# Approval and Promulgation of Implementation Plans; New Mexico; Federal Implementation Plan for Interstate Transport of Pollution Affecting Visibility and Best Available Retrofit Technology Determination

> Briefs, arguments, decisions, and more.

URL: https://www.frixlaw.com/law-library/documents/fr%3A2011-20682

## Record

- **Collection:** Federal Register
- **Document type:** Rule
- **Published:** August 22, 2011
- **Citation:** 76 FR 52388

## Text

ENVIRONMENTAL PROTECTION AGENCY
40 CFR Part 52
EPA-R06-OAR-2010-0846; FRL-9451-1
Approval and Promulgation of Implementation Plans; New Mexico; Federal Implementation Plan for Interstate Transport of Pollution Affecting Visibility and Best Available Retrofit Technology Determination

AGENCY:

Environmental Protection Agency (EPA).

ACTION:

Final rule.

SUMMARY:

EPA is disapproving a portion of the State Implementation Plan (SIP) revision received from the State of New Mexico on September 17, 2007, for the purpose of addressing the “good neighbor” requirements of section 110(a)(2)(D)(i) of the Clean Air Act (CAA or Act) for the 1997 8-hour ozone National Ambient Air Quality Standards (NAAQS or standards) and the 1997 fine particulate matter (PM
2.5
) NAAQS. In this action, EPA is disapproving the New Mexico Interstate Transport SIP provisions that address the requirement of section 110(a)(2)(D)(i)(II) that emissions from New Mexico sources do not interfere with measures required in the SIP of any other state under part C of the CAA to protect visibility. We have found that New Mexico sources, except the San Juan Generating Station, are sufficiently controlled to eliminate interference with the visibility programs of other states. EPA is promulgating a Federal Implementation Plan (FIP) to address this deficiency by implementing nitrogen oxides (NO
X
) and sulfur dioxide (SO
2
) emission limits necessary at the San Juan Generating Station (SJGS), to prevent such interference. EPA found in January 2009 that New Mexico had failed to submit a SIP addressing certain regional haze (RH) requirements, including the requirement for best available retrofit technology (BART). The Clean Air Act required EPA to promulgate a FIP to address RH requirements by January 2011. This FIP addresses the RH BART requirement for NO
X
for SJGS. In addition, EPA is implementing sulfuric acid (H
2
SO
4
) hourly emission limits at the SJGS, to minimize the contribution of this compound to visibility impairment. This action is being taken under section 110 and part C of the CAA.

DATES:

This final rule is effective on: September 21, 2011.

ADDRESSES:

EPA has established a docket for this action under Docket ID No. EPA-R06-OAR-2010-0846. All documents in the docket are listed in the Federal eRulemaking portal index at
http://www.regulations.gov
and are available either electronically at
http://www.regulations.gov
or in hard copy at EPA Region 6, 1445 Ross Ave., Dallas, TX 75202-2733. To inspect the hard copy materials, please schedule an appointment during normal business hours with the contact listed in the
FOR FURTHER INFORMATION CONTACT
section. A reasonable fee may be charged for copies.

FOR FURTHER INFORMATION CONTACT:

Joe Kordzi, EPA Region 6, (214) 665-7186,
kordzi.joe@epa.gov.

SUPPLEMENTARY INFORMATION:

Throughout this document wherever “we,” “us,” “our,” or “the Agency” is used, we mean the EPA. Unless otherwise specified, when we say the “San Juan Generating Station,” or “SJGS,” we mean units 1, 2, 3, and 4, inclusive.

Overview

The Clean Air Act requires states to prevent air pollution from sources within their borders from impairing air quality and visibility in other states. The Act also requires states to reduce pollution from significant sources whose emissions reduce visibility in the nation's pristine and wilderness areas (such as the Grand Canyon), and contribute to regional haze. When a state has not adopted plans as required by these provisions, EPA must put such a plan in place, known as a Federal Implementation Plan (FIP).

In this action, EPA is finalizing a FIP for New Mexico to address emissions from one source: the San Juan Generating Station coal-fired power plant. EPA is finding that the other New Mexico pollution sources are adequately controlled to eliminate interference with the clean air visibility programs of other states. This FIP can be replaced by a state plan that EPA finds meets the applicable Clean Air Act requirements. The federal plan will remain in effect no longer than necessary.

In December 2010, EPA proposed to disapprove a portion of the New Mexico Interstate Transport State Implementation Plan (SIP), specifically the New Mexico Interference with Visibility SIP, and proposed a source-specific FIP to cut pollution from San Juan Generating Station to address adverse visibility impacts.

The federal plan also addresses a portion of EPA's 2-year obligation under the Clean Air Act's Regional Haze Rule to implement a federal plan when the state failed to meet the January 2009 deadline. This shortfall is being addressed by establishing emissions limits representing Best Available Retrofit Technology (BART) for nitrogen oxide (NOx) pollution at the San Juan Generating Station power plant.

The federal plan will require the San Juan Generating Station to cut emissions to improve scenic views at 16 of our most treasured parks including the Grand Canyon, Mesa Verde and Bandelier National Monument. Pollution from this power plant impacts four states including Arizona, Utah, Colorado, and New Mexico. Improved air quality also results in public health benefits.

Public Service Company of New Mexico (PNM) owns the San Juan Generating Station power plant. The power plant has four coal-fired generating units. It is located in San Juan County, 15 miles west of Farmington in northwest New Mexico. The thirty-year-old San Juan Generation Station power plant is one of the largest sources of NOx pollution in the United States.

The federal plan requires the San Juan Generating Station coal-fired power plant to reduce nitrogen oxide and sulfur dioxide pollution to 0.05 pounds per million BTU and 0.15 pounds per million BTU respectively.

By addressing nitrogen oxide pollution requirements of both Interstate Transport and the Regional Haze Rule, PNM will meet these two Clean Air Act requirements for NOx emission limits for the power plant with only one round of improvements. This regulatory certainty will help guide PNM's business decisions regarding capital investments in pollution controls.

EPA evaluated reliable and proven pollution technologies as part of its decision. EPA determined Selective Catalytic Reduction (SCR) to be the most cost-effective pollution control to achieve the emission reductions outlined in the federal plan. Evaluation of a less expensive alternative, Selective Non Catalytic Reduction (SNCR), showed that SNCR at the San Juan Generating Station coal-fired power plant achieves far less reduction in pollution and less visibility improvement, and does not fully meet the requirement of the Act for Best Available Retrofit Technology (BART).

EPA held an extended public comment period on this action, an open house, and a public hearing. After careful review of information provided during the public comment period, EPA revised its calculation of the associated cost investment from $229 million to $345 million. Also, in consideration of comments about the time to comply with the new emissions limits, EPA

extended the time for compliance with the nitrogen oxide pollution emission limit from 3 years to 5 years, the maximum period allowed by the Clean Air Act.

This investment will reduce the visibility impacts due to this facility by over 50% at each one of the 16 national parks and wilderness areas in the area, and promote local tourism by decreasing the number of days when pollution impairs scenic views. Although today's action is taken to address visibility impairments, PNM will also reduce public health impacts by cutting NO
X
pollution by over 80% by installing reliable pollution-control technology on its four coal-fired power generation units over the next five years.

EPA will review the regional haze plan that the State submitted in July 2011, and if there is significant new information that changes our analysis, EPA will make appropriate revisions to today's decision.

Detailed Outline

I. Summary of Our Proposal

II. Final Decision

A. Interstate Transport

B. NO
X
BART Determination for the San Juan Generating Station (SJGS)

C. Compliance Timeframe

III. Analysis of Major Issues Raised by Commenters

A. Comments on the Costs of the NO
X
BART Determination

B. Comments on our Proposed NO
X
BART Emission Limits

C. Comments on our Proposed SO
2
Emission Limit

D. Comments on our Proposed H
2
SO
4
and Ammonia Emission Limits and Other Pollutants

E. Comments on the Emission Limit Compliance Schedule

F. Comments on the Conversion of the SJGS to a Coal-to-Liquids Plant With Carbon Capture as a Means of Satisfying BART

G. Comments on Health and Ecosystem Benefits, and Other Pollutants

H. Miscellaneous Comments

I. Comments in Favor of Our Proposal

J. Comments Arguing Our Proposal Would Hurt the Economy and/or Raise Electricity Rates

K. Comments Arguing Our Proposal Would Help the Economy

L. Comments Requesting an Extension to the Public Comment Period

M. Comments Requesting We Defer Action in Favor of a New Mexico SIP Submittal

N. Comments Generally Against Our Proposal

O. Comments on Legal Issues

P. Modeling Comments

IV. Statutory and Executive Order Reviews

I. Summary of Our Proposal

On January 5, 2011, we published the proposal on which we are now taking final action. 76 FR 491. We proposed to disapprove a portion of the SIP revision received from the State of New Mexico on September 17, 2007, for the purpose of addressing the “good neighbor” provisions of the CAA section 110(a)(2)(D)(i) with respect to visibility for the 1997 8-hour ozone NAAQS and the PM
2.5
NAAQS. Having proposed to disapprove these provisions of the New Mexico SIP, we proposed a FIP to address the requirements of section 110(a)(2)(D)(i)(II) with respect to visibility to ensure that emissions from sources in New Mexico do not interfere with the visibility programs of other states. We proposed to find that New Mexico's sources, other than the San Juan Generating Station (SJGS), are sufficiently controlled to eliminate interference with the visibility programs of other states, and for the SJGS, we proposed specific SO
2
and NO
X
emissions limits that will eliminate such interstate interference. For SO
2
, we proposed to require the SJGS to meet an emission limit of 0.15 pounds per million British Thermal Units (lb/MMBtu). For NO
X
, we proposed to implement a NO
X
emission limit of 0.05 lbs/MMBtu, based on our BART determination, as discussed below.

Separate from our proposal under Section 110 of the CAA, we simultaneously evaluated whether the SJGS met certain other related requirements under the Regional Haze (RH) program under Sections 169A and 169B of the CAA. Regional Haze SIPs were due December 17, 2007. In January 2009, we made a finding that New Mexico had failed to submit a RH SIP addressing the requirements of 40 CFR 51.309(d)(4) and (g). 74 FR 2392 (January 15, 2009). Under the CAA, we are required to promulgate a FIP within two years of the effective date of a finding that a State has failed to submit a SIP unless the State submits a SIP and we approve that SIP within the two year period. CAA § 110(c). At the time of the proposed FIP, New Mexico had not yet submitted a substantive RH SIP addressing, among other things, the requirement that certain stationary sources install BART for NO
X
. (On July 5, 2011, New Mexico submitted a RH SIP, which we discuss later in this Notice.) Based on our evaluation of the RH BART requirements of section 40 CFR 51.309(d)(4), we proposed to find that the SJGS is subject to BART under section 40 CFR 51.309(d)(4), and/or 51.308(e). We proposed a FIP which contained NO
X
BART limits for the SJGS based on our proposed NO
X
BART determination. We proposed to require that the SJGS meet a NO
X
emission limit of 0.05 lb/MMBtu individually at Units 1, 2, 3, and 4. We noted this NO
X
limit is achievable by installing and operating Selective Catalytic Reduction (SCR).

We proposed that both the NO
X
and SO
2
emission limits be measured on the basis of a 30 day rolling average. We also proposed hourly average emission limits of 1.06 × 10
−4
lb/MMBtu for H
2
SO
4
and 2.0 parts per million volume dry (ppmvd) ammonia adjusted to 6 percent oxygen, to minimize the contribution of these compounds to visibility impairment. We solicited comments on a range of 2-6 ppmvd for ammonia, and 1.06 × 10
−4
to 7.87 × 10
−4
lb/MMBtu for H
2
SO
4
. Additionally, we proposed monitoring, record-keeping and reporting requirements to ensure compliance with these emission limitations.

Lastly, we proposed that compliance with the emission limits must be within three (3) years of the effective date of our final rule. We solicited comments on alternative timeframes, up to five (5) years from the effective date our final rule. In our proposal, we did not address whether the state had met other requirements of the RH program, which we will address in later actions. Please see our proposal for more details.

II. Final Decision

A. Interstate Transport

We are disapproving the portion of the SIP revision received from the State of New Mexico on September 17, 2007, for the purpose of addressing the “good neighbor” provisions of the CAA section 110(a)(2)(D)(i) with respect to visibility for the 1997 8-hour ozone NAAQS and the PM
2.5
NAAQS. The 2007 SIP submission by New Mexico anticipated that the State would submit a substantive RH SIP to meet the requirements of section 110(a)(2)(D)(i)(II).

