3

MN	-5.14	10,702	0.53	412.3	407.2	402.0	396.9	391.7	386.6	381.5

MO	3.53	-6,280	0.12	779.4	782.9	786.5	790.0	793.5	797.0	800.6

NC	8.00	-15,302	0.63	709.6	717.6	725.6	733.6	741.6	749.6	757.6

NE	0.92	-1,612	0.11	234.4	235.4	236.3	237.2	238.1	239.1	240.0

NJ	6.69	-13,250	0.90	134.6	141.3	148.0	154.7	161.4	168.1	174.8

NY	17.76	-35,264	0.82	283.7	301.4	319.2	337.0	354.7	372.5	390.2

OH	3.79	-6,281	0.07	1,305.3	1,309.1	1,312.9	1,316.7	1,320.5	1,324.3
1,328.1

PA	26.84	-52,648	0.85	1,077.2	1,104.0	1,130.9	1,157.7	1,184.5	1,211.4
1,238.2

SC	10.75	-21,104	0.79	413.2	423.9	434.6	445.4	456.1	466.9	477.6

TN	-1.88	4,376	0.04	608.5	606.6	604.8	602.9	601.0	599.1	597.2

VA	-3.42	7,203	0.21	346.0	342.6	339.2	335.7	332.3	328.9	325.5

WI	1.60	-2,711	0.05	491.8	493.4	495.0	496.6	498.2	499.8	501.4

WV	-5.83	12,568	0.33	890.6	884.8	879.0	873.1	867.3	861.5	855.6



Table 6: Heat Input (million mmbtu) from All Three EGU Categories for
Ozone Season Estimated Using the Regression Equation.

State	Slope of Linear Regression (million mmbtu/year)	Intercept	r2 Value
2002	2003	2004	2005	2006	2007	2008

AL	4.68	-8,936	0.32	424.0	428.7	433.4	438.1	442.7	447.4	452.1

AR	4.99	-9,850	0.41	136.8	141.8	146.8	151.7	156.7	161.7	166.7

CT	1.77	-3,507	0.35	34.3	36.1	37.8	39.6	41.4	43.2	44.9

DE	0.04	-54	0.00	26.4	26.4	26.4	26.5	26.5	26.6	26.6

FL	20.38	-40,256	0.90	542.2	562.5	582.9	603.3	623.7	644.1	664.4

GA	14.80	-29,231	0.64	393.4	408.2	423.0	437.8	452.6	467.4	482.2

IL	3.33	-6,214	0.20	447.3	450.6	453.9	457.2	460.6	463.9	467.2

IN	1.23	-1,911	0.04	553.2	554.4	555.7	556.9	558.1	559.4	560.6

KS	-2.84	5,880	0.65	184.5	181.6	178.8	175.9	173.1	170.3	167.4

KY	2.15	-3,883	0.08	416.1	418.3	420.4	422.6	424.7	426.9	429.0

LA	8.84	-17,590	0.67	112.9	121.7	130.6	139.4	148.3	157.1	166.0

MD	-1.27	2,671	0.18	128.7	127.4	126.1	124.9	123.6	122.3	121.1

MI	2.10	-3,891	0.10	319.7	321.8	323.9	326.0	328.1	330.2	332.3

MS	6.21	-12,325	0.80	116.7	122.9	129.1	135.4	141.6	147.8	154.0

NC	5.29	-10,265	0.50	317.5	322.8	328.1	333.4	338.7	343.9	349.2

NJ	4.01	-7,956	0.82	63.9	67.9	71.9	75.9	79.9	83.9	87.9

NY	8.39	-16,679	0.74	123.4	131.8	140.2	148.6	157.0	165.4	173.8

OH	0.01	552	0.00	567.3	567.3	567.3	567.3	567.3	567.3	567.3

OK	6.80	-13,402	0.75	212.4	219.2	226.0	232.8	239.6	246.4	253.2

PA	12.52	-24,588	0.69	470.3	482.8	495.3	507.8	520.4	532.9	545.4

SC	5.58	-10,974	0.62	192.0	197.5	203.1	208.7	214.3	219.8	225.4

TN	-0.12	517	0.00	268.2	268.1	268.0	267.9	267.7	267.6	267.5

TX	39.05	-77,113	0.85	1,067.4	1,106.5	1,145.5	1,184.6	1,223.6	1,262.7
1,301.7

VA	-0.14	434	0.00	154.1	154.0	153.9	153.7	153.6	153.4	153.3

WV	-0.50	1,387	0.01	377.3	376.8	376.3	375.8	375.3	374.8	374.3



On a state-by-state basis, the difference between the actual heat input
and the estimated heat input using the regression equation was
calculated for each year.  The differences for SO2, annual NOx, and
ozone season NOx can be found in Tables 7, 8, and 9, respectively.  Some
of these yearly differences are positive, while others are negative. 
Assessing the differences between actual and estimated heat input across
all years for each state (and pollutant), a representative difference
was estimated using the “standard deviation”.  

The standard deviation is defined as the square root of the variance
(the sum of the square of the differences divided by the number of
samples minus one).  The state- and pollutant-specific representative
differences defined using the standard deviation were used for the
remaining steps in the variability analyses for the preferred method. 
In using the standard deviation as a representative difference, we
assume that:  (1) differences between the actual and modeled heat inputs
are “normally” distributed; (2) yearly mean values are independent;
and, (3) distribution of hourly values is the same (i.e., the
“within-year” variance is the same for each year). 

For each state, the standard deviation of the differences is a measure
of the year-to-year (1-year) variance in heat input.  In essence, it
suggests that, on average, 68% of all the year-specific heat input
values could be expected to be within one standard deviation from the
expected value (either higher or lower).  The standard deviation in the
heat input in million mmbtu rounded to three significant digits (as well
as a percentage of the average heat input value) can be found in Tables
7, 8, and 9.  

Table 7: Difference Between Heat Input (million mmbtu) Measured and
Estimated for Coal-Fired Units for Each Year Using the Regression
Equation.

State	2002	2003	2004	2005	2006	2007	2008	Standard Deviation of Heat
Input (million mmbtu)	Average Heat Input (2002-2008 (from Table 1)
Standard Deviation as a Fraction of Average HI	95% Confidence Level
Variability in HI (million mmbtu)	95% Confidence Level Variability in HI
(as a Fraction of Avg. HI)

AL	-24.1	19.7	-18.0	26.1	11.2	17.3	-32.2	24.0	802.1	0.03	47.0	0.059

CT	-6.2	2.7	4.1	2.3	2.3	-4.2	-1.0	3.9	27.9	0.141	7.7	0.276

DE	-1.6	-0.8	2.5	1.6	-2.4	3.7	-2.9	2.6	50.4	0.051	5.1	0.100

FL	-8.8	31.0	-41.2	1.0	19.8	20.1	-21.9	25.9	696.0	0.037	50.7	0.073

GA	-1.8	-15.2	-13.5	29.7	11.0	29.4	-39.7	25.4	841.9	0.03	49.7	0.059

IA	13.0	3.9	-4.8	-12.7	-23.9	7.4	17.0	14.6	397.1	0.037	28.7	0.073

IL	-22.5	-6.4	43.8	3.8	-11.0	2.4	-10.1	21.3	998.2	0.021	41.7	0.041

IN	-14.9	-12.9	13.9	26.1	11.9	-4.0	-20.2	17.5	1,255.0	0.014	34.3	0.027

KS	-2.4	6.3	-0.8	1.1	-17.9	18.4	-4.8	11.0	404.1	0.027	21.6	0.053

KY	10.7	-13.2	-14.8	9.0	17.2	-0.5	-8.4	12.6	961.3	0.013	24.7	0.025

LA	-5.2	-3.4	3.3	10.4	5.2	-6.7	-3.6	6.4	124.1	0.051	12.5	0.100

MA	1.9	-3.3	-5.7	7.5	-0.6	6.7	-6.5	5.7	107.6	0.053	11.1	0.104

MD	-1.8	1.4	-4.6	1.4	4.5	9.0	-10.0	6.2	272.3	0.023	12.1	0.045

MI	-2.7	0.8	0.5	8.5	-16.1	16.9	-7.9	10.7	712.5	0.015	21.0	0.029

MN	-14.1	16.0	-0.2	1.6	0.0	0.8	-4.0	8.9	378.6	0.023	17.4	0.045

MO	-34.0	10.2	10.7	22.6	21.7	-0.6	-30.6	23.4	761.0	0.031	45.9	0.061

NC	2.7	-6.5	-4.0	14.3	-11.1	11.3	-6.8	9.7	705.3	0.014	19.1	0.027

NE	-6.1	5.6	-1.3	6.2	2.2	-9.1	2.6	5.8	232.1	0.025	11.4	0.049

NJ	-3.3	-6.4	-0.5	13.7	5.0	2.4	-11.0	8.1	78.5	0.103	15.8	0.202

NY	-6.2	3.7	2.5	-2.8	3.1	10.8	-11.1	7.3	216.2	0.034	14.2	0.067

OH	-0.2	24.0	-46.9	19.0	-4.0	19.8	-11.8	24.8	1,296.1	0.019	48.5	0.037

PA	-20.1	-5.4	14.7	17.8	9.7	15.9	-32.7	19.9	1,072.4	0.019	39.1	0.037

SC	-1.6	-5.3	4.6	4.1	-0.1	5.3	-7.1	4.9	403.7	0.012	9.7	0.024

TN	23.8	-24.6	-8.7	-10.4	11.5	23.1	-14.7	19.3	596.8	0.032	37.8	0.063

VA	12.3	-6.3	-12.7	-1.5	-2.4	17.9	-7.3	11.1	299.9	0.037	21.7	0.073

WI	-17.1	5.6	16.4	10.5	-7.7	-7.5	-0.3	11.7	462.6	0.025	22.9	0.049

WV	9.5	15.4	-26.1	-9.7	-12.5	24.5	-1.1	17.6	871.0	0.02	34.5	0.039



Table 8: Difference Between Total Heat Input (million mmbtu) Measured
and Estimated from All Three EGU Categories for Each Year Using the
Regression Equation.

State	2002	2003	2004	2005	2006	2007	2008	Standard Deviation of Heat
Input (million mmbtu)	Average Heat Input (2002-2008 (from Table 2)
Standard Deviation as a Fraction of Average HI	95% Confidence Level
Variability in HI (million mmbtu)	95% Confidence Level Variability in HI
(as a Fraction of Avg. HI)

AL	0.2	3.0	-20.5	-0.7	14.6	38.3	-35.0	23.6	937.3	0.025	46.2	0.049

CT	-3.3	-9.7	6.3	6.1	12.0	0.8	-12.2	8.9	85.2	0.105	17.5	0.206

DE	-0.7	-1.8	2.4	2.6	-4.4	4.7	-2.8	3.3	58.3	0.056	6.5	0.110

FL	-8.8	10.1	-33.1	6.9	40.5	33.4	-48.9	32.8	1,276.2	0.026	64.4	0.051

GA	18.8	-32.3	-32.9	30.1	18.9	51.8	-54.5	39.6	917.9	0.043	77.6	0.084

IA	13.0	0.8	-7.7	-6.8	-21.2	11.3	10.5	12.6	408.1	0.031	24.8	0.061

IL	-10.4	-19.8	27.5	19.1	-12.6	19.3	-23.1	21.2	1,034.7	0.020	41.5	0.039

IN	-9.2	-14.8	4.8	29.8	8.7	3.0	-22.3	17.3	1,285.3	0.013	33.9	0.025

KS	-1.4	6.7	-2.3	-0.1	-17.7	19.1	-4.4	11.2	406.8	0.027	21.9	0.053

KY	16.8	-19.7	-21.6	15.1	16.6	5.7	-12.9	17.5	973.2	0.018	34.3	0.035

LA	-47.8	6.1	33.2	33.6	18.6	-14.1	-29.5	31.5	298.6	0.106	61.8	0.208

MA	-30.0	17.3	3.1	11.7	15.5	15.1	-32.7	21.9	232.7	0.094	42.9	0.184

MD	-3.3	2.0	-4.9	5.0	2.7	8.9	-10.4	6.6	282.8	0.023	12.9	0.045

MI	2.2	-5.3	-8.7	19.0	-16.5	23.7	-14.5	15.9	741.5	0.021	31.1	0.041

MN	-13.3	15.9	-5.6	4.9	-0.1	8.0	-9.8	10.4	396.9	0.026	20.4	0.051

MO	-27.3	5.7	5.8	23.6	19.0	3.6	-30.3	21.1	790.0	0.027	41.3	0.053

NC	9.2	-11.3	-7.2	13.6	-14.5	18.1	-7.9	13.2	733.6	0.018	25.9	0.035

NE	-6.1	5.1	-2.7	8.1	2.5	-7.2	0.3	5.7	237.2	0.024	11.2	0.047

NJ	4.7	-8.4	3.5	0.2	-2.6	4.6	-1.9	4.8	154.7	0.031	9.4	0.061

NY	4.6	-4.1	-9.3	-14.1	25.9	20.6	-23.6	18.1	337.0	0.054	35.5	0.106

OH	4.3	21.2	-54.6	23.3	-6.9	30.3	-17.6	29.6	1,316.7	0.023	58.1	0.045

PA	-17.0	-14.0	24.2	14.7	0.8	29.5	-38.3	24.6	1,157.7	0.021	48.2	0.041

SC	5.0	-21.2	7.4	9.5	5.3	7.2	-13.2	12.1	445.4	0.027	23.7	0.053

TN	27.2	-26.1	-12.7	-10.3	12.2	24.7	-15.0	21.1	602.9	0.035	41.3	0.069

VA	10.5	-9.0	-9.7	3.8	-8.3	25.9	-13.2	14.2	335.7	0.042	27.9	0.082

WI	-16.5	0.7	3.4	31.9	-5.9	-2.3	-11.3	15.7	496.6	0.032	30.7	0.063

WV	9.8	15.1	-26.9	-9.8	-11.5	25.7	-2.4	18.0	873.1	0.021	35.2	0.041



Table 9: Difference Between Total Heat Input Measured and Estimated
(million mmbtu) from All Three EGU Categories for Ozone Season Using the
Regression Equation.

