WEST Associates

1
WEST Associates Assessment of Hg MACT Floor
Variability

• CAAAC Mercury MACT
• Working Group
• Washington, DC
• March 4, 2003

2
Who Is WEST Associates?
AZ Arizona Electric Power Cooperative Pinnacle
West Capital Corp. Salt River Project Tucson
Electric Power Co. CA Glendale Public Service
Dept. Los Angeles Dept. of Water
Power Southern California Edison OR PacifiCorp ID
Idaho Power Company ND Basin Electric Power
NM Public Service Co of NM, Xcel
Energy Tri-State G T NV Nevada Power Co/
Sierra Pacific Power Co. CO Colorado Springs
Utilities Xcel Energy Platte River Power
Authority Tri-State G T UT PacifiCorp/Utah
Power and Light WY Basin Electric, PacifiCorp,
Xcel Energy Tri-State G T
3
WEST Associates efforts to-date

• September 2002
• Mercury Emissions from Western Coal-fired Power
  Plants Nature, Extent, and Fate
• Unique Western Concerns Related to the Role of
  Chlorine contents of Coal on Hg Emissions
• Recommended that MACT standard reflect these
  issues.
• Statistical Analysis to address coal chemistry
  issues

4
WESTs Hg MACTData Analysis Goals

• Determine from 80 unit source test ICR III
  database, statistically robust datasets for
  potential MACT subcategories (ENSR ANOVA Study)
• Coal rank (bit., sub-bit., lignite)
• Coal Hg content
• Hg/Cl ratio
• Develop a statistically valid approach to
  integrate operational variability using the ICR
  II fuel chemistry database to calculate Hg MACT
  floors

5
WESTs Hg MACTData Analysis Goals

• This study used the regulatory framework
• the average emission limitation achieved by the
  best performing twelve percent of existing
  sources and
• is achievable under the most adverse
  circumstances which can reasonably be expected to
  recur.

6
Variability in MACT Floor Determinations(Conceptu
al Illustration)
(Hg, Cl, Btu Content)
(Soot blowing, load following)t
7
80 Unit Source Test ICR III Does Not Capture
Variability

• Three 1-hr source tests are only a snapshot in
  time taken under steady state operations
• Three 1-hr source tests do not represent actual
  emissions over any longer operating time
• Based only on limited coal chemistry, and
  operational variability occurring during the 3
  tests
• ICR III source tests represent only a fraction of
  total variability

8
WESTs Multi-variable Method

• Uses ICR III source test and control
  effectiveness data from 12 best performing units
  by coal rank, plus annual coal chemistry data
  from ICR II
• Integrates key drivers of variability Coal Hg,
  Cl, Btu content (annual variability)
• Multi-variable Method is based on a 5 step
  statistical analytical process

9
Multi-variable Method5 Step Analytical Process

• STEP 1
• 80 source test units sorted by coal rank
• FBC units petroleum coke units combination fuel
  units removed (15 total)
• Leaves 29 bit. Units 26 subbit. units 10
  lignite units
• STEP 2
• Within each coal rank, units sorted in ascending
  order of stack tested Hg emissions ( Hg/TBtu)
• Best performing 12 of units the 5 units with
  lowest emissions in each coal rank
• Note Significant differences occur in averages
  of Cl (ppm) and Hg (/TBtu) between coal ranks.

10
Multi-variable Method5 Step Analytical Process

• STEP 3
• To account for intra-unit variability,
  correlation equations were developed to relate Hg
  emissions to coal chlorine content
• For each control configuration (e.g., FF/SDA,
  etc.) determined relationship between Hg removal
  and coal Cl concentration using ICR III stack
  test database for all tested units (not only the
  best performing units)

11
Multi-variable Method5 Step Analytical Process

• STEP 3 (cont.)

Figure 1
12
Multi-variable Method5 Step Analytical Process

• STEP 4
• For each best performing unit, controlled Hg
  emissions calculated by multiplying
  uncontrolled Hg emissions by (1-Hg removal
  fraction)
• ICR II test data (Btu and Hg content) used to
  calculate uncontrolled emissions
• Hg removal fraction derived in one of two ways
• If good correlation (from step 3), correlation
  equation used to calculate Hg removal fraction
• If poor correlation, ICR III source test Hg
  removal fraction used
• Process repeated for each set of measured coal
  composition data from ICR II database (I.e., Hg,
  Btu and Cl measurements) yielding a range of Hg
  emissions for each unit over time

13
Variability in Coal Hg Content
14
Multi-variable Method5 Step Analytical Process

• STEP 5
• For each best performing unit, calculated mercury
  emissions sorted from smallest to largest to
  obtain a frequency distribution
• 95 value of this distribution assumed to
  represent the operation of the unit under the
  most adverse circumstances reasonably expected to
  recur for each unit
• The 95 upper confidence limit (UCL) of the mean
  of these adverse-case emissions is reported as
  the Hg MACT floor

15
Multi-variable Method5 Step Analytical Process

• STEP 5 (cont)

16
Elements of VariabilityNot Captured by this
Method

• Analysis of fuel variability accounts for some,
  but not all, of the variability in the stack
  testing of each unit in ICR III
• Stack test measurement error (/- 20-25)
• Intermittent maintenance events (e.g., operation
  of air heater soot blowers) affect Hg emission
  rates
• Source tests conducted at static load load
  following can change results

17
Results of Multi-variable Hg MACT Floor Method
Coal Rank MACT Floor (lb Hg/TBtu)
Bituminous 2.26
Subbituminous 5.75
Lignite 10.15

• Potential national Hg reduction 15 t/yr 31

18
Statistical Rationale for Alternate MACT Floors

• Could replace Coronado with Comanche in list of
  top 5 best performing sub bituminous plants
• Hg rate for Coronado is only 6 less than
  Comanche
• Measured Hg removal data for Comanche show much
  less scatter than data for Coronado
• The removal for Coronado was found to be
  negative for all 3 source tests
• Could use simple average of top 5 best performing
  lignite units (5 out of 10 units).
• Need for 95 UCL for inter-unit variability among
  10 units is less

19
Alternate Hg MACT Floorsfor Subbituminous
Lignite
Coal Rank MACT Floor (lb Hg/TBtu)
Bituminous 2.26
Subbituminous 4.15
Lignite 8.20

• Potential national Hg reduction 17 t/yr 36

20
Conclusions of Multi-variable Method Hg MACT
Floor Study

• Multi-variable method uses the maximum amount of
  information from both ICR II and ICR III
  databases in the determination of variability in
  a MACT floor
• First known study to comprehensively bridge
  between the ICR III source test, and ICR II
  annual coal chemistry data
• Our MACT Floor levels represent statistically
  robust estimates of the variability of Hg
  emissions as a result of annual variability of
  coal chemistry
• Variability of coal chemistry accounts for only
  one driver of variability. The MACT Floor
  results likely underestimate most adverse
  circumstances which can reasonably be expected to
  recur at a unit meeting a mercury MACT limit.
• This technical analysis conforms to regulatory
  requirements

21
Supplemental Slides

• Additional
• Correlation Equations

22
Multi-variable Method5 Step Analytical Process

• STEP 3 (cont.)

Figure 2
23
Multi-variable Method5 Step Analytical Process

• STEP 3 (cont.)

Figure 3
24
Multi-variable Method5 Step Analytical Process

• STEP 3 (cont.)

Figure 4
25
Multi-variable Method5 Step Analytical Process

• STEP 3 (cont.)

Figure 5