Section 110(a)(2)(D)(i)(II) of the CAA requires that states have a SIP, or submit a SIP revision, containing provisions “prohibiting any source or other type of emission activity within the state from emitting any air pollutant in amounts which will * * * interfere with measures required to be included in the applicable implementation plan for any other State under part C [of the CAA] to protect visibility.” States were required to submit a SIP by December 2007 with measures to address regional haze—visibility impairment that is caused by the emissions of air pollutants from numerous sources located over a wide geographic area. Under the RH program, each State with a Class I area must submit a SIP with reasonable progress goals for each such area that provides for an improvement in visibility for the

most impaired days and ensures no degradation of the best days. (The “Class I” federal areas
1

affected by the SJGS include 16 of our most treasured parks, such as the Grand Canyon, Mesa Verde, and Bandelier National Monument. Emissions from this power plant impact four states including Arizona, Utah, Colorado, and New Mexico.)

1
CAA 42 U.S.C. 7472(a). The list of mandatory class I federal areas where visibility is an important value is codified at 40 CFR part 81 subpart D.

Because of the often significant impacts on visibility from the interstate transport of pollutants, we interpret the “good neighbor” provisions of section 110 of the CAA described above as requiring states to include in their SIPs measures to prohibit emissions that would interfere with the reasonable progress goals set to protect Class I areas in other states. This is consistent with the requirements in the RH program which explicitly require each State to address its share of the emission reductions needed to meet the reasonable progress goals for surrounding Class I areas. 64 FR 35714, 35735 (July 1, 1999). States working together through a regional planning process are required to address an agreed upon share of their contribution to visibility impairment in the Class I areas of their neighbors. 40 CFR 51.308(d)(3)(ii).

The States in the West, including New Mexico, worked through a regional planning organization, the Western Regional Air Partnership (WRAP), to develop strategies to address regional haze. To help the State in establishing reasonable progress goals, the WRAP modeled future visibility conditions. The WRAP modeling assumed emissions reductions from each State, based on extensive consultation among the States as to appropriate strategies for addressing haze. In setting reasonable progress goals, States in the West generally relied on this modeling. As explained in the notice of proposed rulemaking, we believe that the analysis conducted by the WRAP provides an appropriate means for designing a FIP that will ensure that emissions from sources in New Mexico are not interfering with the visibility programs of other states, as contemplated in section 110(a)(2)(D)(i)(II).

As a result of our disapproval of New Mexico's SIP, submitted to meet the requirements of section 110(a)(2)(D)(i)(II) with respect to visibility, we are promulgating a FIP to ensure that emissions from New Mexico sources do not interfere with the visibility programs of other states. We find that New Mexico sources, other than the SJGS, are sufficiently controlled to eliminate interference with the visibility programs of other states because the federally enforceable emission limits for these sources are consistent with those relied upon in the WRAP modeling. The SO
2
and NO
X
emissions relied upon in the WRAP modeling for the SJGS, however, are not federally enforceable. Therefore, we are establishing federally enforceable SO
2
emissions limits that will address these discrepancies and eliminate interstate interference based on current emissions that satisfy the assumptions in the WRAP modeling. We are finalizing our proposal to require the SJGS to meet an SO
2
emission limit of 0.15 lb/MMBtu, the rate assumed in the WRAP modeling. We proposed a 30 day rolling average for units 1, 2, 3, and 4 of the SJGS. However, in response to a comment we received, we are changing our proposed averaging period for these emission limits from a straight 30 day calendar average to one calculated on the basis of a Boiler Operating Day (BOD).

Besides not being federally enforceable, the NOx emissions that were assumed in the WRAP modeling cannot be achieved without additional NOx controls for the SJGS to prevent interference with visibility pursuant to the requirements of section 110(a)(2)(D)(i)(II) of the CAA. We are choosing, however, not to use the WRAP assumptions to make a determination on the enforceable NOx controls necessary to prevent visibility interference, as we are doing for the SO
2
controls. Instead, we are addressing NOx control for the SJGS by fulfilling our duty under the BART provisions of the RH rule to promulgate a RH FIP for New Mexico to address, among other elements of the visibility program, the requirement for BART.
2

We do not believe it is prudent to delay a NOx BART determination for the SJGS, because we have determined that the BART requirements are more stringent than the visibility transport requirements. Separating the visibility transport and BART rulemakings could result in near-term requirements for the utility to install one set of controls and capital expenditures, to only satisfy our obligation under section 110(a)(2)(D)(i)(II), followed shortly thereafter by different requirements for controls and capital expenditures to satisfy our obligation under BART. This could result in unnecessary costs and confusion.

2

See
74 FR 2392.

We did receive a New Mexico RH SIP submittal on July 5, 2011, but it came several years after the statutory deadline, and after the close of the comment period on today's action.
3

In addition, because of the missed deadline for the visibility transport, we are under a court-supervised consent decree deadline with WildEarth Guardians of August 5, 2011, to have either approved the New Mexico SIP or to have implemented a FIP to address the 110(a)(2)(D)(i) provision. It would not have been possible to review the July 5, 2011 SIP submission, propose a rulemaking, and promulgate a final action by the dates required by the consent decree. Notwithstanding these facts, we did comment during the State's public comment period for their proposed RH SIP in May 2011 and we did evaluate the technology advocated as BART in the State's proposed RH SIP: SNCR, as discussed in further detail elsewhere in this Notice.

3
A State Regional Haze SIP was due under the CAA by Dec. 17, 2007, and EPA was obligated to either approve an RH SIP or promulgate a FIP by January 15, 2011.
See
CAA Section 110(c)(1)(B).

B. NOx BART Determination for the San Juan Generating Station (SJGS)

We find that the SJGS is subject to BART under sections 40 CFR 51.309(d)(4), and/or 51.308(e). In this action, we are adopting a FIP that partially addresses the BART requirements of the RH program for New Mexico. We are finalizing our proposal to require the SJGS to meet a NOx emission limit of 0.05 lb/MMBtu individually at Units 1, 2, 3, and 4. As we discuss elsewhere in our response to comments, we find there is ample support for this decision. However, in response to a comment we received, we are changing our proposed averaging period for these emission limits from a straight 30 day calendar average to one calculated on the basis of a boiler operating day (BOD). We also received a comment requesting we revise our proposed unit-by-unit NOx limitation, and replace it with a plant wide average NOx limitation. As we note in our response to this comment, although we are open to combining the BOD and plant wide averaging schemes, this presents a significant technical challenge in having a verifiable, workable, and enforceable algorithm for calculating such an average. Due to our obligation to ensure the enforceability of the emission limits we are imposing in our FIP, we leave it to New Mexico to take up this matter in a future SIP revision, should they deem it worth pursuing. We are confident this issue

can be addressed prior to the installation of the emission controls required to satisfy our FIP.

We are also finalizing our proposal requiring the SJGS to meet an H
2
SO
4
emission limit of 2.6 × 10
−4
lb/MMBtu to minimize its contribution to visibility impairment. We are promulgating monitoring, record-keeping and reporting requirements to ensure compliance with this emission limit. As discussed in our response to comments, after careful consideration of the comments we received concerning our proposal to require the SJGS to meet an hourly average emission limit of 2.0 parts ppmvd for ammonia, we have determined that neither an ammonia limit, nor ammonia monitoring is warranted, and we are not finalizing ammonia limits or monitoring requirements.

C. Compliance Timeframe

We originally proposed a compliance schedule of 3 years for SJGS for the NO
X
, SO
2
, ammonia, and H
2
SO
4
emission limits, and solicited comments on alternative timeframes of less than 3 years and up to 5 years (the maximum allowed under the statute).
4

As noted above, we are no longer requiring an ammonia emission limit. Also, as discussed in our response to comments, we carefully considered comments urging a longer compliance schedule due to site-specific issues such as the congestion of existing equipment (which could slow the retrofit process), historical information on SCR installation times, and our own observation of the site conditions,
5

and we now conclude that a longer compliance schedule is more appropriate. Consequently, compliance with the NO
X
, SO
2
, and H
2
SO
4
emission limits will now be required within 5 years—rather than 3 years—of the effective date of our final rule. (This issue is discussed in further detail in Section III.E., below.)

4
76 FR 491, 504.

5
See San Juan Generating Station Site Visit, 5/23/11, which is viewable in the docket. As explained in a letter, dated May 17, 2011, the visit was solely for the purpose of reviewing and responding to comments. It was not an opportunity to introduce additional comments, and we did not receive any comments as a result of this visit.

III. Analysis of Major Issues Raised by Commenters

Our January 5, 2011 proposal included a 60 day public comment period, which ended on March 7, 2011. We subsequently extended that comment period until April 4, 2011.
6

We also held an open house and a public hearing in Farmington, NM, on February 17, 2011.
7

We received in excess of 13,000 comments.

6
76 FR 12305.

7
76 FR 1578.

In light of the very large number of comments received and the significant overlap between many comments, we have grouped some comments together. We have summarized and provided responses to each significant argument, assertion, and question contained within the totality of the comments. Full responses to comments can be found in our
Complete Response to Comments for NM Regional Haze/Visibility Transport FIP.

A. Comments on the Costs of the NO
X
BART Determination

We received many comments related to various aspects of our cost analysis that fell into four major categories. First, we received general comments opining on the appropriateness of our cost analysis. Second, we received comments that were technical and related to specific line items in the cost analysis (
e.g.,
additional steel, SCR bypass, sorbent injection,
etc.
). Third, we received comments that expressed general concern that the costs of the controls would be passed to the SJGS's customer base in the form of electricity rate increases. Fourth, we received comments that opined on the use of the Regional Haze Rule's (RHR) reliance on the EPA Air Pollution Control Cost Manual (the Cost Manual) to estimate the cost of the SCR installations. We address the more significant comments within these categories individually below.

1. General Cost Comments

Comment:
The National Park Service (NPS) and the U.S. Forest Service (USFS) separately presented a great deal of information in support of their opinions that Public Service Company of New Mexico's (PNM) contractor, Black &Veatch (B&V) overestimated the cost of installing SCR on the units of the SJGS. PNM is a part owner and the operator of the SJGS. The following is a combined summary of their separate comments.

The NPS and the USFS cited a large number of well-documented recent industry studies or surveys, which they use to conclude that PNM has overestimated its SCR costs, expressed in dollars per kilowatt. They stated that PNM has not provided valid information to justify their higher cost estimates for SCR installation at the SJGS. Additionally, the USFS stated PNM's contractors went against our guidance which recommends using the Cost Manual to ensure a transparent and consistent means to conduct cost analyses across the nation. The USFS took issue with PNM's estimation of indirect (soft) costs which include: engineering costs; construction and field expenses (
e.g.,
costs for construction supervisory personnel, office personnel, rental of temporary offices,
etc.
); contractor fees; and start-up and performance test costs. Also, the NPS stated that B&V's improperly escalated costs and its calculations did not consider the weakening of labor markets that has occurred since they set up their spreadsheets in 2007.

Response:
We found that PNM raised some legitimate points about costs, and as discussed elsewhere in this notice, we have adjusted several of our cost estimates upward based on those points. However, in large part, we agree with the NPS that PNM's estimated costs for installing SCR on the units of the SJGS are higher than justified. Please see our other responses to comments for more details on how we have adjusted our cost estimates. The following table illustrates our revised costs in terms of $/kW. These costs agree with the ranges presented by the NPS and the USFS in their comments, which can be viewed in our
Complete Response to Comments for NM Regional Haze/Visibility Transport FIP
document:

Table 1—EPA Revised Estimated Costs of Installing SCR on the Units of the SJGS

Unit 1
Unit 2
Unit 3
Unit 4

Proposed ($/kW)
$144
$155
$116
$110

Final ($/kW)
211
234
179
165

We note, that as required by the BART Guidelines, “[i]n order to maintain and improve consistency, cost estimates should be based on the
OAQPS Control Cost Manual,
[now renamed “EPA Air Pollution Control Cost Manual, Sixth Edition, EPA/452/B-02-001, January 2002] where possible.” 70 FR at 39166 (July 6, 2005). As explained more fully in our
Complete Response to Comments for NM Regional Haze/Visibility Transport FIP
document, we also agree with the USFS that owner's costs are not an appropriate cost item to include in a BART cost estimate, as owners costs are not included in the Cost Manual.