State	2002	2003	2004	2005	2006	2007	2008	Standard Deviation of Heat
Input (million mmbtu)	Average Heat Input (2002-2008 (from Table 3)
Standard Deviation as a Fraction of Average HI	95% Confidence Level
Variability in HI (million mmbtu)	95% Confidence Level Variability in HI
(as a Fraction of Avg. HI)

AL	3.4	0.1	-9.8	-10.8	14.1	22.5	-19.5	14.8	438.1	0.034	29.0	0.067

AR	-12.7	10.7	2.5	-9.7	16.8	8.6	-16.2	12.9	151.7	0.085	25.3	0.167

CT	-1.0	-7.4	5.6	3.2	4.9	1.2	-6.6	5.3	39.6	0.133	10.3	0.261

DE	1.6	-2.2	-1.0	2.5	-3.0	3.7	-1.7	2.6	26.5	0.097	5.0	0.190

FL	6.4	-7.6	-19.6	11.8	15.3	12.2	-18.5	15.0	603.3	0.025	29.4	0.049

GA	10.4	-25.3	-15.9	24.0	16.9	22.2	-32.3	23.8	437.8	0.054	46.7	0.106

IL	7.6	-20.4	-4.3	20.8	0.3	10.7	-14.6	14.5	457.2	0.032	28.3	0.063

IN	4.0	-9.1	-15.0	21.5	8.3	0.5	-10.1	12.6	556.9	0.023	24.7	0.045

KS	-1.3	-0.7	-2.0	4.7	-0.9	6.9	-6.7	4.5	175.9	0.026	8.8	0.051

KY	17.1	-18.2	-19.4	11.8	10.5	9.3	-11.2	15.6	422.6	0.037	30.6	0.073

LA	-18.2	-0.1	14.4	10.5	13.5	-6.4	-13.7	13.3	139.4	0.095	26.1	0.186

MD	6.1	-8.6	-4.4	6.6	-0.3	4.5	-4.0	5.9	124.9	0.047	11.6	0.092

MI	11.4	-9.7	-19.2	13.9	0.4	14.4	-11.2	13.7	326.0	0.042	26.8	0.082

MS	1.1	-6.8	0.3	1.9	6.1	8.1	-10.8	6.7	135.4	0.050	13.2	0.098

NC	15.3	-14.3	-14.0	8.2	1.7	7.5	-4.4	11.4	333.4	0.034	22.4	0.067

NJ	4.8	-7.9	2.1	0.2	0.0	2.2	-1.3	4.0	75.9	0.053	7.9	0.104

NY	2.6	-10.5	-3.9	3.9	17.7	5.1	-14.9	10.8	148.6	0.073	21.3	0.143

OH	16.4	-4.5	-28.9	7.1	-5.4	29.3	-14.0	19.4	567.3	0.034	38.0	0.067

OK	-8.0	4.1	-6.0	15.6	4.9	-5.3	-5.3	8.6	232.8	0.037	16.8	0.073

PA	-4.9	-14.1	-0.8	21.5	10.2	17.9	-29.9	18.3	507.8	0.036	35.8	0.071

SC	11.1	-19.1	1.7	5.5	1.5	2.8	-3.4	9.5	208.7	0.046	18.6	0.090

TN	22.9	-23.9	-11.8	2.2	5.5	11.3	-6.3	15.5	267.9	0.058	30.3	0.114

TX	-35.5	-16.5	24.5	44.5	31.8	0.1	-49.0	35.4	1,184.6	0.030	69.4	0.059

VA	5.8	-7.1	-3.0	0.6	-2.1	14.5	-8.8	8.0	153.7	0.052	15.7	0.102

WV	2.2	7.6	-15.3	-0.4	-2.2	15.5	-7.4	10.0	375.8	0.027	19.5	0.053



The standard deviation can also be used to estimate, on average, the
probability of larger variations in heat input (for example a difference
that we would expect less than 1% of the time).  Using the standard
deviation, “confidence levels” representing the variability
difference in heat input that could be expected at different
probabilities were found for each state (for each pollutant).  Several
different levels were examined (notably the 95th and 99th percent
confidence levels).  As an illustrative example, the two-tailed 95th
percent confidence level indicates that we could expect, on average,
that the total heat input for a particular year for a particular state
will be within 1.960 standard deviations of the variation from its mean
value 95 percent of the time.  For this analysis, we focus on results
for the two-tailed 95th percent confidence level for each state.  EPA
made the policy decision that, assessed across the large number of
states included in the program, the 95th percent two-tailed confidence
level was the appropriate confidence level.  EPA believes that using the
95th percent confidence level provides a high degree of confidence that
sources subject to the rule will be able to operate within the
constraints of the variability limits and without electric reliability
problems arising.  

When the alternative method for calculating variability (the maximum
difference over the time period) was applied, the results for many
states were comparable to the results using the 95th confidence level,
showing that the proposed variation was similar to the maximum measured
historical values. The major differences were for states that had growth
(or decline) in heat input over the time period.  The 95% upper
confidence level heat inputs and percentages (the 95% heat input
difference divided by the average heat input over the 2002-2008 time
period) can be found in Tables 7, 8, and 9.  The results from the
alternative method using the maximum difference in heat input from the
7-year average heat input can be found in Tables 1, 2, and 3.

For states where there were differences between the proposed and
alternative approaches, the difference was often a result of substantial
growth in heat input over the 7-year time period.  For example, as seen
in Table 1, for Virginia, the heat input in 2002 is 340.5 million mmbtu.
 By 2008, the heat input had decreased to 264.2 million mmbtu.  The
average heat input over this time period was 299.9 million mmbtu. 
Consequently, the maximum difference from the average was 40.6 million
mmbtu (Table 1).  Using the proposed approach (the 95% confidence level)
and accounting for decline in heat input over time using the regression
equation, the estimated variability in heat input is substantially
reduced (21.7 million mmbtu, as seen in Table 7).

(c) Estimation of the emission rates projected in 2014 used to estimate
variability in emissions from the year-to-year variability in historical
heat input. 

The final step in estimating state-by-state year-to-year (1-year)
variability in emissions is to convert the year-to-year variability in
heat input into variability in pollutant emissions.  This was done by
multiplying the estimated variability in heat input by a representative
pollutant emission rate.  For each state, the state-specific 95th
percent confidence level heat input variability value (converted to
mmbtu) was multiplied by a state-specific emission rate (tons of
pollutant per mmbtu).  The resulting value is the 95th percent
confidence level variability emission value (in tons of pollutant).  

State-specific modeled emission rates were used in the calculation.  The
rates were based on IPM emissions and heat inputs projected to occur in
2014 when levels of controls similar to that for the proposed remedy are
applied to each state.  Modeled emission rates were used, rather than
historical emission rates, because EPA wanted to estimate year-to-year
variability in emissions within the same time and conditions as those of
the proposed transport rule when new emissions controls could
potentially be in operation.

Using IPM estimates for 2014, the emission rates for annual SO2 were
calculated using modeled estimates of total heat input and total
emissions from coal-fired units, while annual and ozone season NOx were
calculated using modeled estimates of total heat input and total
emissions from all three EGU categories.  EPA derived the state-specific
emission rates from IPM projections parsed for 2014; the state-specific
emission rates for annual SO2, annual NOx and ozone season NOx are from
IPM runs that required controls at $1,600/ton for SO2, $500/ton for
annual NOX, and $500/ton for ozone season NOX, respectively.  Tables 10
through 14 list, for each state and the three EGU sectors defined above,
the modeled emissions, the modeled fuel usage (the sum of the heat
input), and the emission rates.  For each state, the state-specific 95th
percent confidence level heat input variability value was multiplied by
the state-specific emission rate.  The resulting 95% confidence level
emission variability value, in units of tons of emissions, can be found
in Tables 10, 12, and 14 for annual SO2, annual NOx, and ozone season
NOx, respectively (see the columns labeled “95% confidence level
variability value”).  The 95% confidence level emission variability
value was normalized by dividing by each state’s emissions, resulting
in a “coefficient of variation” value.  In this TSD, the value is
referred to as the “percentage” variability value.  

The 95% confidence level emission variability value can be compared to
the difference between the emissions for a particular year and the
emissions budget.  We expect the difference to be less than the 95%
emissions variability value, on average, 95 percent of the time. 
Consequently, the 95% emissions variability value represents a
year-to-year variability that is sufficient, most of the time, to
encompass the inherent variability in power sector generation.   That
is, sources within a state should not exceed the state’s budget if
variability is the only reason for emissions increases above the budget.

Subsequent sections in this TSD discuss how the 95% confidence level
variability values were used to construct the variability limits applied
in the proposed Transport Rule assurance provisions. 



Table 10.  State-by-State SO2 Emissions (thousand short tons), Fuel Use
(million mmbtu), and Emission Rate Estimated in IPM for Units Greater
Than or Equal to 25 MW. Also Shown are the Estimated 95th Percent
Confidence SO2 Tonnage and Percentage Values as well as the Alternative
Approach SO2 Tonnage and Percentage Values.

State	Annual SO2 Emissions From Coal-Fired Boilers (thousand tons)
Annual fuel use from coal-fired boilers (million mmbtu)	SO2 Emission
Rate (lbs/mmbtu)	95% Confidence Level Variability Value (tons)	95%
Confidence Level Variability Value (percentage)	Alternative Approach
Value (tons)	Alternative Approach Value (percentage)

AL	103.38	852.69	0.242	5,693	5.9%	3,170	3.3%

CT	2.72	42.05	0.129	498	27.6%	228	12.6%

DE	7.38	62.66	0.236	595	10.0%	923	15.6%

FL	80.73	1,084.86	0.149	3,775	7.3%	2,968	5.7%

GA	94.39	977.67	0.193	4,801	5.9%	6,631	8.2%

IL	159.15	1,140.91	0.279	5,816	4.1%	4,840	3.5%

IN	231.11	1,380.68	0.335	5,733	2.7%	4,375	2.1%

IA	87.27	466.65	0.374	5,363	7.3%	6,378	8.6%

KS	44.99	380.57	0.236	2,555	5.3%	2,771	5.8%

KY	158.93	1,083.19	0.293	3,628	2.5%	3,767	2.7%

LA	92.78	387.40	0.479	2,992	10.0%	2,500	8.4%

MD	42.67	436.37	0.196	1,186	4.5%	557	2.1%

MA	8.44	111.12	0.152	844	10.4%	572	7.0%

MI	196.34	770.31	0.510	5,359	2.9%	5,975	3.3%

MN	40.10	386.00	0.208	1,807	4.5%	3,371	8.6%

MO	185.27	823.10	0.450	10,337	6.1%	5,087	3.0%

NE	68.91	287.50	0.479	2,730	4.9%	1,489	2.7%

NJ	14.23	155.23	0.183	1,449	20.2%	1,255	17.4%

NY	45.11	261.98	0.344	2,450	6.7%	3,401	9.1%

NC	88.43	931.48	0.190	1,809	2.7%	2,001	3.0%

OH	200.28	1,433.48	0.279	6,781	3.7%	3,299	1.8%

PA	150.88	1,362.73	0.221	4,325	3.7%	3,636	3.1%

SC	73.44	487.02	0.302	1,457	2.4%	2,858	4.7%

TN	109.09	617.16	0.354	6,688	6.3%	4,927	4.7%

VA	60.60	407.67	0.297	3,222	7.3%	6,039	13.5%

WV	138.89	1,029.88	0.270	4,652	3.9%	3,748	3.2%

WI	78.48	495.03	0.317	3,630	4.9%	3,215	4.4%

Table 11.  Annual NOx Emissions (thousand short tons) and Fuel Use
(million mmbtu) Estimated in IPM, for Units Greater Than or Equal to 25
MW.

State	                              Annual NOx Emissions	               
                Annual Fuel Usage

	Coal-Fired Boilers (thousand tons) 	Combined Cycle Turbines (thousand
tons)	Simple Cycle CT (thousand tons)	Oil & Gas Boilers (thousand tons) 
All Units (thousand tons)	Coal-Fired Boilers (million mmbtu)	Combined
Cycle Turbines (million mmbtu)	Simple Cycle CT (million mmbtu)	Oil & Gas
Boilers (million mmbtu)	All Units (million mmbtu)

AL	60.805	0.888	0.030	 	61.724	856.091	122.084	1.449	 	979.624

CT	2.051	0.716	0.043	0.000	2.809	42.050	103.449	1.986	0.000	147.485

DE	3.725	0.584	0.017	0.000	4.327	60.215	26.258	0.685	0.000	87.158

FL	98.493	7.814	2.839	12.350	121.496	1,057.478	645.095	56.530	107.598
1,866.701

GA	44.781	0.437	0.128	0.000	45.346	984.344	57.183	6.240	0.000	1,047.766

IL	55.470	0.123	0.203	0.000	55.796	1,134.540	9.555	7.355	0.000	1,151.449

IN	111.365	0.058	0.047	0.000	111.470	1,392.974	7.880	1.393	0.000
1,402.248

IA	50.717	0.034	0.140

50.891	480.860	3.991	1.687

486.539

KS	37.399

0.000	0.000	37.399	377.821

0.000	0.000	377.821

KY	72.244

0.011

72.256	1,085.560

0.289

1,085.849

LA	34.268	0.754	0.090	1.197	36.310	384.835	75.389	3.188	11.737	475.149

MD	19.643	0.026	0.105	0.000	19.774	428.182	2.058	5.361	0.000	435.601

MA	5.584	1.171	0.019	0.000	6.774	111.117	204.806	0.742	0.000	316.666

MI	63.816	0.575	0.171	0.000	64.562	812.765	23.096	5.095	0.000	840.956

MN	32.037	0.069	0.075	0.000	32.181	373.950	10.424	3.089	0.000	387.463

MO	78.419	0.018	0.000

78.437	843.016	2.056	0.000

845.071

NE	32.967	0.010	0.151	0.000	33.128	298.044	1.494	1.887	0.000	301.425

NJ	10.622	0.952	0.517	0.000	12.091	155.481	112.882	7.173	0.000	275.536

NY	13.240	2.356	0.651	7.561	23.808	261.921	239.245	9.342	202.341	712.849

NC	59.417	0.112	0.145

59.674	914.612	3.404	6.215

924.231

OH	99.323	0.159	0.145

99.627	1,450.382	18.752	4.609

1,473.743

PA	112.489	0.657	0.074	0.000	113.220	1,353.586	105.537	2.365	0.000
1,461.488

SC	34.301	0.201	0.030	0.000	34.532	491.149	18.015	1.338	0.000	510.501

TN	28.272

0.403

28.675	616.230

6.243

622.473

VA	27.524	0.491	0.189	0.000	28.204	399.917	46.858	5.747	0.000	452.522

WV	54.084

0.000

54.084	1,010.968

0.000

1,010.968

WI	39.771	0.031	0.152	0.000	39.953	521.020	4.279	0.608	0.000	525.908



Table 12.  State-by-State Annual NOx Emission Rate Calculated Using the
IPM Estimates Presented in Table 11. Also Shown Are the Estimated 95th
Percent Confidence NOx Tonnage and Percentage Values as well as the
Alternative Approach NOx Tonnage and Percentage Values.

State	Annual NOx Emission Rate (lb/mmbtu)	95% Confidence Level
Variability Value  (tons)	95% Confidence Level Variability Value
(percentage)	Alternative Approach Value (tons)	Alternative Approach
Value  (percentage)

AL	0.126	2,912	4.9%	3,951	6.7%

CT	0.038	333	20.6%	305	18.8%

DE	0.099	321	11.0%	398	13.7%

FL	0.130	4,190	5.1%	7,941	9.6%

GA	0.087	3,359	8.4%	4,980	12.5%

IL	0.097	2,010	3.9%	2,496	5.0%

IN	0.159	2,699	2.5%	2,369	2.3%

IA	0.209	2,591	6.1%	4,015	9.4%

KS	0.198	2,166	5.3%	2,237	5.6%

KY	0.133	2,283	3.5%	1,716	2.6%

LA	0.153	4,723	20.8%	2,873	12.6%

MD	0.091	586	4.5%	353	2.8%

MA	0.043	918	18.4%	934	18.8%

MI	0.154	2,390	4.1%	2,682	4.7%

MN	0.166	1,691	5.1%	2,178	6.6%

MO	0.186	3,832	5.3%	2,191	3.0%

NE	0.220	1,226	4.7%	889	3.4%

NJ	0.088	412	6.1%	795	11.7%

NY	0.067	1,187	10.6%	1,874	16.7%

NC	0.129	1,673	3.5%	2,199	4.6%

OH	0.135	3,928	4.5%	2,559	2.9%

PA	0.155	3,732	4.1%	6,447	7.2%

SC	0.135	1,603	5.3%	1,938	6.4%

TN	0.092	1,904	6.9%	1,512	5.4%

VA	0.125	1,738	8.2%	1,297	6.2%

WV	0.107	1,886	4.1%	1,463	3.1%

WI	0.152	2,333	6.3%	2,426	6.4%

Table 13.  Ozone Season NOx Emissions (thousand short tons) and Fuel
Use (million mmbtu) Estimated in IPM for Units Greater Than or Equal to
25 MW.