Comment:
PNM and its consultants estimated the cost of retrofitting SJGS with SCRs to be between $194 million and $261 million per unit (depending on the unit) with a total cost of $908 million for all four units. EPA maintains that SCRs can be purchased and installed for much less—between $52 million and $63 million per unit for a total of about $229 million. EPA's estimates of annual operating costs for the SCRs are also much lower than PNM's estimate. PNM's analysis indicates annual operating costs for all four SCRs would be approximately $114 million per year, whereas EPA expects PNM to be capable of operating the SCRs for only about $28 million per year. In short, EPA believes that SCRs cost $679 million less, or one quarter of the amount estimated by PNM. The commenter calls our cost estimate into question, since the disparity between these two estimates is large.

Response:
B&V estimated it would cost between $446/kW and $559/kW to retrofit SCR on the SJGS units. Five industry studies conducted between 2002 and 2007 have reported the installed unit capital cost of SCRs to be $79/kW to $316/kW, where the upper end of the range is for very complex retrofits that are severely site constrained.
8

Others have noted the anomalously high costs reported for SJGS.
9 10

We revised our cost estimates based on some comments highlighted in comments, but even with those changes, our revised costs for SCR are from $165/kW to $234/kW,
11

still well within the accepted range of expected costs for such controls.
12

8
Revised BART Cost Effectiveness Analysis for Selective Catalytic Reduction at the Public Service Company of New Mexico San Juan Generating Station, November 2010, pp. 28-29.

9
Comments submitted by United States Department of Interior, National Park Service, dated 3/31/11.

10
New Mexico Environment Department, Appendix A, NMED, Air Quality Bureau, BART Determination, Public Service Company of New Mexico, San Juan Generating Station, Units 1-4, 6/21/10.

11
See Exhibit 1, RTC Revised Cost Analysis.

12
Please see our Complete Response to Comments for NM Regional Haze/Visibility Transport FIP document.

B&V's SJGS costs are unusually high for four principal reasons: (1) Using a methodology (
e.g.,
Allowance for Funds Used During Construction (AFUDC)) that has been disallowed under EPA”s Cost Manual methodology and specifically disallowed for SCR (see discussion at footnote 28); (2) consistently using assumptions at the upper end of the range for key SCR components (
e.g.,
SCR backpressure; stiffening design pressure); (3) including costs for equipment that is not necessary for a SCR (
e.g.,
balanced draft conversion, sorbent injection, SCR bypass); and (4) using excessive contingencies. The BART Guidelines require that “documentation” be provided for “any unusual circumstances that exist for the source that would lead to cost-effectiveness estimates that would exceed that for recent retrofits.”
13

The B&V analysis does not support its unusually high cost estimates.

13
70 FR at 39168 (July 6, 2005).

Further, much of the information that could have supported a claim that site specific issues at SJGS result in costs that are outside of the normal range is missing. Specifically, the B&V analysis lacked information such as project schedules, general arrangement site plans showing SCR and duct layout, requests for proposal (RFPs), vendor proposals, and a complete description of existing facilities.

Instead of preparing a site-specific SCR design, B&V in most circumstances made a worst case, upper bound assumption that, taken together, result in overall costs that are significantly outside of the normal range for SCR. However, B&V provided no record support for their decision to choose the upper end of the range for nearly every aspect of the cost of SCRs. It is unlikely that so many upper bound assumptions could be justified, and if B&V believed that they were justified, they should have explored that proposition in a risk analysis. Therefore, we believe that our approach to considering site specific conditions that would lead to costs outside of the normal range, is justified.

Comment:
Private citizens submitted comments that the costs to PNM will be, alternatively, $250, $500, or $750 million dollars, and that PNM's estimates are overstated, and that any investment in the plant is an investment in the future, and that the plant and its jobs will not be threatened by the proposed emission reductions.

Response:
As we discuss elsewhere in our response to comments, we agree that the cost of installing SCR on the four units of the SJGS is considerably lower than PNM estimated.

Comment:
The CAA visibility provisions, EPA's own RH regulations, and the preambles to those rules all envision a “source-by-source” approach to BART, which by its nature must account for site-specific challenges at each facility. However, despite the significant amount of information provided by PNM in its original BART analysis, in subsequent exchanges with the New Mexico Environment Department (NMED) and EPA, and in meetings between EPA and PNM specifically to discuss the site-specific challenges at SJGS, EPA did not to take into account many of the most significant costs that are essential in calculating an accurate cost estimate of installing SCRs at SJGS.

Response:
We agree that a source-by-source analysis is appropriate, but we do not believe that B&V provided an acceptable analysis. First, the B&V costs were extrapolated from other facilities, based on confidential information that was not provided in response to our requests. Second, the B&V costs were estimated using worst-case upper bounds in lieu of making a site-specific estimate, as discussed above. Third, their costs included components that are not required at this site, and further assumed contingency factors beyond those normally expected. Therefore, we believe, with the exception of certain issues related to site congestion that are addressed separately in other comments, site-specific conditions were properly considered.

Comment:
To justify the approach based entirely on the median of different control technologies, EPA downplays the complicated process of designing and constructing an SCR, thereby not only ignoring the technology itself, but also the site specific-factors that must be considered at SJGS. SCRs at SJGS would have to be constructed so that each SCR can be positioned at the proper point in the flue gas stream, which will significantly complicate the foundation and supports that will be needed, resulting in additional costs of $35,630,000 that EPA failed to recognize or consider.

Response:
All SCRs have to be constructed so that each SCR can be positioned at the proper point in the flue gas stream, with proper foundation and supports; this is not unique to the SJGS. Over 300 retrofit SCRs have been installed since the early 1990s in the

United States. Accordingly, constructability issues are well understood. Standard design and construction management methods have been developed from these 300+ existing installations.
14

This experience would inform the design and construction of the SJGS SCR, resulting in significant economies compared to the estimates presented by B&V based on a very rough preliminary design that has not been optimized for constructability. The record does not identify any unusual site-specific conditions that would result in direct installation costs for SJGS that are substantially higher than upper bound direct installation costs reported by other SCR design firms for similarly complex sites. In fact, B&V has provided no support in the record for its assumptions. Finally, the design costs are not a direct installation cost, but rather indirect costs discussed elsewhere in our response to comments.

14
J.A. Hines and others, Design for Constructability—A Method for Reducing SCR Project Costs, Mega, 2001, available at:
http://www.babcock.com/library/pdf/br-1720.pdf; see also
Institute of Clean Air Companies (ICAC), White Paper, Selective Catalytic Reduction (SCR) Control of NO
X
Emissions from Fossil Fuel-Fired Electric Power Plants, May 2009, EPA-R09-OAR-2009-0598-0032 and Walter Nischt and others, Update of Selective Catalytic Reduction Retrofit on a 675 MW Boiler at AES Somerset, ASME International Joint Power Generation Conference, July 24-25, 2000, available at:
http://www.babcock.com/library/pdf/br-1703.pdf.

Comment:
EPA suggests that the engineering needed to design four SCRs can be completed all at the same time, thus saving time and money. While some economies may arise with a multiple SCR installation, as lessons learned in designing and installing one SCR are applied to the next, a three-year deadline would require PNM to design all four SCRs at the same time. Designing all four SCRs at once would require four separate design and construction teams, which would eliminate the opportunity to apply any experience gained. As a result, the costs associated with designing the SCRs will be much higher on a shorter timeframe, not lower as EPA appears to suggest. The short, three-year deadline also allows no time for additional design work that may be needed to address unforeseen engineering challenges that are likely to arise at each unit.

Response:
We disagree with this comment and believe it mischaracterizes our analysis. In our proposal, we simply noted that “multiple unit discounts may apply to much of this equipment.”
15

Multiple unit discounts were not assumed in our revised cost analysis. It is well established that economies arise from constructing multiple units at a single site. Economies will arise, for example, from common equipment that would serve all four units, such as the ammonia injection system and the control system. Economies arise from shop and material discounts based on quantity. Our cost analysis, however, did not assume any discount for multiple unit discounts. Regardless, for other reasons as stated elsewhere in our response to comments, we are finalizing a schedule which calls for compliance with the emission limits within 5 years—rather than 3 years—of the effective date of our final rule.

15
Revised BART Cost Effectiveness Analysis for Selective Catalytic Reduction at the Public Service Company of New Mexico San Juan Generating Station, November 2010, p. 5.

Comment:
The proposed FIP costs do not acknowledge, or take into account, the $330 million incurred in the past five years implementing a comprehensive emission control plan at SJGS. EPA's proposed BART determination for the SJGS is too expensive and EPA should accept the recently installed pollution control equipment at the SJGS as BART.

Response:
We did, as part of our NO
X
BART evaluation, consider the controls previously installed by PNM as a result of its March 10, 2005 consent decree with the Grand Canyon Trust, Sierra Club, and NMED. These controls included the installation of low-NO
X
burners with overfire air ports, a neural network system, and a pulse jet fabric filter. However, when making the NO
X
BART determination, we are obligated by the RHR to examine additional retrofit technologies.
16

In so doing, we have determined that SCR is cost effective and results in significant visibility improvements at a number of Class I areas, over and above the existing pollution controls currently installed.

16
“You are expected to identify potentially applicable retrofit control technologies that represent the full range of demonstrated alternatives.” 70 FR at 39164.

Comment:
EPA proposes to conclude that, because the SJGS currently is subject to a federally enforceable permit limit of 0.30 lb/MMBtu for NO
X
, which is less restrictive than the WRAP modeling's assumed NO
X
rates for those units (as characterized by EPA), additional NO
X
emission controls are required. EPA, however, proposes on this basis to determine that the BART emission limit for units 1 through 4 at SJGS is not 0.27 (or 0.28) lb/MMBtu but is instead 0.05 lb/MMBtu based on the application of SCR technology. As a result, EPA discontinues its evaluation of other technologies before fully assessing their relative cost-effectiveness and other factors mandated by section 169A(g)(2) of the CAA. EPA's analytical approach is in conflict with its own BART rules and is inconsistent with a logical approach to assessing relative cost-effectiveness of various technology options.

Response:
We disagree with this commenter's characterization of our analysis. As discussed in our proposal (76 FR 491), once we established that units 1, 2, 3, and 4 of the SJGS were subject to BART, we conducted a full five factor BART analysis (40 CFR 51.308(e)(1)(ii)(A)), rather than relying on the WRAP modeling. In conducting the BART analysis, we identified all available retrofit control technologies, including Selective Non Catalytic Reduction (SNCR), considering the technology available, the costs of compliance, the energy and non-air quality environmental impacts of compliance, any pollution control equipment in use at the source, the remaining useful life of the source, and the degree of improvement in visibility which may reasonably be anticipated to result from the use of such technology. In so doing, we did assess other NO
X
control technologies.
17

17
76 FR at 499.

Comment:
Several commenters stated EPA should follow its own promulgated RHR and follow New Mexico's recommendation for BART determinations These commenters are referring to the proposal that was sent to New Mexico's Environmental Improvement Board on February 11, 2011 (later formally submitted to EPA on July 5, 2011). The proposed revision to the SIP finds that BART for SJGS is SNCR—not SCR. One commenter believed that the application of the 2005 BART Guidelines supports a NO
X
emission rate for the SJGS of between 0.23 to 0.39 lb/MMBtu, as opposed to our proposed FIP of 0.05 lb/MMBtu, which requires costly SCR technology. One commenter stated the presumptive limits should be required “unless you [the BART-determining authority] determine that an alternative control level is justified based on consideration of the statutory factors.” 70 FR at 39171. Except for cyclone boilers (which are not present at SJGS), this commenter noted, our presumptive NO
X
BART limits are not based on application of SCR; as noted above, they are instead based on the use of combustion controls. Further, EPA determined that when current combustion control technology would be insufficient to meet the presumptive limits, it would

be appropriate to “consider whether advanced combustion control technologies such as rotating opposed fire air should be used to meet these [presumptive] limits.”
Id.
at 39172. Another commenter asserted that a proper BART assessment would take the presumptive limits into account by beginning with the assumption that the established presumptive limit for these units is appropriate, and then would proceed with an analysis of whether the least stringent control options could achieve that limit. A five-factor BART analysis of increasingly stringent control options could then properly assess incremental costs (and cost-effectiveness) and any benefits of requiring more stringent controls.

Response:
We note the RHR states:

For each source subject to BART, 40 CFR 51.308(e)(1)(ii)(A) requires that States identify the level of control representing BART after considering the factors set out in CAA section 169A(g), as follows:

States must identify the best system of continuous emission control technology for each source subject to BART taking into account the technology available, the costs of compliance, the energy and non-air quality environmental impacts of compliance, any pollution control equipment in use at the source, the remaining useful life of the source, and the degree of visibility improvement that may be expected from available control technology.
18

18
70 FR at 39158.

The RHR also states:

States, as a general matter, must require owners and operators of greater than 750 MW power plants to meet these BART emission limits. We are establishing these requirements based on the consideration of certain factors discussed below. Although we believe that these requirements are extremely likely to be appropriate for all greater than 750 MW power plants subject to BART, a State may establish different requirements if the State can demonstrate that an alternative determination is justified based on a consideration of the five statutory factors.
19

19
70 FR at 39131.

We followed the five statutory factors when assessing NO
X
BART at the SJGS, in determining that a different level of BART control was warranted.
20

This analysis included an examination of whether other technologies should be BART for the SJGS. We also performed our BART evaluation on the basis of increasingly stringent levels of control and assessed incremental costs and cost effectiveness. Thus, we do not believe we improperly truncated the NO
X
BART assessment for the SJGS.

20
76 FR 491, 499.

We received a New Mexico RH SIP on July 5, 2011. This SIP does contain a revised BART analysis that concludes that NO
X
BART for the SJGS should be SNCR and an emission rate of 0.23 lb/MMBtu on a 30-day rolling average. We will review the State RH SIP submittal, and if there is significant new information that changes our analysis, we will make appropriate revisions to today's decision. However, the State RH SIP recommends SNCR as BART, and we have considered that technology in the context of responding to other comments in this notice. For the reasons discussed in our proposal (76 FR 491), and in other responses to comments, we have concluded that BART for the SJGS is an emission limit of 0.05 lbs/MMBtu, based on a 30 BOD average, more stringent than the levels achievable by the SNCR technology recommended by the State.

Comment:
To meet a three-year deadline, PNM would have to prefabricate as much of the SCRs as possible. In addition, a three-year deadline would also require significant overtime hours, expedited material costs, double “heavy long-lift” crane costs, and a larger construction workforce overall. Because these costs would never be incurred in the normal course of installing SCRs, PNM did not include these costs in its analysis, but they would be unavoidable in the event a three-year deadline is required. Such a short construction deadline would also exacerbate the shortage of skilled labor caused by the significant number of similar projects that are either ongoing or planned for the near future in the region. The failure to account for the additional labor costs associated with such a short timeframe, particularly given other factors affecting the market for skilled labor, renders both the three-year deadline and the cost estimate prepared by EPA unrealistic.

Response:
The information in the record does not demonstrate a shortage of labor necessary to complete the installation of SCRs at the SJGS. However, as stated elsewhere in our response to comments, we have modified the schedule for compliance with the emission limits to now require compliance within 5 years—rather than 3 years—from the effective date of our final rule. We believe this compliance schedule will provide adequate time to schedule the necessary labor resources for the installation of controls at the SJGS.

Comment:
The NPS recommends that in addition to the $/ton metric, we evaluate the visibility metric $/deciview as an additional tool to report the benefits of emissions controls. The NPS contends that BART is not necessarily the most cost-effective solution. Instead, it represents a broad consideration of technical, economic, energy, and environmental (including visibility improvement) factors. The NPS notes that one of the options suggested by the BART Guidelines to evaluate cost-effectiveness is $/deciview. The NPS believes that visibility improvement must be a critical factor in any program designed to improve visibility. The NPS goes on to provide several examples of $/deciview calculations.

Two other comments recommend we employ the $/deciview metric. One commenter states EPA has not appropriately considered the costs of compliance for any proposed BART for the SJGS because it relies on a $/ton metric. The commenter maintains that cost should be related to the amount of visibility improvement that it is projected to achieve and proposes the $/dv as the means for making a rational comparison of the relative cost-effectiveness of control measures.

This commenter also states that a method that aggregates projected visibility improvement in each affected class I area is not appropriate for several reasons. That approach masks the fact that it is cumulative over time and space and does not represent actual change at any one class I area. That approach also ensures an artificially low measure of cost-effectiveness simply by allowing the control cost to be divided by a larger value. The commenter suggests that a $/dv metric expressed as a range of the values for each affected class I area would be an appropriate means for comparing cost-effectiveness of different controls. The commenter states that EPA's current measure of cost-effectiveness in terms of $/ton is virtually meaningless in the context of the RH program. Thus, EPA's assessment of the $/ton costs of BART candidates for the SJGS is flawed because the premise for its use is faulty,
i.e.,
a change in emissions is not a suitable surrogate to represent a change in visibility.

Another commenter believes that a dollar per deciview of visibility improvement metric would be more in line with the overall goal of the RH program, namely to improve visibility in national parks and wilderness areas. To properly gauge cost-effectiveness, EPA must consider the fact that installing SCRs at San Juan will cost between $78 million and $336 million per deciview, depending on the Class I area.

Response:
The BART Guidelines require that cost effectiveness be calculated in terms of annualized dollars per ton of pollutant removed, or

$/ton.
21

The commenters are correct in that the BART Guidelines list the $/deciview ratio as an additional cost effectiveness measure that can be employed along with $/ton for use in a BART evaluation. However, the use of this metric further implies that additional thresholds of acceptability, separate from the $/ton metric, be developed for BART determinations for both single and multiple Class I analyses. We have not used this metric because (1) We believe it is unnecessary in judging the cost effectiveness of BART, (2) it complicates the BART analysis, and (3) it is difficult to judge. We conclude it is sufficient to analyze the cost effectiveness of potential BART controls using $/ton, in conjunction with the modeled visibility benefit of the BART control. We have addressed the commenter's statement that we should not aggregate visibility improvement over Class I areas elsewhere in our response to comments.

21
70 FR 39167.

2. Comments on Specific Cost Line Items

The comments that follow have been summarized to capture each one's main points and most of the references have been removed. The reader is encouraged to refer to our
Complete Response to Comments for NM Regional Haze/Visibility Transport FIP
for more details and references.

Comment:
The NPS stated that PNM has improperly rejected use of the Cost Manual in favor of methods not allowed by EPA. The NPS states the SCR cost estimates submitted by PNM are severely lacking in the types of specific information needed to give them credibility. The NPS goes on to provide a great deal of detailed information that supports their opinion that specific cost items were overestimated. This information includes the following cost item categories:

• Appropriateness of using the Cost Manual.

• Problems in B&V's scaling of cost items from another project.

• Ductwork and ammonia grid costs.

• Reactor box and breaching.

• Expansion joints.

• Sonic horns.

• Elevator.

• Structural steel.

• SCR bypass.

• Catalyst.

• NO
X
monitoring.

• Auxiliary electrical system upgrades.

• Instrumentation and control systems.

• Air preheaters.

• Balanced draft conversion.

• Contingencies.

• Operating Labor.

• Reagent.

• Auxiliary power demand.

• Catalyst life.

• Interest rate.

• Effect on cost of PNM's assumption of an emission rate of 0.07 lbs/MMBtu.

The NPS concluded their critique of PNM's cost estimate with their own estimate of an average cost of $2,600/ton for the four units of the SJGS.

Response:
We agree with the general contention that many individual cost items for the installation of SCR on the units of the SJGS were overestimated by PNM. Please see elsewhere in our response to comments for our opinion regarding the appropriate estimated costs for these and other cost items. We note that the NPS estimate of an average cost of $2,600/ton for the four units of the SJGS closely agrees with our own revised estimate.

Comment:
EPA failed to account for the costs associated with ensuring sufficient auxiliary power to operate SCRs at SJGS. EPA discounted by nearly 80 percent the estimated cost of the auxiliary power upgrades needed to power the SCRs. The theory behind this sharply discounted cost estimate is that the SCRs will only be responsible for approximately 20 percent of the total draft pressure of the units and that therefore the cost of the auxiliary power upgrades should be allocated in similar fashion. Without SCRs, no additional auxiliary power would be needed. As such, those costs must be included in the cost of the SCRs, as they represent one of the site-specific concerns that could make the installation of SCR at SJGS more difficult than other units. The decision by EPA to exclude these costs underestimates the cost of SCRs for SJGS by $73,175,000.

Response:
We disagree that installing SCRs would by itself trigger the need to upgrade the auxiliary power system, especially to the extent proposed by PNM. The upgrade benefits the entire auxiliary power system. The modifications, for example, include new transformers, switchgear, and motor control centers that will serve the entire fan auxiliary loads of both the Consent Decree projects and the SCR.
22

The modifications also include replacing the existing fans with upgraded units. These fans will service more than just the SCRs.

22
B&V 10/22/10 Cost Analysis, Sec. 3.0 and 11/4/10 Norem E-mail to Kordzi, Re: Questions on PNM's Revised Cost Estimate for the SJGS SCR Project, Response to Question 3, Table 3 of attachment 1.

This comment advocates attributing 100% of the cost of the auxiliary power system upgrade, recognized after the fact, to the last project to be implemented, the SCR. We did not “discount” the cost of the auxiliary power system by 80%, but rather distributed it among the control projects planned around the same time that triggered its need according to each control's contribution to draft pressure lost. This recognizes that the upgrade provides benefits to the entire system and includes elements that are more than strictly necessary because of the installation of the SCR. Therefore, it is not appropriate to attribute the entire cost of the upgrade to the SCR project. We believe our approach is consistent with standard engineering practices.

Comment:
EPA failed to account for additional costs associated with protecting the air preheater following an SCR Installation. Ammonia reacts with sulfur in the flue gas downstream of the SCR forming ammonium bisulfate (ABS), which condenses in the air preheater. ABS is an acidic substance that forms a sticky deposit on heat transfer surfaces, resulting in both corrosion of the equipment and the collection of fly ash that plug passages, which ultimately impairs the efficiency and reliability of the unit. As such, the installation of a retrofit SCR generally requires a modification to the air preheater to allow for easier cleaning of the basket surfaces in order to protect the heat transfer elements against the potential damage that might otherwise result from ABS. EPA deleted the costs of protecting the air preheater in its SCR cost analysis, “pending compelling justification that they are required for the SCR.” EPA's cost analysis recognizes that modifications to the air preheater are generally required for “units that burn high sulfur coal,” but EPA assumes that such modifications are not necessary “for a properly designed SCR on a boiler that burns low sulfur coal.” EPA is correct that, in spite of the quoted discussion above, Sargent & Lundy did not recommend air preheater modifications in the SCR cost analysis for the Navajo Generating Station. However, that recommendation was based on the specific emission characteristics at Navajo Generating Station, which differ significantly from those at SJGS.

Response:
This comment attempts to distinguish the emission characteristics of Navajo Generating Station and the SJGS by pointing to differences in the coal quality to support air preheater modifications at SJGS but not at Navajo. We obtained and analyzed the Navajo design basis coal quality. The

differences in coal quality are either not material (sulfur, heat content) or mitigate the potential impacts of ammonium bisulfate plugging (higher ash at SJGS). The key factors that determine whether ammonium bisulfate plugging will occur are not coal quality, but rather the amount of sulfur trioxide (SO
3
) and ammonia in the exhaust gases that reach the air preheater and the air preheater temperature regime. The formation of ammonium bisulfate depends on the relative amounts of ammonia and SO
3
in the exhaust gases. When the molar ratio is more than 2:1, ammonium sulfate (not ammonium
bi
sulfate) is preferentially formed. The average molar ratio for both SJGS and Navajo over the catalyst lifetime is much higher than 2:1. Thus, ammonium sulfate would be preferentially formed. Ammonium sulfate is a dry powder at all air preheater operating temperatures and does not create a fouling problem. Thus, consistent with Sargent & Lundy's conclusion for the nearby Navajo Station, which burns a similar coal, ammonium bisulfate fouling would not be expected and we do not believe that upgrades are justified for the air preheaters due to SCR installation.

Comment:
The installation of SCR at SJGS would increase the resistance in the flue gas path for the units. To overcome that additional resistance, PNM would need to install new higher capacity fan rotors and motors because the SCRs will add an additional pressure drop in the system of 10 inches of water gauge (w.g.). This change in pressure and higher fan pressure ratings would increase the potential risk of a boiler implosion during transient (upset or malfunction) conditions. The analysis prepared by B&V of the expected cost of an SCR retrofit includes the costs to mitigate the implosion risk by converting to balanced draft and stiffening the boiler and associated flue gas path. EPA concludes that additional boiler stiffening would not be required, stating simply that “a balance draft conversion with the proposed stiffening is not part of an SCR project.”