State	                              Ozone Season NOx Emissions	         
                      Ozone Season Fuel Usage

	Coal-Fired Boilers (thousand tons)	Combined Cycle Turbines (thousand
tons)	Simple Cycle CT (thousand tons)	Oil & Gas Boilers (thousand tons)
All Units (thousand tons)	Coal-Fired Boilers (million mmbtu) 	Combined
Cycle Turbines (million mmbtu) 	Simple Cycle CT (million mmbtu) 	Oil &
Gas Boilers (million mmbtu) 	All Units (million mmbtu) 

AL	26.193	0.615	0.030	 	26.838	370.7	85.0	1.4	 	457.1

AR	10.730	1.021	0.036	0.000	11.787	169.1	60.5	1.0	0.0	230.6

CT	0.894	0.292	0.017	0.000	1.203	18.3	42.4	0.6	0.0	61.3

DE	1.564	0.255	0.016	0.000	1.835	25.3	11.4	0.6	0.0	37.3

FL	46.385	3.962	1.367	12.580	64.294	465.6	321.2	30.4	110.1	927.3

GA	19.583	0.344	0.128	0.000	20.055	431.9	43.0	6.2	0.0	481.2

IL	23.649	0.106	0.203	0.000	23.958	485.2	8.1	7.3	0.0	500.6

IN	47.093	0.049	0.047	0.000	47.188	589.4	6.2	1.4	0.0	597.1

KS	16.200

0.000	0.000	16.200	163.5

0.0	0.0	163.5

KY	29.843

0.011

29.855	450.2

0.3

450.5

LA	14.992	0.638	0.090	1.197	16.918	168.3	55.2	3.2	11.7	238.5

MD	8.217	0.016	0.095	0.000	8.328	178.3	1.1	4.0	0.0	183.4

MI	27.539	0.498	0.175	0.000	28.212	352.6	16.6	5.2	0.0	374.4

MS	7.651	0.167	0.049	0.000	7.866	67.5	26.4	2.5	0.0	96.4

NJ	4.522	0.537	0.446	0.000	5.506	65.2	56.3	5.4	0.0	126.9

NY	5.759	1.168	0.641	3.946	11.514	114.1	113.2	7.4	95.6	330.3

NC	25.468	0.084	0.132

25.684	390.7	2.4	5.3

398.4

OH	42.052	0.110	0.145

42.308	608.8	13.0	4.6

626.4

OK	19.566	1.895	0.165	2.378	24.003	257.2	75.4	2.0	16.0	350.7

PA	48.089	0.310	0.074	0.000	48.474	579.1	50.5	2.4	0.0	631.9

SC	14.729	0.164	0.030	0.000	14.922	211.6	14.8	1.3	0.0	227.8

TN	11.542

0.423

11.965	251.4

6.4

257.8

TX	56.728	6.237	0.645	4.824	68.434	1,022.3	375.3	14.2	118.7	1,530.5

VA	11.927	0.308	0.125	0.000	12.360	172.2	26.3	3.6	0.0	202.0

WV	24.235

0.000

24.235	429.9

0.0

429.9

WI	17.104	0.027	0.152	0.000	17.283	223.8	3.7	0.6	0.0	228.1

Table 14.  State-by-State Ozone Season NOx Emission Rates Calculated
Using the IPM Estimates Presented in Table 13. Also Shown Are the
Estimated 95th Percent Confidence NOx Tonnage and Percentage Values as
well as the Alternative Approach NOx Tonnage and Percentage Values for
Ozone Season.

State	Ozone Season NOx Emission Rate (lb/mmbtu)	95% Confidence Level
Variability Value (tons)	95% Confidence Level Variability Value
(percentage)	Alternative Approach Value (tons)	Alternative Approach
Value (percentage)

AL	0.117	1,702	6.7%	1,871	7.3%

AR	0.102	1,292	16.7%	1,116	14.4%

CT	0.039	203	26.1%	131	16.9%

DE	0.098	248	19.0%	187	14.3%

FL	0.139	2,037	4.9%	3,674	8.8%

GA	0.083	1,945	10.6%	2,159	11.8%

IL	0.096	1,355	6.3%	994	4.5%

IN	0.158	1,952	4.5%	1,697	3.9%

KS	0.198	873	5.1%	721	4.1%

KY	0.133	2,028	7.3%	900	3.2%

LA	0.142	1,849	18.6%	1,585	16.0%

MD	0.091	527	9.2%	451	8.0%

MI	0.151	2,016	8.2%	1,400	5.7%

MS	0.163	1,073	9.8%	1,679	15.2%

NJ	0.087	341	10.4%	466	14.2%

NY	0.070	741	14.3%	909	17.5%

NC	0.129	1,443	6.7%	1,166	5.4%

OH	0.135	2,569	6.7%	1,982	5.2%

OK	0.137	1,150	7.3%	1,068	6.7%

PA	0.153	2,747	7.1%	3,293	8.5%

SC	0.131	1,221	9.0%	915	6.7%

TN	0.093	1,408	11.4%	1,082	8.7%

TX	0.089	3,104	5.9%	3,498	6.6%

VA	0.122	960	10.2%	868	9.2%

WV	0.113	1,102	5.3%	816	3.8%

(d) Procedure for identifying a single set of variability parameters (a
tonnage and a percentage limit) to uniformly apply to all states in the
program.

From the state-by-state 1-year 95th confidence level tonnage and
percentage emission variability values in Tables 10, 12, and 14, EPA
identified a single set of variability limits (i.e., a tonnage and a
percentage) for each pollutant (i.e., SO2, annual NOX, and ozone season
NOX) to apply to all covered states.  For this analysis, EPA assumes
that, on average, each state is meeting its proposed budget, but is also
subject to inherent year-to-year variability in electric power system
operations that could lead to short-term increases (or decreases) in
emissions up to the variability limits.  In identifying a single set of
percentage and tonnage variability limits to apply across all states,
EPA assumes that for some future year, each state would experience
conditions such that it would need to utilize (but not exceed) its
1-year 95% confidence level emissions variability tonnage value .  Thus,
EPA has identified a set of minimum variability limits (a tonnage and a
percentage) that when compared against the 95% confidence level tonnage
and percentage values for each state, all states could (and would be
required to) meet at least one of the two limits (the tonnage or the
percentage).

Preamble section IV.F provides EPA’s rationale for identifying a
single tonnage and percentage combination to apply to all covered
states.  Preamble section IV.F also provides EPA’s rationale for
identifying both a tonnage limit and a percentage limit.  As explained
in section IV.F, the effect of identifying both a tonnage and percentage
is to ensure that each state is allowed adequate inherent variability
while minimizing the total amount of emissions allowed; this approach
addresses the difficulty that smaller states with fewer units could face
if only percentages were used to set the limits (or that larger states
with many units could face if only tonnages were used).  Most of the
details of EPA’s approach to determine the 1-year variability limits
are provided in section IV.F in the preamble.  This TSD presents some
additional information.

To identify the 1-year tonnage and percentage limits, EPA looked at a
wide range of percentage and tonnage combinations, and chose for further
investigation combinations that provided states sufficient variability
(based on historic variability) while minimizing the total allowed
emissions.  For annual SO2 and ozone season NOx, the tonnage limit
criteria were examined in 300 ton increments, while the percentage
criteria were applied in 2% increments.  For annual NOX, the tonnage
limit criteria were applied in 500 ton increments, while the percentage
criteria were applied in 5% increments.  All combinations of these
criteria were considered.  For each pairing of percentage and tonnage
limits, the first step is to determine, based on the estimates of each
state’s historical variability values, whether any of the states would
exceed both of the limits (i.e., both the tonnage and the percentage). 
If more than four states were not able to meet one of the limits, this
number of states was recorded (this can be seen in grayed squares in
Tables 15, 16, and 17 for annual SO2, annual NOx and ozone season NOx,
respectively).  If four or fewer states were not able to meet one of the
limits, the state abbreviations are listed in Tables 15, 16, and 17 (the
cells are also grayed).  If one of the percentage and tonnage limits
could be met by each state (and thereby the combination of limits would
be applicable to all states), both limits were applied to the state, and
the larger of the two limits was chosen.  This would then be the
emission variability limit used for that state.  The combinations of
limits that were applicable (i.e., where the 95% confidence level
variability values for all states was below at least one of the two
limits) are shown in Tables 15, 16, and 17 by the white shading of
cells, while combinations of limits where the 95% confidence level
variability value for at least one state exceeds both limits are shown
by the gray shading of cells.

The difference between the emission variability limit and the
state-specific 95% confidence level emissions variability value (from
Tables 10, 12, and 14) for each state was calculated, and, for each
percentage and tonnage pairing, the total difference for all states was
summed.  The total differences for all states for each combination of
tonnage and percentage limits can be seen in white cells in Tables 15,
16, and 17 for annual SO2, annual NOx and ozone season NOx,
respectively.  In these tables, white cells represent possible
combinations of percentage and tonnage limits that could successfully be
met by all states included in the proposed rule. The optimal solution
(marked in yellow in the table) was one where:  (a) all states included
in the proposed rule are able to meet at least one of the criteria, and
(b) the sum of the total emissions was minimized.

Table 15. The Effects of Various Combinations of the Proposed Upper 95%
Confidence Level Tonnage and Percentage Variability Limits on Annual EGU
SO2 Emissions (See Notes Below).

	Percentage Limit

Tonnage Limit	2	4	6	8	10	12	14	16	18	20

1,000	24	16	8	LA, NJ, 	NJ, 	NJ, 	NJ, 	NJ, 	NJ, 	NJ, 

1,300	23	15	8	LA, NJ, 	NJ, 	NJ, 	NJ, 	NJ, 	NJ, 	NJ, 

1,600	21	14	7	LA, 	121,876	165,006	208,135	251,265	294,395	337,549

1,900	19	13	7	LA, 	123,076	166,206	209,335	252,465	295,595	338,725

2,200	19	13	7	LA, 	124,276	167,406	210,535	253,665	296,795	339,925

2,500	18	12	6	LA, 	125,476	168,606	211,735	254,865	297,995	341,125

2,800	16	10	6	LA, 	126,845	169,806	212,935	256,065	299,195	342,325

3,100	15	9	5	86,605	128,452	171,006	214,135	257,265	300,395	343,525

3,400	14	8	FL, IA, MO, TN, 	88,897	130,252	172,449	215,335	258,465
301,595	344,725

3,700	12	7	FL, IA, MO, TN, 	91,444	132,074	174,056	216,552	259,665
302,795	345,925

4,000	11	6	IA, MO, TN, 	94,281	134,174	175,856	218,052	260,865	303,995
347,125

4,300	11	6	IA, MO, TN, 	97,417	136,565	177,656	219,661	262,156	305,195
348,325

4,600	10	6	IA, MO, TN, 	100,861	139,123	179,644	221,461	263,656	306,395
349,525

4,900	8	5	IA, MO, TN, 	104,461	141,895	181,833	223,261	265,266	307,760
350,725

5,200	8	5	IA, MO, TN, 	108,304	144,895	184,233	225,113	267,066	309,260
351,925

5,500	6	AL, IL, MO, TN, 	MO, TN, 	112,429	148,190	186,802	227,213
268,866	310,871	353,363

5,800	IL, MO, OH, TN, 	IL, MO, TN, 	MO, TN, 	116,629	151,719	189,509
229,500	270,666	312,671	354,863

6,100	MO, OH, TN, 	MO, TN, 	MO, TN, 	121,187	155,319	192,509	231,900
272,683	314,471	356,476

6,400	MO, OH, TN, 	MO, TN, 	MO, TN, 	125,987	159,123	195,662	234,481
274,783	316,271	358,276

6,700	MO, OH, 	MO, 	MO, 	130,956	163,129	198,977	237,181	277,168	318,152
360,076

7,000	MO, 	MO, 	MO, 	136,056	167,329	202,577	240,122	279,568	320,252
361,876

7,300	MO, 	MO, 	MO, 	141,156	171,529	206,177	243,135	282,160	322,435
363,676

7,600	MO, 	MO, 	MO, 	146,256	176,127	209,942	246,435	284,860	324,835
365,722

7,900	MO, 	MO, 	MO, 	151,511	180,927	213,842	249,836	287,736	327,235
367,822

8,200	MO, 	MO, 	MO, 	156,911	185,763	218,029	253,436	290,736	329,839
370,103

8,500	MO, 	MO, 	MO, 	162,311	190,863	222,229	257,036	293,907	332,539
372,503

8,800	MO, 	MO, 	MO, 	167,980	195,963	226,429	260,762	297,207	335,349
374,903

9,100	MO, 	MO, 	MO, 	173,680	201,063	231,067	264,662	300,694	338,349
377,518

9,400	MO, 	MO, 	MO, 	179,489	206,163	235,867	268,730	304,294	341,380
380,218

9,700	MO, 	MO, 	MO, 	185,696	211,282	240,667	272,930	307,894	344,680
382,963

10,000	MO, 	MO, 	MO, 	191,996	216,682	245,670	277,130	311,581	347,980
385,963

10,300	MO, 	MO, 	MO, 	198,296	222,082	250,770	281,330	315,481	351,553
388,963

10,600	186,027	186,027	188,625	204,596	227,482	255,870	286,007	319,430
355,153	392,152

10,900	194,127	194,127	195,825	210,896	233,118	260,970	290,807	323,630
358,753	395,452

11,200	202,227	202,227	203,564	217,196	238,818	266,070	295,607	327,830
362,400	398,811

11,500	210,327	210,327	211,364	223,801	244,518	271,170	300,477	332,030
366,300	402,411

11,800	218,427	218,427	219,164	230,701	250,404	276,452	305,577	336,230
370,200	406,011

12,100	226,527	226,527	226,964	237,601	256,638	281,852	310,677	340,947
374,330	409,611

12,400	234,627	234,627	234,764	244,501	262,938	287,252	315,777	345,747
378,530	413,219

12,700	242,727	242,727	242,727	251,401	269,238	292,652	320,877	350,547
382,730	417,119

13,000	250,827	250,827	250,827	258,301	275,538	298,256	325,977	355,347
386,930	421,019

13,300	258,927	258,927	258,927	265,201	281,838	303,956	331,077	360,384
391,130	425,031

13,600	267,027	267,027	267,027	272,101	288,138	309,656	336,223	365,484
395,887	429,231



*Numbers in white cells represent the sum (across all states) of the
differences in emissions between the state-specific 95% confidence level
variability values and the state-specific variability limit (in tons);
the variability limit selected for each state is the larger of either
the percentage limit or the tonnage limit.

** If the cell is grey, it means that there is a state or several states
whose state-specific 95% confidence level variability values exceed both
the tonnage and percentage limits.  The cell either lists the states
that could exceed the limits, or lists the number of states.

*** If the cell is yellow, this is the combination tonnage and
percentage limits that minimize the sum of the differences between the
state-specific 95% confidence level variability values and the
state-specific variability limit (in tons).Table 16. The Effects of
Various Combinations of the Proposed Upper 95% Confidence Level Tonnage
and Percentage Variability Limits on Annual EGU NOx Emissions (See Notes
Below).