Response:
The basis for selecting 10 in. w.g. for a 77% NO
X
removal SCR is not explained or documented in the record. The overall SCR system pressure drop consists of losses from the SCR catalyst, static mixers, and duct work. Determining the pressure drop due to the SCR requires a more advanced design than presented in the B&V BART analysis. Instead, B&V appears to have assumed that the pressure drop due to the SCR would be 10 in. w.g., which is at the upper end of the usual range of 3 to 10 in. w.g. The B&V record, for example, contains no duct arrangement drawings; no catalyst vendor quotes; does not identify the type of catalyst,
e.g.,
honeycomb or plate; does not specify the catalyst pitch; and is silent as to static mixers, all important factors in determining the pressure drop due to the SCR. Thus, we do not believe there is a basis for the 10 in. w.g. used to cost boiler stiffening and to justify balanced draft conversion. This pressure drop likely has not been optimized and could be significantly reduced by catalyst selection (
e.g.,
by using honeycomb with large pitch) and ductwork design. Therefore, we do not concur that the record supports a pressure drop of 10 in w.g. for the SCR.

Comment:
Installation of SCR's at SJGS will increase boiler and duct implosion potential due to increased draft system requirements and fan pressure ratings. SCRs will trigger the need to choose between either designing to the general standard of +/− 35 inches w.g. (which is typical for a newly designed power plant) or performing a “more complete and rigorous analysis” to determine whether PNM will qualify for an exception from the generally-applicable implosion protection standard through the use of alternative methods. To date, neither PNM nor its consultants have fully determined whether an alternative to the +/− 35 inches w.g. standard would suffice following installation of an SCR, due to the significant amount of time and expense that would be associated with that analysis. Therefore, B&V included the cost of stiffening the boilers to +/− 35 inches w.g. in its analysis. EPA's failure to properly account for the boiler stiffening costs underestimates the cost of the SCR retrofits for SJGS by $55,718,000 in capital costs for boiler stiffening and properly sized fans and motors.

Response:
This comment acknowledges that the boiler stiffening costs represent a worst case estimate. The magnitude of these costs is unusual. The BART Guidelines require that unusual costs be documented in the record. These costs are stated without providing the underlying engineering calculations. PNM states that the boilers were stiffened to negative pressure differentials of 18 in. w.g. during the Consent Decree projects. The 10 in. w.g. estimate is a worst-case upper bound that is not supported by vendor quotes and SCR design. We agree some cost for code compliance is warranted. However, the worst case used in B&V's analysis is unreasonable and unsupported, given the SCR's potential upper bound contribution of 10 in. w.g. Absent the “more complete and rigorous analysis” to support upper bounds for both an SCR pressure differential and stiffening to +/− 35 in w.g., we feel stiffening costs should have been based on no more than the SCR's contribution to the increase from current conditions of 18 in. w.g. to 35 in. w.g. Thus, we modified our cost analysis to estimate the stiffening cost based on the SCR's maximum contribution to the increase from 18 in. w.g. to 35 in. w.g. or by 59%. This increased our estimate of the capital cost to install SCRs by $19,258,318.

Comment:
EPA failed to account for the cost of installing the initial layers in the SCR. The cost analysis prepared by B&V included the cost of the initial layers of catalyst in the capital cost and including the replacement layers in the annual operating cost calculation. EPA, however, appears to have misunderstood the analysis and assumed that the initial catalyst layers were double-counted. As a result, it subtracted the initial catalyst cost from the capital cost calculation, without adding it to the annual cost calculation. As such, EPA's failure to include the cost of the initial layers of catalyst in its analysis underestimates the cost of installing SCRs at SJGS by $33,556,000.

Response:
We agree with this comment. We have revised our cost analysis to include the initial catalyst charge.

Comment:
Sorbent injection will be needed if PNM must install SCRs at SJGS, and the EPA cost analysis should reflect those costs. Sorbent injection systems are often used at coal-fired power plants equipped with SCRs to help reduce emissions of sulfuric acid mist that are an unavoidable byproduct of the chemical reactions that occur in an SCR. Sulfuric acid mist resulting from SCR operation has been known to cause a visible plume at some units in the industry. Although the installation of SCRs may not result in such a plume at SJGS, the sorbent injection system would be needed to ensure a visible plume does not materialize. The failure to address the sulfuric acid mist created by the SCR can reduce any visibility benefits associated with an SCR.

Response:
We disagree with this comment. B&V updated its cost analysis in October 2010. This is the most recent version of B&V's cost analysis, which was critiqued in our Technical Support Document (TSD) in our proposal. This analysis did not include any costs for sorbent injection. In its June 21, 2010 BART Determination, NMED concluded that BART for SJGS was SCR plus sorbent injection to remove SO
3
and requested a sorbent injection cost analysis from PNM. However, we

disagreed and concluded that sorbent injection was not required due to the low sulfur content of the coal, availability of low conversion SCR catalyst, and our calculations. We see no reason to change that view. The reasons advanced in this comment for requiring sorbent injection to control sulfuric acid mist (SAM) are not applicable to the SJGS SCR. Visible plume issues have only been experienced at units that burn high sulfur coal, containing greater than 2+% sulfur and typically over 3% sulfur,
e.g.,
Gavin, Ghent. The coal burned at SJGS contains 0.77% sulfur, much lower than the amount of sulfur that has resulted in visible plume issues elsewhere and is considered to be low sulfur. No explanation is provided for why the commenter believes a plume may “materialize” on installing SCR. If the SCR is properly designed to address SJGS's coal, a plume should not materialize. Low conversion catalysts capable of achieving an SO
2
conversion as low as 0.1% per layer of catalyst in the high dust, hot (>650 F) position and 0.5% across the entire SCR reactor are common in higher sulfur and other applications. Even lower levels can be achieved if the catalyst is regenerated.

Comment:
EPA's calculation of sulfuric acid emissions is incorrect. EPA estimated sulfuric acid mist emission levels based on a document prepared by the Electric Power Research Institute (EPRI), which describes a formula used by many utilities to estimate sulfuric acid emissions. However, in applying that formula, EPA assumed an ammonia slip value of 2.0 parts per million (ppm), even though actual ammonia slip varies over the life of a catalyst layer from very low values up to 2.0 ppm as the catalyst ages. A more appropriate assumption for ammonia slip is the 0.75 ppm value recommended by the EPRI formula, which better represents the expected ammonia slip over the life of a catalyst. Using that assumption, the sulfuric acid emissions from SJGS are calculated to be twice that assumed by EPA. As a result, EPA's attempt to justify its decision to delete the costs of sorbent injection based on minimal sulfuric acid mist emissions is incorrect.

Response:
The commenter is correct in that the EPRI report does suggest that a value of 0.75 ppm should be used. We note that the ammonia slip of an SCR is minimal when the catalyst is new and increases as the catalyst ages. In order to be conservative, we recalculated the sulfuric acid emission rate, based on zero ammonia slip, to be 2.6 X10
−4
lb/MMBtu, compared to our original value of 1.06 X10
−4
lb/MMBtu at 2ppm ammonia slip. The 2.0 ppm we selected in our proposed visibility modeling was based on the maximum slip from PNM's design specifications. This revised sulfuric acid emission rate remains significantly lower than that estimated by NMED and is a minimal level of sulfuric acid emissions. We continue to conclude that sorbent injection is not required due to the low sulfur content of the coal, availability of low conversion SCR catalysts, removal by existing control equipment and our revised calculations.

Comment:
The EPA also cites to the results of a stack test performed at the Navajo Generating Station in November 2009 to conclude that actual sulfuric acid mist emissions are lower than would be estimated using the EPRI Method. However, the air quality control industry generally considers sulfuric acid testing to be very prone to inaccuracy because the test methods used are susceptible to bias. Also, sulfuric acid emissions vary significantly from unit to unit because emissions removal is dependent on many variables including temperature, moisture, process operation, air quality control equipment, ambient conditions, and the quality of the testing. As mentioned above, SJGS and the Navajo Generating Station differ significantly in many of these respects. Therefore, it is not appropriate to use test results from Navajo Generating Station to make assumptions about SJGS.

Response:
We believe this comment mischaracterizes our analysis. We did not use test results from the Navajo Generating Station to make assumptions about the SJGS. Rather, we compared sulfuric acid mist emissions calculated for Navajo using the EPRI procedure with a stack test at Navajo in accordance with EPA Method 8A procedures. Thus, we compared Navajo EPRI estimates with Navajo test data to judge the accuracy of the EPRI procedure. This comparison suggests that the EPRI method may overestimate sulfuric acid mist emissions when firing a similar coal if PNM's assumptions are used. This analysis supports the conclusion that the EPRI method and parameters we used provide a better estimation of sulfuric acid emissions than the methodology and parameters utilized by PNM and NMED in their analysis, which overestimates these emissions. We also note that PNM estimates for sulfuric acid emissions that were reported to the Toxic Release Inventory in recent years are much lower than those estimated by PNM for their BART analysis.

Comment:
It is appropriate to include sorbent injection costs in the SCR cost analysis because sorbent injection may be required by law. The Prevention of Significant Deterioration (PSD) program under the CAA requires major sources to install additional controls to address any significant net emissions increases resulting from a physical change to an emissions unit. Because the SCR will constitute a physical change to the SJGS emission units, and could have the potential to result in a significant net emissions increase in sulfuric acid mist, additional controls could be required by the PSD program. If triggered, the PSD program would require the installation of “best available control technology,” which for sulfuric acid mist emission increases would likely include a sorbent injection system. Although there remains some uncertainty as to whether the SCR would trigger PSD permitting requirements, PNM believes it is appropriate to include the cost of the system in the SCR cost analysis, and the failure to include those costs underestimates the cost of the SCRs by $12,118,000.

Response:
For the reasons outlined elsewhere in our response to comments, we believe the level of sulfuric acid generated at the SJGS will be so low that sorbent injection will not be needed. However, it is possible that the installation of SCR on all four units of the SJGS could generate enough additional sulfuric acid that a PSD review could be triggered. EPA is not the permitting authority for sources in New Mexico but we believe it is reasonable to anticipate that a subsequent BACT analysis for sulfuric acid emissions at the SJGS will determine that no additional controls are required because despite the projected increase in sulfuric acid emissions, emissions are expected to remain low. In considering SCR for controlling NOx, EPA specifically considered the issues of sulfuric acid formation. In our review, we believe that the emission limits for NOx can be achieved through the use of lower reactivity catalyst, thus mitigating the formation of sulfuric acid across the catalyst bed. We have set an emission limit for emissions of sulfuric acid that restricts the increase of sulfuric acid. According to the two most recent Toxic Release Inventory (TRI) reports submitted by SJGS, the total sulfuric acid emissions are very low (17.77 TPY for 2009, and 27.5 TPY for 2008). Based on our calculations, we believe the current emissions of sulfuric acid to be significantly lower than these reported values due to the low sulfur content of the coal and the removal of sulfuric acid in the installed control equipment, including wet scrubbers and fabric filters. We project, with the

implementation of SCR using a low reactivity catalyst that total emissions of sulfuric acid will remain below 22 tons/year.
23

In this particular case, sorbent injection technology is unlikely to be cost-effective on a cost per ton basis of sulfuric acid mist removed. Again, we note that the New Mexico Environmental Department is the permitting authority and has the primary responsibility to implement the New Source Review program which includes the PSD permitting process, and the issuance of the applicable permit. NMED will be responsible for determining if PSD will be triggered for increases in sulfuric acid emissions or other NAAQS pollutants and in determining the BACT for such increases.

23
Based on our emission limit of 2.6×10
−4
lb/MMBtu and conservatively assuming each unit operates 100% of the year (8760 hr/yr).

Comment:
EPA failed to account for the additional steel that will be needed due to site congestion at the SJGS. EPA assumed that the “complexity factor” applied to the structural steel cost in PNM's cost analysis was a “contingency factor.” As such, EPA assumed that PNM had double-counted contingency costs by using both the “complexity factor” for structural steel and a more general “contingency factor” overall. PNM asks EPA to reconsider the analysis provided by B&V, given that the engineers at B&V made several site visits to SJGS and designed the SCRs for the St. John's River Power Park (SJRPP). The pictures of SJRPP and SJGS provided by B&V illustrate the differences in site congestion. EPA underestimated the cost of its BART proposal by $35,087,000 by failing to accurately account for site congestion.

Response:
A complexity factor is a subset of a contingency factor as it estimates unknown costs. PNM applied a complexity factor of 1.2 for Units 1 and 4 and 1.5 for Units 3 and 4. We regard these factors as rough estimates that cannot be fully determined until the SCR is designed. We visited the SJGS plant on May 19, 2011.
24

This visit confirmed that the site is congested. However, this does not confirm that the cost of structural steel for Units 1 and 4 would be 1.2 times higher than at SJRPP, and 1.5 times higher for Units 2 and 3, as this comment contends. The materials provided by PNM do not contain any plot plans or design drawing for SJRPP (or SJGS) that would allow one to conclude anything about the cost of structural steel at one facility compared to the other. Photographs attached to the PNM comments indicate more room for crane access at SJRPP than at SJGS, but this does not address the capital cost of the structural steel framework, only the cost of constructing it.