	Percentage Limit

Tonnage Limit 	5	10	15	20	25	30	35	40	45	50

1,000	12	LA, NY, 	LA, 	LA, 	224,069	280,207	336,368	392,653	448,937
505,222

1,500	11	LA, 	LA, 	LA, 	225,323	281,212	337,346	393,484	449,622	505,761

2,000	7	LA, 	LA, 	LA, 	227,129	282,712	338,601	394,490	450,622	506,761

2,500	5	LA, 	LA, 	LA, 	229,129	284,680	340,230	395,990	451,879	507,768

3,000	FL, GA, LA, MO, 	LA, 	LA, 	LA, 	231,326	286,680	342,230	397,781
453,379	509,268

3,500	FL, LA, MO, 	LA, 	LA, 	LA, 	234,079	288,816	344,230	399,781
455,331	510,881

4,000	FL, LA, 	LA, 	LA, 	LA, 	237,079	291,420	346,306	401,781	457,331
512,881

4,500	LA, 	LA, 	LA, 	LA, 	240,079	294,420	348,806	403,796	459,331
514,881

5,000	76,672	98,561	140,230	189,970	243,079	297,420	351,760	406,296
461,331	516,881

5,500	89,877	107,933	146,230	194,258	246,302	300,420	354,760	409,101
463,786	518,881

6,000	103,377	117,683	152,673	199,208	250,118	303,420	357,760	412,101
466,441	521,276

6,500	116,877	128,211	160,138	204,649	254,118	306,587	360,760	415,101
469,441	523,781

7,000	130,377	139,211	168,266	210,514	258,666	310,267	363,760	418,101
472,441	526,781

7,500	143,877	150,471	177,153	216,576	263,666	314,267	366,872	421,101
475,441	529,781

8,000	157,377	161,971	186,461	223,105	269,092	318,448	370,415	424,101
478,441	532,781

8,500	170,877	173,748	195,961	230,420	274,798	323,123	374,415	427,157
481,441	535,781

9,000	184,377	186,036	205,836	238,391	280,798	328,123	378,415	430,657
484,441	538,781

9,500	197,877	198,968	216,336	246,745	287,001	333,535	382,793	434,563
487,442	541,781

10,000	211,377	211,968	227,128	255,745	293,537	339,083	387,581	438,563
490,942	544,781

10,500	224,877	224,968	238,128	264,989	300,806	345,083	392,581	442,638
494,712	547,781

11,000	238,377	238,377	249,267	274,489	308,645	351,083	397,978	447,138
498,712	551,228

11,500	251,877	251,877	260,767	283,989	316,692	357,426	403,478	452,039
502,712	554,860

12,000	265,377	265,377	272,267	293,989	325,337	363,969	409,367	457,039
506,983	558,860

12,500	278,877	278,877	283,934	304,489	334,337	371,192	415,367	462,421
511,496	562,860

13,000	292,377	292,377	295,934	315,045	343,517	378,898	421,367	467,921
516,496	566,860

13,500	305,877	305,877	308,365	326,045	353,017	386,898	427,851	473,651
521,496	571,328

14,000	319,377	319,377	321,263	337,045	362,517	395,155	434,401	479,651
526,864	575,954

14,500	332,877	332,877	334,263	348,064	372,017	403,929	441,578	485,651
532,364	580,954

15,000	346,377	346,377	347,263	359,564	382,142	412,929	449,152	491,776
537,936	585,954

15,500	359,877	359,877	360,263	371,064	392,642	422,045	457,152	498,276
543,936	591,307

16,000	373,377	373,377	373,377	382,564	403,142	431,545	465,152	504,833
549,936	596,807

16,500	386,877	386,877	386,877	394,120	413,961	441,045	473,618	511,964
555,936	602,307

17,000	400,377	400,377	400,377	406,120	424,961	450,545	482,521	519,464
562,200	608,220

17,500	413,877	413,877	413,877	418,194	435,961	460,127	491,521	527,405
568,700	614,220

18,000	427,377	427,377	427,377	430,694	446,961	470,296	500,574	535,405
575,265	620,220

18,500	440,877	440,877	440,877	443,558	458,361	480,796	510,074	543,581
582,349	626,220

19,000	454,377	454,377	454,377	456,558	469,861	491,296	519,574	552,113
589,849	632,625

19,500	467,877	467,877	467,877	469,558	481,361	501,878	529,074	561,113
597,659	639,125

20,000	481,377	481,377	481,377	482,558	492,861	512,878	538,574	570,113
605,659	645,697

20,500	494,877	494,877	494,877	495,558	504,361	523,878	548,252	579,113
613,659	652,735

21,000	508,377	508,377	508,377	508,558	516,305	534,878	558,449	588,602
622,044	660,235

21,500	521,877	521,877	521,877	521,877	528,305	545,878	568,949	598,102
630,706	667,912

22,000	535,377	535,377	535,377	535,377	540,523	557,158	579,449	607,602
639,706	675,912



*Numbers in white cells represent the sum (across all states) of the
differences in emissions between the state-specific 95% confidence level
variability values and the state-specific variability limit (in tons);
the variability limit selected for each state is the larger of either
the percentage limit or the tonnage limit.

** If the cell is grey, it means that there is a state or several states
whose state-specific 95% confidence level variability values exceed both
the tonnage and percentage limits.  The cell either lists the states
that could exceed the limits, or lists the number of states.

*** If the cell is yellow, this is the combination tonnage and
percentage limits that minimize the sum of the differences between the
state-specific 95% confidence level variability values and the
state-specific variability limit (in tons).

Table 17. The Effects of Various Combinations of the Proposed Upper 95%
Confidence Level Tonnage and Percentage Variability Limits on Ozone
Season EGU NOx Emissions (See Notes Below).

	Percentage Limit

Tonnage Limit	0	2	4	6	8	10	12	14	16	18

800	20	20	20	15	8	5	AR, LA, 	AR, LA, 	AR, LA, 	LA, 

1,100	17	17	17	13	6	AR, GA, LA, TN, 	AR, LA, 	AR, LA, 	AR, LA, 	LA, 

1,400	12	12	12	9	GA, LA, MI, TN, 	GA, LA, TN, 	LA, 	LA, 	LA, 	LA, 

1,700	10	10	10	7	GA, LA, MI, 	GA, LA, 	LA, 	LA, 	LA, 	LA, 

2,000	6	6	6	KY, MI, OH, PA, 	MI, 	27,836	35,240	43,315	51,653	60,326

2,300	OH, PA, TX, 	OH, PA, TX, 	OH, PA, TX, 	OH, PA, 	27,326	32,529
39,180	46,694	54,787	63,096

2,600	PA, TX, 	PA, TX, 	PA, TX, 	PA, 	33,326	37,824	43,491	50,524	58,149
66,258

2,900	TX, 	TX, 	TX, 	36,882	39,326	43,628	48,591	54,724	61,909	69,603

3,200	44,115	44,115	44,115	44,115	45,528	49,628	54,086	59,554	66,069
73,320

3,500	51,615	51,615	51,615	51,615	52,338	55,628	59,931	64,725	70,517
77,413

3,800	59,115	59,115	59,115	59,115	59,538	61,628	65,931	70,348	75,617
81,666

4,100	66,615	66,615	66,615	66,615	66,738	68,081	71,931	76,234	80,910
86,580

4,400	74,115	74,115	74,115	74,115	74,115	74,993	77,931	82,234	86,610
91,680

4,700	81,615	81,615	81,615	81,615	81,615	82,193	84,034	88,234	92,536
97,172

5,000	89,115	89,115	89,115	89,115	89,115	89,393	90,645	94,234	98,536
102,872

5,300	96,615	96,615	96,615	96,615	96,615	96,615	97,649	100,234	104,536
108,839

5,600	104,115	104,115	104,115	104,115	104,115	104,115	104,849	106,588
110,536	114,839

5,900	111,615	111,615	111,615	111,615	111,615	111,615	112,049	113,267
116,536	120,839

6,200	119,115	119,115	119,115	119,115	119,115	119,115	119,249	120,305
122,569	126,839

6,500	126,615	126,615	126,615	126,615	126,615	126,615	126,615	127,505
129,141	132,839

6,800	134,115	134,115	134,115	134,115	134,115	134,115	134,115	134,705
135,889	138,839

7,100	141,615	141,615	141,615	141,615	141,615	141,615	141,615	141,905
142,960	145,094

7,400	149,115	149,115	149,115	149,115	149,115	149,115	149,115	149,115
150,160	151,694

7,700	156,615	156,615	156,615	156,615	156,615	156,615	156,615	156,615
157,360	158,510

8,000	164,115	164,115	164,115	164,115	164,115	164,115	164,115	164,115
164,560	165,616

8,300	171,615	171,615	171,615	171,615	171,615	171,615	171,615	171,615
171,760	172,816

8,600	179,115	179,115	179,115	179,115	179,115	179,115	179,115	179,115
179,115	180,016

8,900	186,615	186,615	186,615	186,615	186,615	186,615	186,615	186,615
186,615	187,216

9,200	194,115	194,115	194,115	194,115	194,115	194,115	194,115	194,115
194,115	194,416

9,500	201,615	201,615	201,615	201,615	201,615	201,615	201,615	201,615
201,615	201,616

9,800	209,115	209,115	209,115	209,115	209,115	209,115	209,115	209,115
209,115	209,115

10,100	216,615	216,615	216,615	216,615	216,615	216,615	216,615	216,615
216,615	216,615

10,400	224,115	224,115	224,115	224,115	224,115	224,115	224,115	224,115
224,115	224,115

10,700	231,615	231,615	231,615	231,615	231,615	231,615	231,615	231,615
231,615	231,615

11,000	239,115	239,115	239,115	239,115	239,115	239,115	239,115	239,115
239,115	239,115

11,300	246,615	246,615	246,615	246,615	246,615	246,615	246,615	246,615
246,615	246,615

11,600	254,115	254,115	254,115	254,115	254,115	254,115	254,115	254,115
254,115	254,115

11,900	261,615	261,615	261,615	261,615	261,615	261,615	261,615	261,615
261,615	261,615

12,200	269,115	269,115	269,115	269,115	269,115	269,115	269,115	269,115
269,115	269,115

12,500	276,615	276,615	276,615	276,615	276,615	276,615	276,615	276,615
276,615	276,615

12,800	284,115	284,115	284,115	284,115	284,115	284,115	284,115	284,115
284,115	284,115

13,100	291,615	291,615	291,615	291,615	291,615	291,615	291,615	291,615
291,615	291,615

13,400	299,115	299,115	299,115	299,115	299,115	299,115	299,115	299,115
299,115	299,115



*Numbers in white cells represent the sum (across all states) of the
differences in emissions between the state-specific 95% confidence level
variability values and the state-specific variability limit (in tons);
the variability limit selected for each state is the larger of either
the percentage limit or the tonnage limit.

** If the cell is grey, it means that there is a state or several states
whose state-specific 95% confidence level variability values exceed both
the tonnage and percentage limits.  The cell either lists the states
that could exceed the limits, or lists the number of states.

*** If the cell is yellow, this is the combination tonnage and
percentage limits that minimize the sum of the differences between the
state-specific 95% confidence level variability values and the
state-specific variability limit (in tons).

From the tables above, EPA determined that a number of tonnage and
percentage combinations resulted in fairly similar total emissions.  For
each pollutant, a percentage limit of 10 percent coupled with a tonnage
limit specific to the pollutant was a combination amongst those that
would result in the lowest total emissions, so EPA chose 10 percent. 
The tonnage limits for each pollutant, as shown below, were established.
The procedure used to identify these tonnage and percentage limits
ensures that every state is able to meet at least one of the limits,
while minimizing total EGU emissions.  The resulting 1-year tonnage and
percentage limits (which are also presented in the preamble) are as
follows:

SO2 – 1,700 tons or 10 percent of state’s budget

Annual NOX – 5,000 tons or 10 percent of state’s budget

Ozone season NOX – 2,100 tons or 10 percent of state’s budget

As described in the preamble, after determining for each pollutant a
1-year tonnage and 1-year percentage limit, EPA assigned each state one
of these values – either the tonnage limit or the percentage limit,
whichever was greater for that state.  In other words, for SO2, every
state has a 1-year variability limit of 1,700 tons or 10 percent of the
state’s SO2 budget, whichever is greater.  For annual NOX, every state
has 1-year variability limit of 5,000 tons or 10 percent of the
state’s annual NOX budget, whichever is greater.  And, for ozone
season NOX, every state has a 1-year variability limit of 2,100 tons or
10 percent of the state’s ozone season NOX budget, whichever is
greater.

For example, Connecticut’s annual SO2 budget is 3,059 tons, and 10
percent (the percentage limit for SO2) of 3,059 tons is 306 tons. 
Because 1,700 tons (the tonnage limit for SO2) is greater than 306 tons,
Connecticut’s 1-year SO2 variability limit is 1,700 tons.  The 1-year
variability limits for SO2, annual NOX, and ozone season NOX emissions
for each covered state that result from this approach are presented in
section IV.F in the preamble.

As discussed in preamble section IV.F, the EPA also requests comment on
an alternative calculation method for determining 1-year variability
limits.  The alternative method would use the results of the proposed
method but add a “ceiling percentage” equal to the maximum 95%
confidence level percentage of variability among all covered states as
observed in the historic heat input data described previously.  The
percentage variability limits for all states are shown in Tables 10, 12,
and 14 for annual SO2, annual NOx and ozone season NOx, respectively. 
The alternative 1-year variability limits resulting from this
calculation method are presented in preamble section IV.F.  EPA
explained in the preamble its rationale for considering this alternative
calculation method.

  The ceiling percentages for each pollutant, based on the historic
data, are as follows:

SO2 – 28 percent of state’s budget (equal to the value for
Connecticut)

Annual NOX – 21 percent of state’s budget (equal to the value for
Louisiana)

Ozone season NOX – 27 percent of state’s budget (equal to the value
for Connecticut)

Under this alternative calculation method, for SO2 emissions, a
state’s 1-year variability limit would be 1,700 tons as long as 1,700
tons is between 10 and 28 percent of the state’s SO2 emissions budget.
 If 1,700 tons is greater than 28 percent of the state’s SO2 budget,
then the state’s 1-year variability limit is set at 28 percent of the
state’s SO2 budget.  If 1,700 tons is less than 10 percent of the
state’s SO2 budget, then the state’s 1-year variability limit is set
at 10 percent of the state’s SO2 budget.  The alternative calculation
method would be applied to determine annual and ozone season NOX 1-year
variability limits in the same manner as for the SO2 limits.

Tables 18, 19, and 20 demonstrate application of this alternative method
to determining 1-year variability limits for SO2 emissions, annual NOX
emissions, and ozone season NOX emissions, respectively.  In Table 18,
the first column lists the state and the second column lists the
state’s 2014 SO2 emissions budget.  The third column lists, for each
state, 10 percent (the percentage limit for SO2) of the state’s SO2
budget.  The fourth column lists the tonnage limit for SO2 emissions,
which is 1,700 tons for all states, as discussed above.  The fifth
column lists, for each state, 28 percent (the percentage ceiling for
SO2) of the state’s SO2 budget.  And finally, the sixth column lists
the state’s alternative 1-year variability limit.  The columns in
Tables 19 and 20 follow the same pattern but for annual NOX and ozone
season NOX emissions, respectively.

Again, using Connecticut’s 1-year SO2 variability limit as an example,
Connecticut’s annual SO2 budget is 3,059 tons.  Ten percent (the
percentage limit for SO2) of 3,059 tons is 306 tons.  Twenty-eight
percent (the percentage ceiling for SO2) of 3,059 tons is 857 tons. 
Because 1,700 tons (the tonnage limit for SO2) is greater than 857 tons,
Connecticut’s 1-year SO2 variability limit is set at 857 tons, as
shown in Table 18.  In contrast, using the proposed approach would
result in a 1-year SO2 variability limit of 1,700 tons for Connecticut,
as discussed above and in preamble section V.F.