24
See San Juan Generating Station Site Visit, 5/23/11.

The BART Guidelines require that “documentation” be provided for “any unusual circumstances that exist for the source that would lead to cost-effectiveness estimates that would exceed that for recent retrofits.” We specifically asked PNM to identify any retrofit constraints and support them with engineering calculations, drawings, and photographs. PNM has not provided specific documentation that supports the use of their chosen structural steel complexity factors. Nevertheless, based on the information that was provided, we have modified our cost analysis to use B&V's estimate for structural steel, which includes the “complexity factors” cited in this comment, as B&V produced designs for both facilities.

Comment:
EPA failed to account for the SCR bypass that will be necessary to protect the SCR during startup on oil. EPA assumed that SJGS could initiate startup of its units on oil without fouling the catalyst in the SCR. EPA's justification for the removal of this cost line item was that fuel oil is efficiently burned in modern low NOx burners with oil igniters, citing two coal-fired units that have shown the ability to startup on oil without a bypass and two oil-fired boilers with SCRs that do not have a bypass. Based on these references, EPA concluded that SJGS will be able to startup on oil without risking catalyst fouling resulting from a coating of incompletely combusted fuel oil. The failure to account for the needed SCR bypass system underestimates the cost of installing SCR at SJGS by $126,484,000.

Response:
We disagree with this comment. The removal of SCR bypass costs was based on several factors. First, a noted air pollution handbook concluded (before U.S. ozone season trading programs made them routine): “most applications do not have SCR bypasses, since routines are used during startup and shutdown which preclude their need” (Cho and Dubow),
25

and regulations sometimes prohibit their use. Also, experience in Japan and Germany has shown them to be costly and not required to prevent damage due to low-load oil firing, thermal gradients, and other conditions. We believe a bypass is not required in a properly designed and operated SCR system to prevent SCR catalyst fouling during startup or operation on oil. Two examples were cited in our TSD as part of our proposal to confirm this information. In addition, Sargent & Lundy, the consultant that prepared the design and cost estimate for SCR for the 3 units at Navajo Generating Station, an existing facility of similar age and retrofit complexity that starts up on oil, did not recommend an SCR bypass in its BART analysis.

25
S.M. Cho and S.Z. Dubow, Design of a Selective Catalytic Reduction System for NO
X
Abatement in a Coal-Fired Cogeneration Plant, Proceedings of the American Power Conference, April 13-15, 1992, pp. 717-722.

Comment:
The EPA cost estimate also does not properly estimate annual operating costs for auxiliary power consumption and catalyst replacement rate. B&V estimated the amount of auxiliary power needed to run the SCR to be 16,297 kW (for all four units) at a cost of $0.06095 per kWh, based on a site-specific analysis. Specifically, B&V's calculation was based on the calculation of the additional fan energy (based on flue gas flow rate and estimated pressure drop from the SCR) and the power consumption for the auxiliary equipment (such as the ammonia system). EPA, on the other hand, simply assumed a cost of 5,400 kW at $0.05 per kWh based on a percentage estimate for “typical” SCR installations. This error underestimates the cost of auxiliary power consumption when operating SCRs by $5,388,000.

Response:
EPA disagrees with the comment. First, the claimed “site-specific analysis” was not submitted for inclusion in the record, and thus EPA and the public could not review it. Second, the values that would affect the cost analysis,
e.g.,
duct length, catalyst pressure drop, would be estimates as the SCR system has not yet been designed. In fact, the record does not even contain an arrangement diagram, required to determine duct lengths. Third, the B&V estimate of the amount of auxiliary power needed to run the SCR (16,297 kW) was initially rejected by us as it amounts to 0.9% of the total gross generating capacity of the station, which is high compared to other estimates known to us. An SCR typically uses about 0.3% of a plant's electric output, which would be about 5,400 kW or three times less than assumed in the B&V cost analysis. The BART Guidelines require that unusual costs be documented in the record. PNM did not supply any additional information to support its unusually high estimate.

Fourth, as discussed elsewhere in our response to comments, no support has been provided for PNM's claim of a 10 in. w.g.
26

pressure drop due to the SCR,

which is at the upper end of the usual range of 3 to 10 in. w. g. Fifth, the unit cost of electricity used by B&V, $0.06095/kWh, is much higher than the auxiliary power cost commonly used in cost effectiveness analyses, and thus was not justified. Auxiliary power is the power required to run the plant, or power not sold. Cost effectiveness analyses are based on the cost to the owner to generate electricity, or the busbar cost, not market retail rates. The B&V estimate is based on the average forecasted cost of replacement power for 2007 to 2012.
27

Thus, even if this is the correct site specific cost, it is the wrong metric for a cost effectiveness analysis. We further note that the use of forecast cost is inconsistent with the BART methodology, which is based on current dollars. We conservatively used the upper end of the range of costs assumed in BART cost effectiveness analyses ($0.03/kWh to $0.05/kWh)
28

or $0.050/kWh. After our analysis was complete, PNM responded to a question from us that its average cost of production is $0.047/kWh ($47.83/MWh). This rounds up to 0.05/kWh, the number we used. Thus, we have made no changes to our estimate of auxiliary power demand.

26
10/22/10 B&V Cost Analysis Update, Appendix B; 6/7/07 B&V San Juan BART Analysis, p. B-3.

27
E-mail from Norem to Kordzi, October 21, 2010, Re: PNM Responses to Follow-Up Questions from October 14, 2010 Conference Call Regarding BART Cost Estimate, October 21, 2010 (10/21/10 Responses), Response to Question 9, pp. 3-4.

28
Sargent & Lundy, Sooner Units 1 & 2, Muskogee Units 4 & 5 Dry Flue Gas Desulfurization (FGD) BART Analysis Follow-Up Report, Prepared for Oklahoma Gas & Electric, December 28, 2009, Attach. C, pdf 109; (Gerald Gentleman—$45.65/MWh; White Bluff—$47/MWh; Boardman/Northeastern/Naughton—$50/MWh; Nebraska City—$30/MWh).

Comment:
In its analysis, EPA recognized that the Cost Manual does provide factors to estimate certain “direct installation costs,” namely foundation/supports, handling/erection, electrical, piping, insulation, painting, demolition, and relocation. However, the Control Cost Manual fails to provide factors to estimate these costs for SCR, as recognized in EPA's analysis. EPA indiscriminately took the median of the factors for other control technologies, which vary significantly from SCRs. As a result, EPA's analysis slashes in half the direct installation costs estimated by B&V. For example, the direct costs assumed by EPA for Unit 1 are $8,799,917, but that amount would only cover 159,998 man-hours, or 21 weeks of construction. EPA's own schedule, even though insufficient itself, assumes 38 weeks of construction, nearly double of the amount that EPA's analysis could afford. Thus, EPA's estimate is insufficient for its own estimated construction timeline, much less the 64 to 72 weeks of construction that PNM's experienced consultants predict.

Response:
We disagree with this comment. The B&V direct installation costs were calculated by multiplying total purchased equipment costs by various unsupported percentages, a rough estimating practice referred to as “factoring.” B&V did not submit into the record the basis for the various factors that they used. The percentages that B&V used are demonstrably high. We compared each of B&V's direct costs with those from a major SCR designer's (Babcock Power) database and from similar SCR projects nationwide. Foundation and supports, costed by B&V as 30% of purchased equipment cost, for example, based on its estimate of purchased equipment cost, are two to three times higher than upper bound costs reported by Babcock Power for similar sized units ($8/MW compared with the B&V estimate of $18/MW to $29/MW for SJGS). Based on these comparisons the B&V's costs were excessive. No documentation has been provided to justify the higher B&V costs.

The Cost Manual estimating procedure for direct installation costs is based on the same factoring approach used by B&V. We tabulated the factors for total direct installation costs for all controls reported in the Manual. These ranged from 30% to 85% of the purchased equipment cost. In comparison, B&V assumed direct installation costs were 103% to 113% of total purchased equipment cost.

We calculated direct installation costs for SJGS using the median of this range or 62% of purchased equipment cost. This is consistent with the upper bound Babcock Power estimate for actual retrofit SCR installations and estimates made by others. The B&V estimate is also high compared to direct installation costs that it reported for the SJRPP SCR, which was otherwise used to extrapolate equipment costs to SJGS. The direct installation costs for the SJRPP SCR were 95% of the total purchased cost. We have revised our cost estimate to use this percentage to conform to the balance of the B&V cost estimate.

The B&V estimate assumes a 150-man crew for the entire 21 weeks, a 50-hour workweek for the duration, and a wage of $55/hour. This represents peak staffing and labor rates, even though the number of workers would vary over time. Thus, our estimate of direct installation costs corresponds to a longer duration than claimed. Regardless, it is important to note that this duration corresponds to construction of a much smaller project (less SCR bypass, preheater modifications,
etc.
) than proposed by B&V. Further, for our proposal, we did not estimate construction duration, but rather the length of time from the effective date of the final rulemaking to startup of the SCR or 36 months. We note that we have revised our proposal to allow 60 months from the effective date of the rule allowing additional flexibility in deploying workers. Thus, the basis of this comment's starting point, an EPA estimate of 38 weeks, is incorrect. In addition, the B&V estimate does not contain a schedule, which is required to estimate the staffing and duration of construction.

Comment:
EPA asserts that “[t]he contingencies included in the B&V cost estimates are double-counted and excessive,” based on the misimpression that there are three contingencies “imbedded” in the analysis. However, two of the three allowances are for known costs, and therefore are not “contingencies.” Specifically, the complexity factor for structural steel costs of 1.2 (for Units 1 and 2) and 1.5 (for Units 3 and 4) are known, expected costs, and therefore do not constitute a contingency factor, as noted previously. Also, the $2 million estimated for underground obstructions and the $500,000 estimated for on-site buildings are also known, and therefore do not represent a duplicative contingency factor. Thus, EPA's claim that PNM double-counted its contingency costs is incorrect and underestimates the cost of SCRs at SJGS by $61,978,000.

Response:
This comment explains that the “complexity factor,” site unknowns, and general building requirements are not contingencies, but rather known factors. Based on this explanation and the information we have about the SJGS, we concur that these complexity factors, and the engineering estimates for underground obstructions and on-site buildings, are reasonable and we have modified our cost estimates to reflect B&V's estimates.

Comment:
EPA also claims that the Interest During Construction included in the B&V cost estimates are not allowed by the Cost Manual. Therefore, this cost was eliminated from the cost analysis underlying the proposed FIP. However, this cost item is a real project cost, which will be incurred by PNM to finance the project and must by recovered from the SJGS customers. The rejection of costs associated with Interest During Construction underestimates the cost of the project by $78,300,000.

Response:
The B&V cost analysis include a charge for interest during construction of 7.41% of direct plus indirect costs. This charge is generally

known as the Allowance for Funds Used During Construction (AFUDC) and is specifically disallowed under the Cost Manual methodology and specifically disallowed for SCRs.
29

A cost effectiveness analysis is a regulatory analysis that is based on current annual dollars without any inflation. AFUDC is an accounting method. Assets under construction do not provide service to current customers and thus associated interest and allowed return on equity are not charged to current customers. Instead, AFUDC capitalizes these costs and adds them to the rate base so that they can be recovered from future customers when the assets are used. Thus, these charges represent future cash income to the utility. In other words, AFUDC is the accumulated cost of carrying capital and holding it waiting to spend, so money can be made in the future by selling electricity. Future income should not be charged against the cost of a SCR in a BART cost-effectiveness analysis. These costs are not part of the constant dollar approach found in the Cost Manual and should not be included in BART cost-effectiveness analyses.

29
EPA Air Pollution Control Cost Manual, pdf 486, Table 2.5, E (Allowance for Funds During Construction) = 0.

3. Concerns Over Possible Electricity Rate Increases

Comment:
Both the CAA and EPA BART regulations require consideration of the remaining useful life of a source. Requiring the imposition of possibly $1 billion or more of control technology capital costs at SJGS, a nearly 40-year old plant, presents a likely scenario where the remaining useful life of SJGS is less than the time period needed for amortizing the costs of the control technologies. As such, it could make production at SJGS during its remaining useful life uneconomical in comparison with other existing or future plants. If, in light of SJGS' estimated remaining useful life, it is determined that an investment of such magnitude does not make economic sense, owners of SJGS must evaluate alternate long-term options for meeting obligations to provide a cost-effective, reliable supply of electricity to customers. As such, the significant cost of requiring such SCR at SJGS will substantially increase the cost of electricity produced by SJGS. Over two million electric customers in New Mexico and other western states stand to be directly and adversely affected by the EPA proposal. PNM estimates that the average residential customer will experience a 10 percent increase in rates due solely to EPA's proposed SCR technology. As a result of the Proposed Rule, PNM has indicated that possible sources of replacement power may be needed to ensure it can fulfill its obligation to provide electricity to the citizens of New Mexico.

Response:
The commenter is correct that the remaining useful life of a facility may impact the BART determination. As we note in the BART Guidelines,

The “remaining useful life” of a source, if it represents a relatively short time period, may affect the annualized costs of retrofit controls. For example, the methods for calculating annualized costs in EPA's OAQPS Control Cost Manual require the use of a specified time period for amortization that varies based upon the type of control. If the remaining useful life will clearly exceed this time period, the remaining useful life has essentially no effect on control costs and on the BART determination process. Where the remaining useful life is less than the time period for amortizing costs, you should use this shorter time period in your cost calculations.
30

30
70 FR 39104, 39169.

The BART Guidelines further clarify, “[w]here this affects the BART determination, this date should be assured by a federally- or State-enforceable restriction preventing further operation.”

As part of our review of PNM's BART determination for the SJGS, we met with representatives of PNM and its contractor several times, and communicated numerous times through e-mail and phone. At no point did PNM indicate that it wished to constrain the amortization period for financing BART controls based on the remaining useful life of the facility through the use of a federally enforceable restriction.

Comment:
Several local and county governments and municipal power systems expressed concern that the proposed FIP would require a major capital expenditure that could well exceed $750 million, according to PNM. Such significant costs will drastically increase the cost of power produced by the SJGS and have the potential to increase electricity rates in the communities served by the SJGS. Another commenter stated our NO
X
BART proposal for the SJGS would cost New Mexico or Albuquerque ratepayers $10.20 more a year, or 85 cents a month, which is the price of a candy bar, so cleaning up this decades old air pollution is affordable and now is the time to do it.

Response:
As discussed in our proposal, we disagree with PNM's cost estimate for installing SCR on the four units of the SJGS. Although PNM estimated the total cost to be in excess of $1 Billion, we estimated that cost to be approximately $250 Million. As discussed elsewhere in this notice, taking into consideration various comments, we have refined our estimate to be $344,542,604. In light of the visibility benefits we predict will occur, we consider this to be cost effective. We take our duty to estimate the cost of controls very seriously, and make every attempt to make a thoughtful and well informed determination. However, we do not consider a potential increase in electricity rates to be the most appropriate type of analysis for considering the costs of compliance in a BART determination. Nevertheless, we note that our cost estimate, being about
1/3
that of PNM's will result in significantly less costs being passed on to rate payers.

4. Comments That Opined on Our Reliance on the EPA Air Pollution Control Cost Manual

Comment:
The rejection of PNM's escalation factors is unrealistic. By relying too heavily on the Cost Manual, EPA's analysis not only omits the specific line items, it also omits or alters various estimating factors utilized by B&V in PNM's analysis. EPA relied on the Chemical Engineering Plant Cost Index (CEPCI) to escalate costs from the Cost Manual. However, although that index may be a reasonable tool for a chemical plant, it does not properly account for escalation of costs at power plants. In contrast, B&V developed an appropriate escalation factor with the help of an outside consulting firm specializing in financial analysis and forecasting, which incorporates the complete B&V database of “as-built” costs, the Bureau of Labor Statistics indices, and the consulting firm's database of costs and indices, all tailored specifically to the power generation industry.

Response:
The CECPI, which is published monthly by the magazine,
Chemical Engineering,
has been used for decades in regulatory cost effectiveness analyses and is one of the factors that allows a comparison to be made between cost effectiveness analyses at different facilities. This method was selected by EPA's Office of Air Quality Planning and Standards for use in regulatory cost effectiveness analyses because “this index specifically covers cost items that are pertinent to pollution control equipment (materials, construction labor, structural support, engineering & supervision,
etc.
).”
31

The

B&V escalation index, on the other hand, is proprietary and not subject to public review.

31
E-mail from Larry Sorrels (OAQPS) to Don Shepherd (Park Service) with cc to Anita Lee (EPA

Region 9), dated 7/21/10, concerning the SRP Navajo Generating Station SCR cost estimate.

Comment:
A commenter contends that EPA improperly rejected PNM's cost estimates, because EPA thought them inconsistent with the Cost Manual. The commenter states EPA should consider site-specific costs, even when those costs are not included in the Manual. The commenter further states that EPA did not take “unusual circumstances” into proper account and expresses the view that EPA did not consider site-specific elements that would eliminate available control technologies from consideration.

Response:
We disagree with commenter's view that our cost analysis is improper, but we agree that the Cost Manual is not the only source of information for the BART analysis. For instance, the reference to the Cost Manual in the BART Guidelines clearly recognizes the potential limitations of the Manual and the need to consider additional information sources:

The basis for equipment cost estimates also should be documented, either with data supplied by an equipment vendor (
i.e.,
budget estimates or bids) or by a referenced source (such as the OAQPS Control Cost Manual, Fifth Edition, February 1996, EPA 453/B-96-001). In order to maintain and improve consistency, cost estimates should be based on the OAQPS Control Cost Manual, where possible. The Control Cost Manual addresses most control technologies in sufficient detail for a BART analysis. The cost analysis should also take into account any site-specific design or other conditions identified above that affect the cost of a particular BART technology option.
32

32
70 FR 39104, 39166.

The Cost Manual establishes a methodology for calculating cost effectiveness that allows comparison across multiple units. The regulatory cost is expressed in current real or constant dollars, less inflation. B&V did not follow the regulatory cost method. Instead, it used CUECost, a model that estimates control costs using the levelized cost method developed by the EPRI, which is not approved for BART determinations; extrapolation from several other projects; and its own proprietary and confidential databases not available for public review.

As to unusual circumstances, the BART Guidelines call for “documentation” to be provided for “any unusual circumstances that exist for the source that would lead to cost-effectiveness estimates that would exceed that for recent retrofits.”
33

PNM did not provide any documentation of unusual circumstances related to the BART determinations in any of its cost analysis.

33

Id.
at 39168.

We subsequently toured the SJGS plant site on May 19, 2011.
34

The SJGS site is congested, but not more so than other space-constrained sites where SCR has been retrofit for much less cost than estimated for SJGS.
35

Gibson, a complex, space-constrained retrofit in which the SCR was built 230 feet above the power station using the largest crane in the world
36

only cost $249/kW in 2010 dollars.
37

Similarly, the Belews Creek SCR, one of the largest and most complex SCR retrofit projects in the U.S., involved installing the SCR 280 feet above ground level above the boiler building. This retrofit only cost $202/kW in 2010 dollars,
38 39

compared to cost estimates of $423/kW to $567/kW for SJGS. B&V's estimates of capital cost to retrofit SCR at SJGS ($446/kW-$599/kW) are higher than actual installed cost for Gibson and many other existing retrofit SCRs, including those with extreme retrofit difficulty. The record including the information we have about the site does not document any unusual circumstances that would justify the unusually high costs claimed by B&V for SJGS. Thus, we do not believe that unusual circumstances are warranted.

34
See San Juan Generating Station Site Visit, 5/23/11.

35
Revised BART Cost Effectiveness Analysis for Selective Catalytic Reduction at the Public Service Company of New Mexico San Juan Generating Station, November 2010, pp. 28-29.

36
Bob Ellis, Standing on the Shoulder of Giants, Modern Power Systems, July 2002.

37
McIlvaine, NO
X
Market Update, August 2004. SCR was retrofit on Gibson Units 2-4 in 2002 and 2003 at $179/kW. Assuming 2002 dollars, this escalates to ($179/kW)(550.7/395.6) = $249/kW.
http://www.mcilvainecompany.com/sampleupdates/NoxMarketUpdateSample.htm.

38
Bill Hoskins, Uniqueness of SCR Retrofits Translates into Broad Cost Variation, PowerGen Worldwide, May 2003. Available at:
http://www.power-eng.com/articles/print/volume-107/issue-5/features/uniqueness-of-scr-retrofits-translates-into-broad-cost-variations.html.

39
Escalated from $145/kW: ($145/kw) (560.3/401.7)-$202/kW. Chemical Engineering, April 2011.

Comment:
The exclusive use of the Cost Manual underestimates the expected costs for SCRs at SJGS for several reasons. First, the Manual was last updated in 2002 and Section 4.2, Chapter 2, Selective Catalytic Reduction, was actually written in October 2000. In addition, on page 2-40 of the SCR section, the Manual indicates that the costs presented are based on 1998 dollars. Therefore, the Manual does not reflect more recent experience with SCR installations, the cost of which has skyrocketed. Second, the 2002 version of the Manual was the very first version to specifically address NO
X
controls at all. According to the introduction to the Manual, EPA was at that time “entering new and uncharted territory for part of the Manual” because “previous editions did not discuss NO
X
or SO
2
controls, and [the 2002] edition starts the process of correcting that oversight.” Finally, EPA also admits in the Manual that it had difficulty obtaining information on control costs because most of the information is proprietary—the very type of information to which B&V has ready access.

Response:
As discussed elsewhere in our response to comments, the Cost Manual contains two types of information, general cost analysis methodology and control-specific costing information. This comment addresses the latter. The information on SCR in Chapter 2 of the Cost Manual contains general information on SCR, design procedures, and some cost information. We agree that the cost information does not reflect current market costs. Thus, cost data should be escalated to current dollars using the CECPI before it is used or replaced with site-specific vendor quotes. We did not use any SCR costs data from this chapter in our analysis.

Comment:
The EPA cost estimate only differs from the Cost Manual where doing so would serve to reduce the amount of the cost estimate. For example, EPA applied an SCR life span of 30 years instead of the 20 year life span provided in the Cost Manual. The justification for choosing a different life span than provided for in the Manual is that other facilities have requested 30 year life spans in requests for proposal and some unidentified SCRs in Europe have lasted that long. If such general, anecdotal information were sufficient to convince EPA to stray from the Cost Manual, the EPA analysis should be replete with variations from the outdated Cost Manual. The use of a 30-year lifespan underestimates the cost estimate of SCR by $15,268,000.

Response:
We disagree with this comment and we used the Cost Manual appropriately, as directed by the RHR. We used it for cost factors that for reasons expressed elsewhere in our response to comments, we feel were miscalculated by B&V, but were not otherwise available in the public domain. We did not use any actual cost data from the Cost Manual. In the case of SCR lifetime, the Cost Manual does not recommend a lifetime for an SCR, but rather sets out a calculation example that uses a lifetime of 20 years. In fact, this same calculation makes many other

assumptions that we felt were not applicable to SJGS and if used anyway, would have resulted in lower cost estimates, but which were not used in our analysis.

The lifetime of an SCR, which is a metal frame packed with catalyst modules, is equal to the lifetime of the boiler, which might easily be over 60 years. The lifetime of a retrofit SCR is generally set equal to the remaining useful life of the facility. The record is silent on the remaining useful life of the SJGS units. Further, USGS studies of the coal reserves upon which the SJGS relies indicate that the local coal supply is adequate to support a remaining useful life of 30 years.
40

Many utilities routinely specify 30+ year lifetimes in requests for proposal and to evaluate proposals. In fact, an analysis prepared by B&V for another facility assumed a 40 year SCR lifetime.
41

And finally, Sargent & Lundy assumed a design life of 30 years
42

for the nearby Navajo Generating Station which burns a similar coal. We conclude there is nothing in the record to support a 20 year lifetime for the SCR and believe a 30 year lifetime is justified.

40
Gretchen K. Hoffman and Glen E. Jones, Coal Availability Study—Fruitland Formation in the Fruitland and Navajo Fields, Northwest New Mexico, USGS Open-File 464, January 24, 2002, Available at:
http://geoinfo.nmt.edu/publications/openfile/downloads/ofr400-499/451-475/464/ofr_464.pdf.

41
E-mail from O'Brien to Van Helvoirt, September 28, 2004, Re: Cost Impact, WPS-011904 at WPS-011905.

42
8/17/10 Salt River Project Navajo Generating Station Units 1, 2, 3 SCR and Baghouse Capital Cost Estimate Report (S&L Navajo Cost Analysis), Appendix A, p. 6, Sec. 1.7.