Table 18.  Application of Alternative Calculation Method for Determining
1-Year Variability Limits on SO2 Emissions for 2014 and Later.

State	2014

SO2 Annual Emissions Budget

(tons)	10 Percent of SO2 Budget

(tons)	SO2 Tonnage Limit

(tons)	28 Percent of SO2 Budget

(tons)	Alternative 1-Year Limit

(tons)

Alabama	161,871	16,187	1,700	45,324	16,187

Connecticut	3,059	306	1,700	857	857

Delaware	7,784	778	1,700	2,180	1,700

District of Columbia	337	34	1,700	94	94

Florida	161,739	16,174	1,700	45,287	16,174

Georgia	85,717	8,572	1,700	24,001	8,572

Illinois	151,530	15,153	1,700	42,428	15,153

Indiana	201,412	20,141	1,700	56,395	20,141

Iowa	86,088	8,609	1,700	24,105	8,609

Kansas	57,275	5,728	1,700	16,037	5,728

Kentucky	113,844	11,384	1,700	31,876	11,384

Louisiana	90,477	9,048	1,700	25,334	9,048

Maryland	39,665	3,967	1,700	11,106	3,967

Massachusetts	7,902	790	1,700	2,213	1,700

Michigan	155,675	15,568	1,700	43,589	15,568

Minnesota	47,101	4,710	1,700	13,188	4,710

Missouri	158,764	15,876	1,700	44,454	15,876

Nebraska	71,598	7,160	1,700	20,047	7,160

New Jersey	11,291	1,129	1,700	3,161	1,700

New York	42,041	4,204	1,700	11,771	4,204

North Carolina	81,859	8,186	1,700	22,921	8,186

Ohio	178,307	17,831	1,700	49,926	17,831

Pennsylvania	141,693	14,169	1,700	39,674	14,169

South Carolina	116,483	11,648	1,700	32,615	11,648

Tennessee	100,007	10,001	1,700	28,002	10,001

Virginia	40,785	4,079	1,700	11,420	4,079

West Virginia	119,016	11,902	1,700	33,324	11,902

Wisconsin	66,683	6,668	1,700	18,671	6,668

Table 19.  Application of Alternative Calculation Method for
Determining 1-Year Variability Limits on NOX Annual Emissions for 2014
and Later.

State	2014

NOX Annual Emissions Budget

(tons)	10 Percent of NOX Annual Budget

(tons)	NOX Annual Tonnage Limit

(tons)	21 Percent of NOX Annual Budget

(tons)	Alternative 1-Year Limit

(tons)

Alabama	69,169	6,917	5,000	14,525	6,917

Connecticut	2,775	278	5,000	583	583

Delaware	6,206	621	5,000	1,303	1,303

District of Columbia	170	17	5,000	36	36

Florida	120,001	12,000	5,000	25,200	12,000

Georgia	73,801	7,380	5,000	15,498	7,380

Illinois	56,040	5,604	5,000	11,768	5,604

Indiana	115,687	11,569	5,000	24,294	11,569

Iowa	46,068	4,607	5,000	9,674	5,000

Kansas	51,321	5,132	5,000	10,777	5,132

Kentucky	74,117	7,412	5,000	15,565	7,412

Louisiana	43,946	4,395	5,000	9,229	5,000

Maryland	17,044	1,704	5,000	3,579	3,579

Massachusetts	5,960	596	5,000	1,252	1,252

Michigan	64,932	6,493	5,000	13,636	6,493

Minnesota	41,322	4,132	5,000	8,678	5,000

Missouri	57,681	5,768	5,000	12,113	5,768

Nebraska	43,228	4,323	5,000	9,078	5,000

New Jersey	11,826	1,183	5,000	2,483	2,483

New York	23,341	2,334	5,000	4,902	4,902

North Carolina	51,800	5,180	5,000	10,878	5,180

Ohio	97,313	9,731	5,000	20,436	9,731

Pennsylvania	113,903	11,390	5,000	23,920	11,390

South Carolina	33,882	3,388	5,000	7,115	5,000

Tennessee	28,362	2,836	5,000	5,956	5,000

Virginia	29,581	2,958	5,000	6,212	5,000

West Virginia	51,990	5,199	5,000	10,918	5,199

Wisconsin	44,846	4,485	5,000	9,418	5,000



Table 20.  Application of Alternative Calculation Method for
Determining 1-Year Variability Limits on NOX Ozone Season Emissions for
2014 and Later.

State	2014

NOX Ozone Season Emissions Budget

(tons)	10 Percent of NOX Ozone Season Budget

(tons)	NOX Ozone Season Tonnage Limit

(tons)	27 Percent of NOX Ozone Season Budget

(tons)	Alternative 1-Year Limit

(tons)

Alabama	29,738	2,974	2,100	8,029	2,974

Arkansas	16,660	1,666	2,100	4,498	2,100

Connecticut	1,315	132	2,100	355	355

Delaware	2,450	245	2,100	662	662

District of Columbia	105	11	2,100	28	28

Florida	56,939	5,694	2,100	15,374	5,694

Georgia	32,144	3,214	2,100	8,679	3,214

Illinois	23,570	2,357	2,100	6,364	2,357

Indiana	49,987	4,999	2,100	13,496	4,999

Kansas	21,433	2,143	2,100	5,787	2,143

Kentucky	30,908	3,091	2,100	8,345	3,091

Louisiana	21,220	2,122	2,100	5,729	2,122

Maryland	7,232	723	2,100	1,953	1,953

Michigan	28,253	2,825	2,100	7,628	2,825

Mississippi	16,530	1,653	2,100	4,463	2,100

New Jersey	5,269	527	2,100	1,423	1,423

New York	11,090	1,109	2,100	2,994	2,100

North Carolina	23,539	2,354	2,100	6,356	2,354

Ohio	40,661	4,066	2,100	10,978	4,066

Oklahoma	37,087	3,709	2,100	10,013	3,709

Pennsylvania	48,271	4,827	2,100	13,033	4,827

South Carolina	15,222	1,522	2,100	4,110	2,100

Tennessee	11,575	1,158	2,100	3,125	2,100

Texas	75,574	7,557	2,100	20,405	7,557

Virginia	12,608	1,261	2,100	3,404	2,100

West Virginia	22,234	2,223	2,100	6,003	2,223



3. Estimating variability over a multi-year time period.  

As described in section 2, for each state, EPA estimated the inherent
year-to-year (1-year) variability in emissions at the 95% confidence
level using the historical year-to-year standard deviation in heat input
along with a modeled emission rate.  From these values EPA identified a
set of variability limits (a percentage and tonnage) and assigned a
particular proposed “1-year” variability tonnage limit to each state
for each year under the proposed Transport Rule.  As described in
section IV.F.1 of the preamble, for each state, EPA proposes to also
assess the difference between each state’s emissions budget and a
3-year rolling average of its emissions, comparing it against a
“3-year” variability limit.  This section of the TSD describes the
method EPA used to determine each state’s 3-year variability limit.

EPA used a standard statistical method to estimate the variability that
could be expected (on average) in a 3-year average of a state’s yearly
emissions, assuming that the emissions for each year are independent of
the following year.  For this analysis, we assume that, for each state,
the year-to-year variability in emissions is described using the 1-year
values in section 2 of this TSD.  

What is our expectation for how a 3-year average of emissions would
compare to the emissions for a single year?  We would expect that the
emissions for a single year could be higher or lower than the budget. 
What happens if we begin to average the emissions from multiple years? 
We would expect that we would average some years that are higher than
the budget with some years that are lower than budget.  The resulting
average would be relatively close to the budget.  Some variation from
the budget value is expected, though, when multiple 3-year time periods
are examined.  How variable could we expect the 3-year average to be
relative to the variability for a single year?  The answer to these
questions can be estimated with the standard deviation of the annual
emissions (i.e., the year-to-year variability) and the number of years
that are being averaged together (using some standard statistics
equations along with some basic assumptions about normality and
independence of the yearly emission values). 

The average variability of a multi-year-average is the average
variability of a single year divided by the square root of the number of
years in the multi-year average.  Thus, the variability of a 3-year
average is equal to the annual variability divided by the square root of
three.  EPA used this approach to determine 3-year variability limits
based on the 1-year limits.

Most general introductory statistics textbooks state the general
equation, and some describe a basic derivation.  Derivation of the
equation is found in those textbooks and the statistics literature. 
However, a rudimentary outline of some of the principles is described
here.  Usually, the problem is cast as combining the variances for
either three different random samples of a variable, or alternatively,
averaging three independent “variables” (where each year in the
three sequential years that are going to be averaged together would be a
different “variable”).  For this example, we will describe three
independent yearly “variables” of the emissions.  These yearly
variables are called Y1, Y2, and Y3.  Each of these yearly variables
could assume a particular emissions value for a particular year.  It is
assumed that the individual realized yearly values are uncorrelated and
independent from each other, that their average and variability can be
described with a normal distribution classified by a mean value (i.e.,
the budget value) and a standard deviation (i.e., the year-to-year
standard deviation).  For explanatory purposes, it will first be assumed
without loss of generality that the distributions of hourly emissions
values from year-to-year is identical (i.e., the “within-year”
variation is the same) and that all three years have the same mean value
and expected variation.  

The expected variance of the average of the three years of emissions can
be estimated.  If the years are assumed to be uncorrelated, and random
samples are drawn independently from each of the year-specific
distributions, the variance in the resulting estimate of the sample mean
can be written as follows:

 

where,

 ) = the variance of the mean of yearly emission values from each of
three independent years, Yi, where i =1 to 3; 

 = (1/3) * (X1 + X2 + X3) 

 = the year-specific variance of three independent years, Yi, where i =
1 to 3

 (where n, is the number of years in the average and is equal to 3).
This form of the equation is used by EPA in the proposed approach to
calculate the 3-year variance in the sample mean from the year-to-year
variance.

Applying these equations on a state-by-state basis using the
state-specific 1-year emission variability limits and the mean values
(the state emissions budgets), EPA calculated the 3-year variability
limits that are presented in section IV.F in the preamble.  These limits
are the 1-year limits divided by the square root of three.  The 3-year
limits can be derived from the state-specific standard deviations and
95% confidence level variability values.  The method for identifying a
set of limits is identical (where all the values have been divided by
the square root of 3).  The results are identical (except all values are
divided by the square root of three).  Consequently, EPA elected to
simply divide the proposed 1-year limits by the square root of three.

Similarly, EPA applied the equations above to define variability for the
2-year variability limits presented as an alternative in preamble
section IV.F.  These limits are the 1-year limits divided by the square
root of two (since we are interested in the variance of just two years,
rather than three).  As discussed in the preamble, the 2-year
variability limits would be applied for the 15 SO2 group 1 states, i.e.,
the more stringent SO2 tier, if EPA were to finalize an alternative
remedy that uses variability limits for the years 2012 and 2013 (the
Transport Rule’s first phase).

As discussed in the preamble, EPA also considered, instead of 2-year
average limits for the 15 SO2 group 1 states in 2012 and 2013, 3-year
average limits for these states starting in 2014 (for Phase 1).  EPA
considered the alternative of 3-year average limits starting in 2014 for
these states because this is the approach EPA would apply to SO2 group 2
states (and all states for NOX emissions) in the event EPA were to
finalize a remedy that uses variability limits in the first phase.  The
group 1 states have different SO2 budgets in 2012 and 2013 than in 2014
and beyond.

2+2+2) = 6481.7/32 = 720.2.

 =(1/3)*(1000+1000+700)=2700/3=900 thousand tons of SO2.

EPA calculated the 3-year average of the 95th percent confidence level
limits for 2014 for the group 1 states using these equations and
compared these to the 2-year average of the 95th percent confidence
level limits that are presented in the preamble.  These 3-year and
2-year limits are shown in Table 21.  As discussed in the preamble, EPA
believes that the 2-year average limit approach is reasonable (and
preferable if the alternative approach is chosen where variability
limits are applied during the first phase).  EPA’s proposed remedy
does not use variability limits during the first phase (2012 and 2013),
however, as explained in the preamble, EPA is also taking comment on an
alternative approach that would use variability limits in Phase 1.  The
2-year average limits presented in the preamble and shown in Table 21
are calculated based on the Phase 1 SO2 emissions budgets and are
intended for use in 2013.  In contrast, the alternative 3-year average
limits shown in Table 21, while intended to limit SO2 emissions in Phase
1, are impacted (reduced) by the tighter Phase 2 SO2 budgets that apply
to these states starting in 2014.  If it is ultimately decided to use
variability limits in Phase 1, EPA believes it is more appropriate to
base the multi-year average limits for Phase 1 on the stringency of the
Phase 1 emissions budgets, not the stringency of the Phase 2 budgets,
thus it is preferable to use the 2-year limits.

Table 21.  Phase 1 Variability Limits for SO2 Group 1 States: Comparison
of 2-Year Average to Alternative 3-Year Average Limits of the 95th
Percent Confidence Level.

SO2 Group 1 State	SO2 Annual Emissions Budgets

(tons)	Phase 1 Multi-Year Variability Limits

(tons)

	2012	2014	2-Year Limits* 	3-Year Limits** 

Georgia	233,260	85,717	16,494	11,361

Illinois	208,957	151,530	14,775	11,070

Indiana	400,378	201,412	28,311	20,033

Iowa	94,052	86,088	6,651	5,281

Kentucky	219,549	113,844	15,524	11,023

Michigan	251,337	155,675	17,772	12,935

Missouri	203,689	158,764	14,403	10,964

New York	66,542	42,041	4,705	3,436

North Carolina	111,485	81,859	7,883	5,922

Ohio	464,964	178,307	32,878	22,710

Pennsylvania	388,612	141,693	27,479	18,918

Tennessee	100,007	100,007	7,072	5,774

Virginia	72,595	40,785	5,133	3,682

West Virginia	205,422	119,016	14,526	10,465

Wisconsin	96,439	66,683	6,819	5,060



* 2-year average limits on 2012-2013 emissions for SO2 group 1 states
for the alternative where variability limits would be used in the first
phase (as presented in preamble section IV.F).

** Alternative approach of 3-year average limits on 2012-2014 emissions
for SO2 group 1 states for the alternative where variability limits
would be used in the first phase.

4. Results of an analysis done using the air quality assessment tool

The objective of this section is to estimate the possible effects on air
quality of the variability in emissions. This analysis was done using
the air quality assessment tool, or AQAT, and uses state-by-state
emissions to estimate downwind state-by-state air quality contributions
at various nonattainment and maintenance monitors (See the Analysis for
Significant Contribution TSD for details on the construction and use of
the AQAT as well as the estimated downwind air quality concentrations
resulting from the proposed remedy).   See preamble section IV.C and the
Air Quality Modeling TSD for a list of the nonattainment and maintenance
monitors.  

For this analysis, EPA varied the SO2 emissions of each upwind state
included in the proposed Transport Rule around the proposed budgets,
simulating the effects of variation in emissions resulting from the
proposed allowed variability under the variability limits, and estimated
the resulting variability in air quality (daily PM2.5 concentrations).  
This analysis focused on variability in emissions related to the daily
PM2.5 concentrations for several reasons.  First, the number of monitors
classified as nonattainment and/or maintenance was larger than for the
other standards.  Second, generally, sites required larger emissions
reductions and larger relative air quality improvements (under the daily
PM2.5  NAAQS, compared with the annual PM2.5 NAAQS).  Lastly, the SO2
emissions reductions from EGUs relative to total SO2 emissions for daily
PM2.5 standard is proportionally larger than NOx emissions reductions
from EGUs relative to total NOx emissions.  Consequently, the
variability in estimated daily PM2.5 concentrations relative to
variation in SO2 emissions is larger than variability in estimated ozone
concentrations relative to variation in NOx emissions.  