Comment:
EPA also justifies its refusal to consider additional line items outside the scope of the Cost Manual on the grounds that “PNM had provided no documentation regarding unique circumstances related to the BART determinations.” That claim is incorrect. EPA's own analysis cites the documentation PNM submitted to demonstrate the unique circumstances at SJGS, referred to by EPA as B&V's “Cost Analysis Manual Commentary.” That document was a response to the cost analysis that was initially prepared by NMED in March 2008 as a response to follow-up questions from NMED regarding the BART determination for SJGS. In addition, PNM also provided significant evidence of the site-specific challenges directly to EPA in response to its questions over the several months during which EPA prepared its BART determination for SJGS. Thus, the assertion by EPA that PNM has failed to sufficiently document the site-specific challenges at SJGS is incorrect.

Response:
The specific items in dispute are discussed elsewhere in our response to comments. The information provided in the “Cost Analysis Manual Commentary” and additionally provided to NMED and us explains how B&V extrapolated costs that it estimated from other facilities to apply to SJGS. The alleged unique, site-specific constraints at SJGS, that would justify extrapolating costs from these other facilities, the St. Johns River Power Project, which burns coke, and Harding Street, were never explained. The record, for example, does not contain any structural steel and duct layout drawings to justify this high contingency and other factors, nor does it contain vendor quotes specific to SJGS's coal and site constraints. In fact, as noted elsewhere, we specifically asked PNM to document site specific constraints but they did not respond.

B. Comments on Our Proposed NO
X
BART Emission Limits

We received a significant number of comments concerning our proposed NO
X
BART emission limit of 0.05 lbs/MMBtu for the SJGS. We have summarized our responses to these comments, but refer the reader to our
Complete Response to Comments for NM Regional Haze/Visibility Transport FIP
document for more detail.

Comment:
PNM stated the BART limit should not be based on daily averages of thirty (30) calendar days, as we proposed, because it believes it would be inconsistent with the BART Guidelines. If calendar days are used, they argue, the average could include as little as one hour of operation if the unit is offline for an outage that lasts longer than thirty days because the first hour of operation would be the only data recorded in the last thirty calendar days. Instead, PNM requested that we consider changing “calendar days” to boiler operating days (BODs) which are days in which the unit ran for at least one hour. That approach would be consistent with the BART Guidelines, which include the following advice to states:

For EGUS, specify an averaging time of a 30-day rolling average, and contain a definition of “boiler operating day” that is consistent with the definition in the proposed revisions to the NSPS for utility boilers in 40 CFR part 60, subpart Da.
43

43
70 FR 49104, 39172.

The BOD would ensure that, when an outage occurs, the emissions following startup will be averaged with the emissions data from before the outage, rather than with the period of time during which the unit did not have any emissions at all because it was offline.

Response:
We agree with this comment that our proposed NO
X
emission limit should be based on BODs, rather than a straight calendar average. In response to this comment, we have reanalyzed our proposed determination that the units of the SJGS can achieve a NO
X
emission limit of 0.05 lbs/MMBtu on a continuous basis, using the BOD concept. We have done this because we believe the same metric should be used to both determine BART and to determine compliance with BART. The results of that analysis are presented in response to another comment. In summary, we continue to believe that NO
X
BART for the units of the SJGS is an emission limit of 0.05 lbs/MMBtu. We have concluded that emission limit should be based on a 30-day BOD rolling average based on any operation in a given day counting toward the average. We believe that averaging scheme complies with the BART Guidelines, which defines a BOD to be “any 24-hour period between 12:00 midnight and the following midnight during which any fuel is combusted at any time at the steam generating unit.”
44

44

Id.

Comment:
The U.S. Forest Service (USFS) expressed its support of our NO
X
BART emission limit of 0.05 lb/MMBtu. The USFS believe this emission limit is adequate and will improve visibility at Class I areas throughout the Four Corners region. Additionally, the USFS feels SCR has already been determined to be BART at several other coal-fired power plants across the United States.

Response:
We agree with the USFS.

Comment:
EPA predetermined the cost-effectiveness of SCR at SJGS “assuming an outlet NO
X
of 0.05 lb/MMBtu.” EPA then proposed that assumed rate as the BART emission limit for SJGS. EPA's assumption is unfounded—the installation of SCRs at SJGS will not enable the units to achieve 0.05 lb/MMBtu on a continuous basis. As such, the proposed 0.05 lb/MMBtu limit cannot be BART for SJGS.

Response:
We disagree with this comment. We initially estimated the cost effectiveness of SCR, assuming an outlet NO
X
of 0.07 lb/MMBtu, to provide a direct comparison with B&V's analysis. Following this, we determined that a BART emission limit of 0.05 lb/MMBtu was appropriate and then refined the cost effectiveness on that basis. The BART level of 0.05 lb/MMBtu was selected based on an examination of continuous emission monitoring

systems (CEMS) data for existing units operating with retrofit SCRs, as we explain elsewhere in our response to comments.

Comment:
In contrast to EPA's NO
X
emission limit assumption of 0.05 lbs/MMBtu, B&V, who has extensive practical experience in actually designing and installing retrofit SCRs determined that a retrofit SCR would only be capable of achieving 0.07 lb/MMBtu on a continuous basis, particularly if required to use the low-oxidation catalyst assumed by EPA to minimize ancillary emission increases associated with SCR.

Response:
We do not believe the claim that B&V “determined that a retrofit SCR would only be capable of achieving 0.07 lb/MMBtu on a continuous basis * * *” is supported in the record by any calculations or arrangement drawings. Rather, the 0.07 lb/MMBtu value is simply stated in the initial June 6, 2007 B&V BART analysis without any explanation as to how it was determined or why 0.07 lb/MMBtu satisfies BART rather than a lower limit.
45

The basis for this limit has been questioned by NMED, the NPS and us since July 2007, but we do not believe that PNM has provided adequate supporting analysis. We do not view an unsupported statement, such as this, questioned on the record by many parties and inconsistent with retrofit SCR experience at numerous facilities, to be sufficient to support a BART determination of 0.07 lb/MMBtu.

45
6/7/07 B&V BART Analysis, Table ES-2, Table 2-3, Table 6-1, 7-1.

We note the NO
X
design basis was 0.05 lbs/MMBtu for the SCR retrofit for the nearby Navajo Generating Station, a facility of a similar age that burns a similar coal, with a more constrained site. As explained elsewhere in our response to comments, we present data that demonstrates that retrofit SCR installations are capable of achieving a NO
X
limit of 0.05 lbs/MMBtu on a continuous basis. Therefore, we believe the statement that a retrofit SCR would only be capable of achieving 0.07 lb/MMBtu on a continuous basis, is factually incorrect.

Comment:
Several commenters stated that our claim that many facilities are using SCR to actually achieve lower emission rates than 0.07 lb/MMBtu (including the Havana Unit 9, Amos Units 1 and 2, Chesterfield Unit 6, Cardinal Units 2 and 3, Colbert Unit 5, Ghent Units 3 and 4, and Mill Creek Unit 3) is incorrect. This commenter states that while these units have shown the ability to reach 0.05 lb/MMBtu or lower at times, those units are unable to do so on a continuous basis. Thus, the commenter claims, if the units cited by us were in fact subject to a 0.05 lb/MMBtu emission limit, those limits would have been violated many times at each unit.

Response:
We disagree with this comment and continue to believe that the NO
X
emission limit we proposed for the four units of the SJGS, 0.05 lbs/MMBtu, is achievable on a continuous basis. In reaching this conclusion, we followed the language in the BART Guidelines:

It is important, however, that in analyzing the technology you take into account the most stringent emission control level that the technology is capable of achieving. You should consider recent regulatory decisions and performance data (
e.g.,
manufacturer's data, engineering estimates and the experience of other sources) when identifying an emissions performance level or levels to evaluate.

In assessing the capability of the control alternative, latitude exists to consider special circumstances pertinent to the specific source under review, or regarding the prior application of the control alternative. However, you should explain the basis for choosing the alternate level (or range) of control in the BART analysis. Without a showing of differences between the source and other sources that have achieved more stringent emissions limits, you should conclude that the level being achieved by those other sources is representative of the achievable level for the source being analyzed.
46

46
70 FR 39104, 39166.

First, we examined “the most stringent emission control level that technology [SCR] is capable of achieving.” As demonstrated below, we concluded that SCR is capable of achieving a NO
X
emission limit of 0.05 lbs/MMBtu. Second, we examined the record to determine if there existed “special circumstances pertinent to the specific source under review” that would prevent the units of the SJGS from achieving this limit, and found none. Third, concluding there was no “showing of differences between the source and other sources that have achieved more stringent emissions limits” that would preclude the application of this limit, we “conclude[d] that the level being achieved by those other sources is representative of the achievable level for the source being analyzed.” The following discussion expands on these points.

In our
Complete Response to Comments for NM Regional Haze/Visibility Transport FIP
document, we provide a detailed discussion of why we believe the commenter, PNM, misquotes our cost evaluation report, which was incorporated into our proposal's TSD. In summary, that report contained a previous study of SCR performance during the ozone season for the period 2003-2006. This study showed that several units were achieving a NO
X
emission limit of 0.05 lb/MMBtu at that time to meet NO
X
SIP Call regulations that were then in force. These SCRs only operated from May to October of each year, the ozone season. The SCRs were bypassed during the remainder of the year as they were not required to meet the NO
X
SIP Call.

PNM presents graphs for each of the ozone season 2003-2006 units for the period January 2008 to November 2010. These graphs suggest that 0.05 lb/MMBtu is exceeded on numerous occasions and imply this was due to a limitation of the equipment to maintain control. However, these graphs appear to be based on calendar operating days. This distinction is significant, as the BOD convention discussed by the BART Guidelines
47

smoothes out the 30-day rolling average outage spikes. Also, these charts include large blocks of time during which the SCRs were turned off because they were not required under the trading programs then in force. Lastly, these charts connect the dots across outage periods, when the SCRs are not in use and improperly include the zero hour days in the averages at elevated levels.

47

Id.
at 39172.

To address this, we analyzed data from EPA's Clean Air Markets Division (CAMD), which compiles CEMS data reported under various trading programs. We analyzed the NO
X
CEMS data for the period 2009-2010 to identify the best performing retrofit units that operate year-round. We ranked the annual average NO
X
emissions for all units in the database for the years 2009 and 2010 from the lowest to the highest NO
X
emissions. We then selected those facilities that had at least one unit in the top 30 group in both years to identify retrofits achieving best performance.

We then developed a spreadsheet program that used the CAMD data and calculated and graphed three types of 30-day rolling averages for most of these best performing units, plus those additional units graphed by PNM for the period 2008-2010 for the Ozone Transport Assessment Group (OTAG) units and 2006-2010 for the Texas units (Parish 7, 8). All of the units we analyzed were retrofitted with SCR.

As Exhibit 2 shows,
48

the averaging conventions we used are: (1) A conventional 30-day calendar rolling average; (2) a 30-day BOD rolling average based on any operation in a given day counting toward the average; and (3) a 30-day BOD rolling average based on only full 24-hour days. We believe that averaging scheme (2) complies with the BART Guidelines, which defines a BOD to be “any 24-hour period between 12:00 midnight and the following midnight during which any fuel is combusted at any time at the steam generating unit.”
49

48
Exhibit 2, Best Performing SCR Retrofit Installations, June 8, 2011.

49
70 FR 39104, 39172.

The Havana Unit 9 data shows that it has operated under 0.05 lbs/MMBtu from mid-2009 to the end of 2010 on a continuous basis. In fact, this unit has operated under 0.035 lbs/MMBtu for much of that time. The Parish Unit 7 data shows that it has operated under 0.05 lbs/MMBtu from mid-2006 to mid 2010 on a continuous basis. In fact, this unit has operated for months at approximately 0.035 lbs/MMBtu, and for approximately 2 years at approximately 0.04 lbs/MMBtu. The Parish Unit 8 data show that it has operated almost continuously under 0.045 lbs/MMBtu since the beginning of 2006. Other units' data show months of continuous operation below 0.05 lbs/MMBtu. We believe this data demonstrates that similar coal fired units that have been retrofitted with SCRs are capable of achieving NO
X
emission limits of 0.05 lbs/MMBtu on a continuous basis.

In addition, it is important to note that most of the NO
X
CEMS data in the CAMD database is generated under cap and trade programs, such as

[Text truncated at 120,000 characters. The full text is on the page linked above.]

---

Source: Frix Law Library, https://www.frixlaw.com/law-library/documents/fr%3A2011-20682. Public record. Not legal advice.