Two approaches were taken to estimate the variation in downwind air
quality at each monitor for daily PM2.5 allowed under the Transport Rule
in 2014 due to the inherent variability in SO2 emissions.  Each of these
approaches will be described, in turn, in the following paragraphs.  To
summarize the two approaches:  The first approach examined the 1-year
variability effects on daily PM2.5 concentration when variations in
emissions from different states are independent from each other.  This
is intended to represent “typical” random variations in emissions
and the resulting typical variations in air quality that might be seen
under the Transport Rule.  The second approach examined the “worst”
case 1-year scenario for each monitor, when the upwind states with the
largest impacts per ton emit at the upper end of the variability limit,
while upwind states with the lowest impacts per ton emit below their
budgets.  This is intended to estimate an upper bound for the effects of
emissions variability on air quality.   For both approaches, the effects
of the inherent variation in emissions on daily PM2.5 concentrations
were estimated to be small. 

For the first approach, the SO2 emissions for each state included in the
proposed control region were allowed to randomly vary around the level
of the budgets.  That is, by chance, one state may increase its
emissions, while another state may decrease its emissions.  This random
variation is intended to represent the inherent “random” variations
in heat input (and emissions) that were characterized in section 2.  As
with the analysis in section 2, variations were assumed to be
“normally” distributed and characterized by a standard deviation
equal to 5.1% of the state’s budget.  This standard deviation is
derived from the two-tailed 95% confidence level 1-year variability
limit (equal to 10% of a state’s budget for many states).  (Note that
this process did not account for the fact that the proposed 1-year
variability limits for some of the smaller states are larger than 10% of
the budgets.)  This process also did not account for banking due to
early reductions.  The reader should keep in mind that for this
approach, when the state-by-state variations in emissions are assessed
across all states included in the proposed Transport Rule region, the
emissions from states that emit over their budgets largely cancel out
with emissions from states that emit below their budgets, with the
result that total region-wide emissions are relatively constant.   

As described in section 2b of this TSD when determining the
state-by-state year-to-year variability in emissions, we assume, on a
state-by-state basis that: (1) from year-to-year, the differences
between the budget and modeled emissions are “normally” distributed;
(2) the yearly emissions are independent from each other and are
independent from the emissions of any other state; and, (3) the
distribution of hourly emissions values is the same from year to year
(i.e., the “within-year” variance is the same each year).

g/m3).  

For an individual monitor, the air quality impact per ton of emissions
can be estimated using the daily PM2.5 2012 base case air quality
sulfate contributions from the CAMx source-apportionment modeling
results and the 2012 base case emissions inventory also used in the
modeling (see the Air Quality Modeling TSD and section IV.C of the
preamble for details on both of these data sets).  For each individual
upwind state contributing to a particular monitor, the estimated impact
per ton of sulfate is the air quality contribution of sulfate from that
state to the particular monitor, divided by the state’s total 2012
base case emission inventory of SO2.  The standard deviation in
emissions (in tons) for each state can be found by multiplying the
proposed 2014 state budget by 0.1 (10%) and dividing by 1.960 (to
convert the variability limit in emissions at the 95% confidence level
back to the standard deviation level).  The state-by-state SO2 budgets
in 2014 (and the budgets multiplied by 0.1) can be found in Table 18.

For each combination of upwind state and downwind monitor, the impact
per ton is then multiplied by the standard deviation of emissions (in
tons).  This product is then squared, becoming the variance.  The
variances from states that were modeled in IPM as well as in the CAMx
source-apportionment air quality modeling, but were not included in the
proposed remedy, were assumed to be equal to zero (i.e., they were
assumed to have no variability in their emissions).  

g/m3) that we would expect to see, on average, at each monitor
location due to the year-to-year variability in emissions (Table 22). 
The estimated variations in air quality are quite small (a fraction of a
percent) relative to the 2012 base case design values.  The locations in
Table 22 are in order of decreasing 2012 base case average design value.

g/m3.  

In conclusion, we found that, even while allowing each state’s
emissions to randomly vary up to 10% of its budget (the 2-tailed 95%
confidence variability level prescribed for many states in the Transport
Rule), the combined downwind air quality impacts were essentially
negligible.  

In the second approach, EPA examined the “worst” case scenario for
each monitor location.  This approach is intended to simulate a
situation where all of the states in the region surrounding a monitor
increase their SO2 emissions to the maximum amount possible (the 1-year
variability limit for each state), while states far away make all of the
emissions reductions to compensate.  In this scenario, the covered
upwind states with the largest air quality impacts per ton of emissions
were modeled to increase their SO2 emissions up to their proposed 2014
SO2 budget variability limit (Table 18). The upwind states with the
lowest air quality impacts per ton of emissions were modeled to reduce
their emissions by the number of tons equal to the variability limit
(Table 18).  This was done such that the total emissions (i.e., the sum
of the budgets with variability changes) remained exactly at the total
emissions level for the proposed region (i.e., the sum of all the state
budgets).  In other words, for this approach, on a monitor-by-monitor
basis, emissions increases in the upwind states that have the largest
air quality impacts per ton are paired with equivalent emissions
decreases in states that are having the lowest air quality impacts per
ton.  A result of this approach is that overall regional emissions do
not change at all from the sum of the budgets.  

The state containing the monitor was included in this analysis (unless
the state was not included in the proposed Transport Rule region).  For
the “middle” state (i.e., the state whose impact per ton value was
such that the total emissions from states with higher impacts per ton
equaled the total emissions from states with lower impacts per ton), the
emissions were set at an intermediate level.  States that were not
included in the proposed Transport Rule region had their variability in
emissions set to zero.  The results of this analysis, seen in Table 22
in the “worst” case columns, show the cumulative air quality impact
from all states and from just the upwind states.  The average values
were 0.30 g/m3 for all states, and 0.20 g/m3 for just the upwind
states, while the maximum values were 0.66 g/m3 and 0.35 g/m3,
respectively.  Again, the air quality impacts, while larger than for the
random variability approaches, are still substantially less than the
difference between the average and maximum design values.  These results
suggest that even under a “worst case” scenario, where nearby states
minimize reductions in emissions, while states far away maximize
reduction, the resulting increases in air quality are small relative to
other factors (i.e., weather). 

Collectively, by assessing these two approaches, it appears that the
magnitude of the variability in air quality resulting from the proposed
levels of variability in emissions is likely to be smaller than other
factors impacting air quality.   These results suggest that the
estimated variations in air quality resulting from the small variations
in emissions (even under “worst-case” scenarios) are not
substantial.  The variations are much smaller than documented
year-to-year variability in air quality (as measured at the monitors and
expressed as the difference between the average and maximum design
values).  Consequently, allowing variation in emissions under the
proposed variability limits in the Transport Rule, while allowing
flexibility for the power sector to address inherent fluctuations in
electric generation, does not overly affect air quality.



g/m3) in 2014 are Shown for Each Downwind Monitor for Daily PM2.5.  
The 2012 Base Case Average and Maximum Design Values as well as the
Estimated Standard Deviation, the 95% Confidence Level, and the Worst
Case Variability Estimates Are Shown.

Monitor Identification Number	Receptor State	Receptor County	2012 Base
Case Average DV (g/m3)	2012 Base Case Maximum DV (g/m3)	Difference
Between Average and Maximum 2012 Base DVs (g/m3)	Standard Deviation
(All States)	Standard Deviation (Upwind States)	95% Confidence Level
(All States)	95% Confidence Level (Upwind States)	Worst Case (All
States)	Worst Case (Upwind States)

420030064	Pennsylvania	Allegheny	58.8	62.3	3.5	0.047	0.038	0.092	0.074
0.195	0.141

261630033	Michigan	Wayne	42.1	42.6	0.5	0.104	0.043	0.205	0.085	0.345
0.159

390350038	Ohio	Cuyahoga	41.2	44.0	2.8	0.086	0.048	0.168	0.094	0.342
0.202

420030093	Pennsylvania	Allegheny	41.1	46.2	5.1	0.060	0.052	0.117	0.102
0.253	0.196

170311016	Illinois	Cook	41.0	44.1	3.1	0.050	0.039	0.098	0.076	0.218
0.156

261630016	Michigan	Wayne	40.6	43.0	2.4	0.093	0.033	0.182	0.065	0.307
0.138

180970043	Indiana	Marion	40.5	42.0	1.5	0.126	0.057	0.248	0.112	0.458
0.237

390170003	Ohio	Butler	40.3	42.3	2.0	0.109	0.089	0.214	0.174	0.472	0.348

180970066	Indiana	Marion	40.3	41.8	1.5	0.132	0.058	0.258	0.113	0.480
0.247

420210011	Pennsylvania	Cambria	40.3	40.7	0.4	0.102	0.055	0.200	0.107
0.389	0.220

180970081	Indiana	Marion	40.1	41.1	1.0	0.117	0.054	0.228	0.106	0.425
0.223

010730023	Alabama	Jefferson	40.0	40.7	0.7	0.132	0.020	0.259	0.040	0.324
0.068

171191007	Illinois	Madison	40.0	40.6	0.6	0.067	0.055	0.132	0.108	0.282
0.208

540090011	West Virginia	Brooke	39.9	40.8	0.9	0.050	0.047	0.097	0.091
0.219	0.187

390618001	Ohio	Hamilton	39.6	40.3	0.7	0.071	0.051	0.140	0.100	0.286
0.188

390350060	Ohio	Cuyahoga	39.4	42.8	3.4	0.072	0.034	0.142	0.066	0.266
0.140

171190023	Illinois	Madison	39.4	40.2	0.8	0.105	0.084	0.206	0.164	0.422
0.298

180970083	Indiana	Marion	39.0	39.3	0.3	0.132	0.059	0.259	0.115	0.485
0.253

550790043	Wisconsin	Milwaukee	38.8	39.7	0.9	0.070	0.060	0.137	0.117
0.318	0.248

180970078	Indiana	Marion	38.7	39.7	1.0	0.126	0.057	0.246	0.111	0.455
0.236

261630019	Michigan	Wayne	38.6	39.1	0.5	0.063	0.032	0.124	0.062	0.235
0.128

170310052	Illinois	Cook	38.5	39.7	1.2	0.049	0.033	0.096	0.064	0.196
0.125

261630015	Michigan	Wayne	38.5	39.1	0.6	0.105	0.050	0.206	0.098	0.365
0.184

390170017	Ohio	Butler	38.5	38.5	0.0	0.090	0.073	0.176	0.144	0.390	0.287

261470005	Michigan	St. Clair	38.4	39.4	1.0	0.059	0.035	0.115	0.068	0.233
0.141

170313301	Illinois	Cook	38.2	41.0	2.8	0.059	0.043	0.115	0.084	0.252
0.173

340172002	New Jersey	Hudson	38.2	38.2	0.0	0.035	0.030	0.069	0.059	0.186
0.132

180190006	Indiana	Clark	38.1	40.2	2.1	0.081	0.055	0.159	0.107	0.341
0.223

261610008	Michigan	Washtenaw	38.1	39.8	1.7	0.067	0.046	0.130	0.090	0.261
0.166

010732003	Alabama	Jefferson	38.1	38.9	0.8	0.083	0.017	0.163	0.034	0.226
0.066

170313103	Illinois	Cook	38.1	38.7	0.6	0.046	0.028	0.090	0.055	0.189
0.118

420031008	Pennsylvania	Allegheny	38.0	39.3	1.3	0.070	0.058	0.137	0.114
0.291	0.215

390610006	Ohio	Hamilton	38.0	38.0	0.0	0.091	0.074	0.179	0.146	0.385
0.281

261250001	Michigan	Oakland	37.9	38.4	0.5	0.082	0.052	0.160	0.102	0.325
0.201

390171004	Ohio	Butler	37.8	38.6	0.8	0.089	0.071	0.175	0.139	0.381	0.274

420710007	Pennsylvania	Lancaster	37.7	40.1	2.4	0.033	0.012	0.066	0.024
0.135	0.074

420070014	Pennsylvania	Beaver	37.7	39.1	1.4	0.057	0.051	0.111	0.100
0.242	0.195

550790010	Wisconsin	Milwaukee	37.7	39.0	1.3	0.049	0.037	0.095	0.072
0.241	0.178

390617001	Ohio	Hamilton	37.7	38.1	0.4	0.085	0.064	0.166	0.126	0.367
0.259

390610014	Ohio	Hamilton	37.5	38.5	1.0	0.081	0.059	0.159	0.116	0.330
0.221

390170016	Ohio	Butler	37.5	37.8	0.3	0.086	0.075	0.169	0.146	0.378	0.293

170316005	Illinois	Cook	37.4	39.8	2.4	0.048	0.031	0.094	0.060	0.193
0.121

180890022	Indiana	Lake	37.3	42.1	4.8	0.058	0.028	0.114	0.054	0.218	0.117

180970079	Indiana	Marion	37.2	38.3	1.1	0.119	0.061	0.234	0.119	0.456
0.255

171192009	Illinois	Madison	37.2	38.2	1.0	0.095	0.076	0.186	0.149	0.388
0.278

390610042	Ohio	Hamilton	37.2	38.0	0.8	0.083	0.058	0.163	0.113	0.342
0.225

360610056	New York	New York	37.1	38.0	0.9	0.040	0.016	0.078	0.032	0.167
0.096

420030116	Pennsylvania	Allegheny	37.1	37.1	0.0	0.060	0.051	0.117	0.100
0.259	0.198

261150005	Michigan	Monroe	37.0	38.0	1.0	0.074	0.063	0.145	0.123	0.321
0.243

210590005	Kentucky	Daviess	37.0	37.0	0.0	0.133	0.110	0.260	0.215	0.494
0.347

550790099	Wisconsin	Milwaukee	36.8	37.7	0.9	0.053	0.044	0.104	0.086
0.260	0.201

191630019	Iowa	Scott	36.8	36.8	0.0	0.083	0.073	0.162	0.143	0.407	0.330

340390004	New Jersey	Union	36.7	37.2	0.5	0.030	0.026	0.059	0.050	0.173
0.127

420031301	Pennsylvania	Allegheny	36.6	38.6	2.0	0.062	0.052	0.121	0.101
0.266	0.199

471251009	Tennessee	Montgomery	36.6	37.9	1.3	0.068	0.058	0.133	0.113
0.320	0.249

390490024	Ohio	Franklin	36.6	37.6	1.0	0.064	0.040	0.126	0.078	0.267
0.168

390811001	Ohio	Jefferson	36.5	39.9	3.4	0.065	0.043	0.126	0.084	0.261
0.166

390350065	Ohio	Cuyahoga	36.5	38.9	2.4	0.072	0.035	0.142	0.070	0.274
0.151

180372001	Indiana	Dubois	36.5	38.0	1.5	0.102	0.055	0.200	0.109	0.380
0.212

171193007	Illinois	Madison	36.5	37.3	0.8	0.103	0.083	0.202	0.163	0.420
0.300

295100087	Missouri	St. Louis City	36.4	36.9	0.5	0.097	0.071	0.191	0.139
0.395	0.264

550790026	Wisconsin	Milwaukee	36.3	40.1	3.8	0.046	0.036	0.091	0.071
0.226	0.170

180890026	Indiana	Lake	36.3	39.3	3.0	0.051	0.036	0.101	0.070	0.218	0.145

391130032	Ohio	Montgomery	36.3	38.5	2.2	0.090	0.068	0.177	0.134	0.393
0.277

245100040	Maryland	Baltimore (City)	36.3	38.3	2.0	0.068	0.013	0.133
0.026	0.187	0.057

170310076	Illinois	Cook	36.3	37.3	1.0	0.066	0.051	0.130	0.099	0.278
0.195

180970042	Indiana	Marion	36.3	37.2	0.9	0.129	0.060	0.252	0.118	0.480
0.257

261630036	Michigan	Wayne	36.3	36.9	0.6	0.069	0.042	0.136	0.083	0.256
0.149

360610128	New York	New York	36.2	38.0	1.8	0.042	0.032	0.083	0.062	0.221
0.167

390490025	Ohio	Franklin	36.1	36.4	0.3	0.062	0.039	0.122	0.077	0.253
0.157

390350045	Ohio	Cuyahoga	36.0	39.0	3.0	0.078	0.040	0.153	0.078	0.302
0.170

211110044	Kentucky	Jefferson	36.0	36.5	0.5	0.094	0.076	0.184	0.149	0.359
0.252

390610043	Ohio	Hamilton	36.0	36.4	0.4	0.091	0.078	0.179	0.154	0.395
0.304

295100007	Missouri	St. Louis City	36.0	36.3	0.3	0.133	0.075	0.261	0.147
0.497	0.282

421330008	Pennsylvania	York	35.9	38.8	2.9	0.058	0.025	0.113	0.049	0.215
0.113

181570008	Indiana	Tippecanoe	35.9	36.9	1.0	0.080	0.050	0.156	0.099	0.345
0.225

180830004	Indiana	Knox	35.9	36.5	0.6	0.093	0.049	0.182	0.096	0.355	0.200

420030008	Pennsylvania	Allegheny	35.9	36.3	0.4	0.052	0.042	0.102	0.081
0.217	0.155

360050080	New York	Bronx	35.9	36.2	0.3	0.041	0.023	0.081	0.045	0.201
0.133

390610040	Ohio	Hamilton	35.8	36.8	1.0	0.084	0.065	0.166	0.128	0.377
0.272

211110043	Kentucky	Jefferson	35.8	36.4	0.6	0.101	0.087	0.197	0.170	0.382
0.282

420430401	Pennsylvania	Dauphin	35.7	37.1	1.4	0.038	0.017	0.074	0.034
0.160	0.095

170310057	Illinois	Cook	35.7	37.0	1.3	0.050	0.032	0.097	0.063	0.210
0.136

090091123	Connecticut	New Haven	35.7	36.6	0.9	0.031	0.030	0.060	0.059
0.223	0.162

290990012	Missouri	Jefferson	35.7	36.5	0.8	0.122	0.074	0.239	0.144	0.474
0.284

340171003	New Jersey	Hudson	35.7	36.1	0.4	0.038	0.029	0.075	0.057	0.217
0.143

170312001	Illinois	Cook	35.6	38.2	2.6	0.064	0.059	0.125	0.116	0.258
0.211

391530017	Ohio	Summit	35.6	37.2	1.6	0.069	0.040	0.135	0.078	0.274	0.164

211110048	Kentucky	Jefferson	35.6	36.4	0.8	0.084	0.068	0.165	0.134	0.331
0.234

291831002	Missouri	Saint Charles	35.5	37.1	1.6	0.107	0.072	0.209	0.141
0.435	0.281

245100049	Maryland	Baltimore (City)	35.5	35.5	0.0	0.066	0.013	0.130
0.025	0.180	0.052

261630001	Michigan	Wayne	35.4	37.8	2.4	0.085	0.056	0.166	0.111	0.345
0.221

360610062	New York	New York	35.3	37.0	1.7	0.033	0.025	0.064	0.050	0.180
0.140

420410101	Pennsylvania	Cumberland	35.3	37.0	1.7	0.033	0.018	0.065	0.035
0.142	0.087

390810017	Ohio	Jefferson	35.3	36.8	1.5	0.062	0.042	0.121	0.082	0.256
0.167

171630010	Illinois	Saint Clair	35.3	35.9	0.6	0.099	0.080	0.195	0.157
0.402	0.288

295100085	Missouri	St. Louis City	35.3	35.7	0.4	0.128	0.073	0.251	0.143
0.484	0.278

181670023	Indiana	Vigo	35.1	36.5	1.4	0.099	0.048	0.195	0.093	0.384	0.212

550250047	Wisconsin	Dane	35.1	36.1	1.0	0.056	0.040	0.109	0.078	0.254
0.179

471650007	Tennessee	Sumner	35.1	36.0	0.9	0.081	0.067	0.159	0.131	0.376
0.285

171971002	Illinois	Will	35.1	35.8	0.7	0.052	0.040	0.101	0.079	0.222
0.158

210290006	Kentucky	Bullitt	35.0	36.3	1.3	0.090	0.073	0.176	0.143	0.377
0.273

170310022	Illinois	Cook	34.9	36.6	1.7	0.070	0.069	0.138	0.134	0.214
0.184

551330027	Wisconsin	Waukesha	34.9	35.6	0.7	0.048	0.042	0.094	0.081	0.225
0.178

550790059	Wisconsin	Milwaukee	34.8	36.3	1.5	0.056	0.047	0.111	0.093
0.261	0.202

245100035	Maryland	Baltimore (City)	34.7	35.5	0.8	0.065	0.012	0.127
0.023	0.171	0.046

390350027	Ohio	Cuyahoga	34.5	36.6	2.1	0.065	0.027	0.128	0.053	0.231
0.115

191390015	Iowa	Muscatine	34.5	36.0	1.5	0.038	0.026	0.074	0.050	0.158
0.104

211451004	Kentucky	McCracken	34.4	36.8	2.4	0.087	0.079	0.171	0.155	0.362
0.289

420030095	Pennsylvania	Allegheny	34.3	36.6	2.3	0.052	0.043	0.102	0.083
0.223	0.165

180431004	Indiana	Floyd	34.3	35.7	1.4	0.097	0.062	0.191	0.121	0.388
0.240

391130031	Ohio	Montgomery	34.3	35.6	1.3	0.065	0.048	0.127	0.094	0.275
0.189

391351001	Ohio	Preble	34.3	35.5	1.2	0.087	0.073	0.170	0.143	0.376	0.284

390950024	Ohio	Lucas	34.2	36.5	2.3	0.058	0.047	0.113	0.093	0.240	0.176

360610079	New York	New York	34.2	36.4	2.2	0.050	0.027	0.099	0.054	0.244
0.161

390990014	Ohio	Mahoning	34.2	35.8	1.6	0.053	0.030	0.103	0.059	0.211
0.127

170310050	Illinois	Cook	34.1	35.8	1.7	0.063	0.055	0.123	0.108	0.273
0.214

110010041	District Of Columbia	District Of Columbia	34.0	35.6	1.6	0.043
0.043	0.085	0.085	0.655	0.193

540090005	West Virginia	Brooke	33.9	36.1	2.2	0.055	0.049	0.107	0.096
0.226	0.178

391550007	Ohio	Trumbull	33.9	35.6	1.7	0.051	0.033	0.100	0.064	0.217
0.141

421255001	Pennsylvania	Washington	33.9	35.5	1.6	0.071	0.068	0.139	0.132
0.288	0.245

420033007	Pennsylvania	Allegheny	33.8	38.5	4.7	0.066	0.056	0.130	0.109
0.281	0.210

240031003	Maryland	Anne Arundel	33.8	36.7	2.9	0.057	0.027	0.112	0.053
0.225	0.126

180390003	Indiana	Elkhart	33.8	35.6	1.8	0.056	0.030	0.110	0.059	0.197
0.104

212270007	Kentucky	Warren	33.7	36.3	2.6	0.086	0.079	0.169	0.154	0.385
0.315

390350034	Ohio	Cuyahoga	33.7	35.7	2.0	0.061	0.036	0.120	0.070	0.246
0.149

170314007	Illinois	Cook	33.6	35.7	2.1	0.065	0.047	0.128	0.093	0.268
0.180

390950026	Ohio	Lucas	33.6	35.6	2.0	0.068	0.050	0.134	0.098	0.299	0.208

110010042	District Of Columbia	District Of Columbia	33.0	35.6	2.6	0.034
0.034	0.067	0.067	0.606	0.159





Appendix A:

Alternative Method for Identifying 1-Year Tonnage and Percentage Limits
Table A-1. The Effects of Various Combinations of the Alternative
Method Tonnage and Percentage Variability Limits on Annual EGU SO2
Emissions (See Notes Below).

	Percentage Limit

Tonnage Limit	2	4	6	8	10	12	14	16	18	20

1,000	22	12	7	7	NJ, VA, 	NJ, VA, 	NJ, 	NJ, 	302,343	345,824

1,300	21	11	6	6	VA, 	VA, 	216,599	259,662	302,879	346,236

1,600	20	11	6	6	VA, 	VA, 	217,799	260,853	303,907	346,997

1,900	20	11	6	6	VA, 	VA, 	218,999	262,053	305,107	348,162

2,200	19	11	6	6	VA, 	VA, 	220,199	263,253	306,307	349,362

2,500	18	10	5	5	VA, 	VA, 	221,399	264,453	307,507	350,562

2,800	17	9	5	5	VA, 	VA, 	222,599	265,653	308,707	351,762

3,100	15	7	5	5	VA, 	VA, 	223,799	266,853	309,907	352,962

3,400	12	5	GA, IA, NY, VA, 	GA, IA, NY, VA, 	VA, 	VA, 	224,999	268,053
311,107	354,162

3,700	10	GA, IA, TN, VA, 	GA, IA, VA, 	GA, IA, VA, 	VA, 	VA, 	226,199
269,253	312,307	355,362

4,000	8	GA, IA, TN, VA, 	GA, IA, VA, 	GA, IA, VA, 	VA, 	VA, 	227,671
270,453	313,507	356,562

4,300	8	GA, IA, TN, VA, 	GA, IA, VA, 	GA, IA, VA, 	VA, 	VA, 	229,308
271,692	314,707	357,762

4,600	7	GA, IA, TN, VA, 	GA, IA, VA, 	GA, IA, VA, 	VA, 	VA, 	231,108
273,192	315,907	358,962

4,900	6	GA, IA, TN, VA, 	GA, IA, VA, 	GA, IA, VA, 	VA, 	VA, 	232,908
274,835	317,214	360,162

5,200	GA, IA, MI, VA, 	GA, IA, VA, 	GA, IA, VA, 	GA, IA, VA, 	VA, 	VA, 
234,708	276,635	318,714	361,362

5,500	GA, IA, MI, VA, 	GA, IA, VA, 	GA, IA, VA, 	GA, IA, VA, 	VA, 	VA, 
236,797	278,435	320,362	362,736

5,800	GA, IA, MI, VA, 	GA, IA, VA, 	GA, IA, VA, 	GA, IA, VA, 	VA, 	VA, 
239,191	280,235	322,162	364,236

6,100	GA, IA, 	GA, IA, 	GA, IA, 	GA, IA, 	165,234	202,313	241,591
282,179	323,962	365,889

6,400	GA, 	GA, 	GA, 	GA, 	169,134	205,498	244,151	284,387	325,762
367,689

6,700	90,919	93,882	112,952	140,982	173,038	208,821	246,862	286,787
327,562	369,489

7,000	99,019	100,931	118,652	146,082	177,238	212,421	249,862	289,187
329,662	371,289

7,300	107,119	108,222	124,781	151,182	181,438	216,021	252,911	291,755
331,983	373,089

7,600	115,219	116,022	131,081	156,282	186,077	219,916	256,211	294,455
334,383	375,044

7,900	123,319	123,822	137,381	161,502	190,877	223,816	259,621	297,411
336,783	377,179

8,200	131,419	131,622	143,681	166,902	195,748	227,881	263,221	300,411
339,360	379,579

8,500	139,519	139,519	150,163	172,363	200,848	232,081	266,821	303,624
342,060	381,979

8,800	147,619	147,619	157,063	178,063	205,948	236,281	270,699	306,924
344,960	384,379

9,100	155,719	155,719	163,963	183,763	211,048	240,968	274,599	310,421
347,960	386,964

9,400	163,819	163,819	170,863	189,466	216,148	245,768	278,525	314,021
351,037	389,664

9,700	171,919	171,919	177,763	195,668	221,248	250,568	282,725	317,621
354,337	392,509

10,000	180,019	180,019	184,663	201,968	226,623	255,614	286,925	321,482
357,637	395,509

10,300	188,119	188,119	191,586	208,268	232,023	260,714	291,157	325,382
361,221	398,509

10,600	196,219	196,219	198,786	214,568	237,474	265,814	295,859	329,282
364,821	401,750

10,900	204,319	204,319	206,024	220,868	243,174	270,914	300,659	333,369
368,421	405,050

11,200	212,419	212,419	213,824	227,228	248,874	276,014	305,459	337,569
372,265	408,422

11,500	220,519	220,519	221,624	234,045	254,574	281,114	310,380	341,769
376,165	412,022

11,800	228,619	228,619	229,424	240,945	260,328	286,343	315,480	346,034
380,065	415,622

12,100	236,719	236,719	237,224	247,845	266,555	291,743	320,580	350,751
384,012	419,222

12,400	244,819	244,819	245,024	254,745	272,855	297,143	325,680	355,551
388,212	423,048

12,700	252,919	252,919	252,919	261,645	279,155	302,585	330,780	360,351
392,412	426,948

13,000	261,019	261,019	261,019	268,545	285,455	308,285	335,880	365,151
396,612	430,848

13,300	269,119	269,119	269,119	275,445	291,755	313,985	340,980	370,245
400,911	434,748

13,600	277,219	277,219	277,219	282,345	298,055	319,685	346,080	375,345
405,642	438,856



*Numbers in white cells represent the sum (across all states) of the
differences in emissions between the state-specific “alternative
approach” variability values and the state-specific variability limit
(in tons); the variability limit selected for each state is the larger
of either the percentage limit or the tonnage limit.

** If the cell is grey, it means that there is a state or there are
several states whose “alternative approach” level variability values
exceed both the tonnage and percentage limits.  The cell either lists
the states that could exceed the limits, or lists the number of states.

*** If the cell is yellow, this is the combination tonnage and
percentage limits that minimizes the sum of the differences between the
state-specific “alternative approach”  variability values and the
state-specific variability limit (in tons).Table A-2. The Effects of
Various Combinations of the Alternative Method Tonnage and Percentage
Variability Limits on Annual EGU NOx Emissions (See Notes Below).

	Percentage Limit

Tonnage Limit 3	5	10	15	20	25	30	35	40	45	50

1,000	13	GA, LA, NY, 	NY, 	160,818	216,772	272,730	328,701	384,804
440,907	497,010

1,500	12	GA, LA, NY, 	NY, 	162,461	218,027	273,737	329,688	385,646
441,604	497,562

2,000	9	GA, LA, 	109,477	164,461	219,830	275,237	330,946	386,655	442,604
498,562

2,500	6	GA, LA, 	112,477	166,710	221,830	277,200	332,570	388,155	443,864
499,573

3,000	5	GA, 	115,477	169,642	224,017	279,200	334,570	389,939	445,364
501,073

3,500	5	GA, 	118,916	172,642	226,807	281,324	336,570	391,939	447,309
502,678

4,000	FL, GA, IA, PA, 	GA, 	123,005	175,642	229,807	283,972	338,631
393,939	449,309	504,678

4,500	FL, GA, PA, 	GA, 	127,840	178,957	232,807	286,972	341,137	395,939
451,309	506,678

5,000	FL, PA, 	91,796	133,376	182,894	235,807	289,972	344,137	398,438
453,309	508,678

5,500	FL, PA, 	101,282	139,376	187,180	239,076	292,972	347,137	401,303
455,745	510,678

6,000	FL, PA, 	111,184	145,758	192,125	242,872	295,972	350,137	404,303
458,468	513,052

6,500	FL, 	121,708	153,315	197,600	246,872	299,195	353,137	407,303
461,468	515,633

7,000	FL, 	132,708	161,315	203,507	251,411	302,850	356,137	410,303
464,468	518,633

7,500	FL, 	143,875	170,203	209,507	256,411	306,850	359,313	413,303
467,468	521,633

8,000	150,982	155,375	179,682	216,017	261,879	311,029	362,828	416,303
470,468	524,633

8,500	164,482	167,069	189,182	223,463	267,639	315,697	366,828	419,432
473,468	527,633

9,000	177,982	179,199	199,284	231,426	273,639	320,697	370,828	422,932
476,468	530,633

9,500	191,482	192,199	209,784	239,610	279,708	326,159	375,203	426,806
479,551	533,633

10,000	204,982	205,199	220,571	248,610	286,276	331,770	379,983	430,806
483,051	536,633

10,500	218,482	218,482	231,571	258,082	293,708	337,770	384,983	434,878
486,784	539,670

11,000	231,982	231,982	242,572	267,582	301,536	343,770	390,438	439,378
490,784	543,170

11,500	245,482	245,482	254,072	277,197	309,536	349,953	395,938	444,269
494,784	546,762

12,000	258,982	258,982	265,572	287,385	318,018	356,535	401,901	449,269
499,052	550,762

12,500	272,482	272,482	277,112	297,885	327,018	363,954	407,901	454,718
503,555	554,762

13,000	285,982	285,982	289,112	308,433	336,482	371,647	413,901	460,218
508,555	558,762

13,500	299,482	299,482	301,308	319,433	345,982	379,647	420,198	466,033
513,555	563,227

14,000	312,982	312,982	314,308	330,433	355,482	387,647	426,794	472,033
518,997	567,841

14,500	326,482	326,482	327,308	341,433	365,250	396,425	434,199	478,033
524,497	572,841

15,000	339,982	339,982	340,308	352,769	375,486	405,425	441,758	484,033
530,164	577,841

15,500	353,482	353,482	353,482	364,269	385,986	414,883	449,758	490,443
536,164	583,277

16,000	366,982	366,982	366,982	375,769	396,486	424,383	457,758	497,052
542,164	588,777

16,500	380,482	380,482	380,482	387,269	407,296	433,883	465,910	504,444
548,164	594,295

17,000	393,982	393,982	393,982	399,156	418,296	443,383	474,832	511,944
554,188	600,295

17,500	407,482	407,482	407,482	411,156	429,296	453,304	483,832	519,869
560,688	606,295

18,000	420,982	420,982	420,982	423,416	440,296	463,586	493,283	527,869
567,311	612,295

18,500	434,482	434,482	434,482	436,416	451,465	474,086	502,783	535,869
574,689	618,295

19,000	447,982	447,982	447,982	449,416	462,965	484,586	512,283	544,239
582,189	624,433

19,500	461,482	461,482	461,482	462,416	474,465	495,159	521,783	553,239
589,980	630,933

20,000	474,982	474,982	474,982	475,416	485,965	506,159	531,358	562,239
597,980	637,570

20,500	488,482	488,482	488,482	488,482	497,465	517,159	541,358	571,683
605,980	644,935

21,000	501,982	501,982	501,982	501,982	509,199	528,159	551,687	581,183
613,980	652,435

21,500	515,482	515,482	515,482	515,482	521,199	539,159	562,187	590,683
622,646	660,091

22,000	528,982	528,982	528,982	528,982	533,199	550,162	572,687	600,183
631,646	668,091



*Numbers in white cells represent the sum (across all states) of the
differences in emissions between the state-specific “alternative
approach” variability values and the state-specific variability limit
(in tons); the variability limit selected for each state is the larger
of either the percentage limit or the tonnage limit.

** If the cell is grey, it means that there is a state or there are
several states whose “alternative approach” level variability values
exceed both the tonnage and percentage limits.  The cell either lists
the states that could exceed the limits, or lists the number of states.

*** If the cell is yellow, this is the combination tonnage and
percentage limits that minimizes the sum of the differences between the
state-specific “alternative approach”  variability values and the
state-specific variability limit (in tons).Table A-3. The Effects of
Various Combinations of the Alternative Method Tonnage and Percentage
Variability Limits on Ozone Season EGU NOx Emissions (See Notes Below).

	Percentage Limit

	Tonnage Limit	0	2	4	6	8	10	12	14	16	18

800	20	20	17	13	9	5	AR, LA, MS, NY, 	AR, LA, MS, NY, 	LA, NY, 	55,182

1,100	12	12	11	8	6	AR, GA, LA, MS, 	AR, LA, MS, 	AR, LA, MS, 	LA, 
56,328

1,400	9	9	8	7	5	GA, LA, MS, 	LA, MS, 	LA, MS, 	LA, 	57,832

1,700	6	6	6	5	FL, GA, PA, 	GA, 	33,340	41,805	50,605	59,639

2,000	FL, GA, PA, TX, 	FL, GA, PA, TX, 	FL, GA, PA, TX, 	FL, GA, PA, TX,
	FL, GA, PA, 	GA, 	36,727	44,851	53,249	61,971

2,300	FL, PA, TX, 	FL, PA, TX, 	FL, PA, TX, 	FL, PA, TX, 	FL, PA, 
33,911	40,645	48,219	56,362	64,733

2,600	FL, PA, TX, 	FL, PA, TX, 	FL, PA, TX, 	FL, PA, TX, 	FL, PA, 
39,183	44,923	52,023	59,711	67,872

2,900	FL, PA, TX, 	FL, PA, TX, 	FL, PA, TX, 	FL, PA, TX, 	FL, PA, 
44,982	49,998	56,223	63,421	71,203

3,200	FL, PA, TX, 	FL, PA, TX, 	FL, PA, TX, 	FL, PA, TX, 	FL, PA, 
50,982	55,464	60,985	67,602	74,865

3,500	FL, 	FL, 	FL, 	FL, 	FL, 	56,982	61,304	66,146	71,973	78,980

3,800	60,374	60,374	60,374	60,374	60,812	62,982	67,304	71,746	77,072
83,180

4,100	67,874	67,874	67,874	67,874	68,012	69,455	73,304	77,625	82,342
88,059

4,400	75,374	75,374	75,374	75,374	75,374	76,272	79,304	83,625	88,027
93,159

4,700	82,874	82,874	82,874	82,874	82,874	83,471	85,431	89,625	93,947
98,609

5,000	90,374	90,374	90,374	90,374	90,374	90,671	92,031	95,625	99,947
104,309

5,300	97,874	97,874	97,874	97,874	97,874	97,874	98,930	101,625	105,947
110,269

5,600	105,374	105,374	105,374	105,374	105,374	105,374	106,130	108,007
111,947	116,269

5,900	112,874	112,874	112,874	112,874	112,874	112,874	113,330	114,651
117,947	122,269

6,200	120,374	120,374	120,374	120,374	120,374	120,374	120,530	121,589
124,016	128,269

6,500	127,874	127,874	127,874	127,874	127,874	127,874	127,874	128,789
130,583	134,269

6,800	135,374	135,374	135,374	135,374	135,374	135,374	135,374	135,989
137,291	140,269

7,100	142,874	142,874	142,874	142,874	142,874	142,874	142,874	143,189
144,249	146,560

7,400	150,374	150,374	150,374	150,374	150,374	150,374	150,374	150,389
151,449	153,160

7,700	157,874	157,874	157,874	157,874	157,874	157,874	157,874	157,874
158,649	159,931

8,000	165,374	165,374	165,374	165,374	165,374	165,374	165,374	165,374
165,849	166,908

8,300	172,874	172,874	172,874	172,874	172,874	172,874	172,874	172,874
173,049	174,108

8,600	180,374	180,374	180,374	180,374	180,374	180,374	180,374	180,374
180,374	181,308

8,900	187,874	187,874	187,874	187,874	187,874	187,874	187,874	187,874
187,874	188,508

9,200	195,374	195,374	195,374	195,374	195,374	195,374	195,374	195,374
195,374	195,708

9,500	202,874	202,874	202,874	202,874	202,874	202,874	202,874	202,874
202,874	202,908

9,800	210,374	210,374	210,374	210,374	210,374	210,374	210,374	210,374
210,374	210,374

10,100	217,874	217,874	217,874	217,874	217,874	217,874	217,874	217,874
217,874	217,874

10,400	225,374	225,374	225,374	225,374	225,374	225,374	225,374	225,374
225,374	225,374

10,700	232,874	232,874	232,874	232,874	232,874	232,874	232,874	232,874
232,874	232,874

11,000	240,374	240,374	240,374	240,374	240,374	240,374	240,374	240,374
240,374	240,374

11,300	247,874	247,874	247,874	247,874	247,874	247,874	247,874	247,874
247,874	247,874

11,600	255,374	255,374	255,374	255,374	255,374	255,374	255,374	255,374
255,374	255,374

11,900	262,874	262,874	262,874	262,874	262,874	262,874	262,874	262,874
262,874	262,874

12,200	270,374	270,374	270,374	270,374	270,374	270,374	270,374	270,374
270,374	270,374

12,500	277,874	277,874	277,874	277,874	277,874	277,874	277,874	277,874
277,874	277,874

12,800	285,374	285,374	285,374	285,374	285,374	285,374	285,374	285,374
285,374	285,374

13,100	292,874	292,874	292,874	292,874	292,874	292,874	292,874	292,874
292,874	292,874

13,400	300,374	300,374	300,374	300,374	300,374	300,374	300,374	300,374
300,374	300,374



*Numbers in white cells represent the sum (across all states) of the
differences in emissions between the state-specific “alternative
approach” variability values and the state-specific variability limit
(in tons); the variability limit selected for each state is the larger
of either the percentage limit or the tonnage limit.

=

€

ƒ

«

¬

Ë

Î

Ô

Õ

Ö

ç

ð

÷

ú

ü

Ð

,

-

.

4

>

E

F

b

~

´

Ã

Í

Î

Ï

Ð

h

h½

  h´

'

hæ

愀Ĥ摧检i਀

S

kdô

ì欀

y	k

#

y	k

#

y	k

#

y	k

#

y	k

#

y	k

#

y	k

#

y	k

#

y	k

#

y	k

#

y	k

#

y	k

#

hæ

hæ

hæ

晆멭᠀

愀Ȥ摧ξÚ᠀

hÛ

hú

hú

hú

h7

hü

h7

h7

yt‰

h‰

h‰

h‰

h‰

h‰

h‰

±	›

±	›

±	›

±	›

±	›

±	›

±	›

±	›

±	›

h‰

h‰

h‰

h‰

h‰

h‰

h‰

h‰

h‰

h‰

h‰

h‰

h o

h o

h o

愀Ĥ摧෈¹ᨀ

A

	

	

 



)

Ñ

Ý

	

b

%

6

a

b

d

e

f

g

s

ä

h o

F

h o

 h

h

h

h

 h

h

h

$

hÄ

kd³

kda

 hÄ

hÄ

hÄ

7

kd

 hÄ

hÄ

hÄ

hÄ

h

h»

h»

h»

h»

h»

h»

h»

(

(

(

(

(

(

(

(

(

(

(

(

(

(

(

(

(

hÄ

h

hÄ

h

h

hÄ

h’

h’

晆桥	ᄀ



 

s

t

u

v

z

¶

·

¸

¹

½

8

9

:

;

?

ˆ

‰

Š

‹

ß

à



 

 

-

:

:

G

P

W

^

e

l

s

t

z



„

‰

Ž

“

š

¡

¨

¯

¶

·

½

Ä

Ë

Ò

Ù

à

ç

ç

î

õ

ü

 

"

)

0

8

9

9

?

F

M

T

[

b

i

p

x

€

ˆ

‰

—

Ÿ

§

¯

·

¿

Ç

Ï

×

ß

à

æ

î

ö

à

á

â

æ

Eö

þ

Ff€

the state-specific “alternative approach”  variability values and
the state-specific variability limit (in tons).

   EPA developed annual SO2 and NOX budgets for each state covered for
the annual and/or 24-hour PM2.5 NAAQS.  Additionally, the EPA developed
ozone season NOX budgets for each state covered for the ozone NAAQS. 
Table III.A-1 in preamble section III lists the states that would be
covered for the PM2.5 and/or ozone NAAQS.

  As discussed in the preamble, for purposes of emissions reductions
requirements in the Transport Rule the EPA proposes to define the ozone
season as May through September.

 On a state-by-state basis, the standard deviation was calculated from
the set of yearly difference values.  As described, the yearly values
were the difference between the actual heat input and the estimated heat
input (using the regression equation) for each year.

 The two-tailed 95th percent confidence level is the equivalent of the
97.5th upper (single-tailed) confidence level.  Hereafter in this TSD,
the “95th“ percent confidence level refers to the two-tailed 95th
percent confidence level.  

 EPA developed the proposed variability limits in parallel with
developing the overall control requirements.  As such, while these IPM
runs used to develop the variability limits assumed reasonable levels of
emissions control, they are not identical to the final control strategy
chosen for the proposed rule.  Whereas IPM output files report
aggregated results for "model" plants (i.e., aggregates of generating
units with similar operating characteristics), parsed files show IPM
results at the generating unit level.  The IPM runs that are the bases
for the 2014 parsed files used to derive the state-specific rates are
designated “TR_SO2_1600”, “TR_NOX_500”, and “TR_NOX_OS_500”
for SO2, annual NOX, and ozone season NOX rates, respectively.  The IPM
runs and parsed files can be found in the docket; Docket ID No.
EPA-HQ-OAR-2009-0491.

  Tables 10, 12, and 14 also show resulting estimated emission
variability values based on the alternative approach (see columns
labeled “Alternative approach value”).

   Similar tables based on the alternative approach (where the max. heat
input over the time period was found relative to the average value over
the period) are in Appendix A.  

 EPA developed the proposed variability limits in parallel with
developing the overall control requirements.  As such, while the
particular “optimal” 1-year tonnage and percentage limits presented
here are close to the optimal values (presented in Tables 15, 16, and
17), they are not exact because during the development of the limits,
changes were made in the states covered by the proposed rule.  The IPM
modeling and CAMx air quality modeling for the proposed rule used these
values.  As discussed in the preamble, EPA requests comment on the
variability limits. 

 Moore, David S. and George P. McCabe. Introduction to the Practice of
Statistics. 2nd ed. New York: W.H. Freeman and Company, 1993. p.395.

 http://en.wikipedia.org/wiki/Variance

 The states included in this analysis were Alabama, Connecticut,
Delaware, District of Columbia, Florida, Georgia, Illinois, Indiana,
Iowa, Kansas, Kentucky, Louisiana, Maryland, Massachusetts, Michigan,
Minnesota, Missouri, Nebraska, New Jersey, New York, North Carolina,
Ohio, Pennsylvania, South Carolina, Tennessee, Virginia, West Virginia,
and Wisconsin.  

  PAGE  1 

