On February 3, 2005 I Received a Package That Included Proposed Plan Approval Conditions

USDAForest Service

Technical Review of the Robinson Power Company

Prevention of Significant Deterioration Application and Draft Plan Approval

March 2005

Robinson Power Company (Robinson) is a 272 megawatt (MW) waste coal fired, circulating fluidized bed (CFB) boiler to be located in Robinson Township, Washington County, Pennsylvaniaabout 12 miles west of Pittsburgh. The project site contains about 37 million tons of gob, covering about 350 acres. Gob will be delivered to the boiler by conveyor, although truck delivery may occur as a back-up. Coarse material will be crushed using a rotary breaker located at the gob pile. The material (blended coarse gob and fines) has a heating value around 4,600 Btu/lb.

Sulfur dioxide (SO2)emissions would be controlled to 3,154 tons per year (tpy) by the SO2 removal inherent in a CFB, plus the use of a Dry Limestone Scrubber. Nitrogen oxide (NOx) emissions would be controlled to 972 tpy using Selective Non-Catalytic Reduction (SNCR). Total PM10would be limited to 146 tpy by a fabric filter (baghouse). The project is also subject to Prevention of Significant Deterioration (PSD) for sulfuric acid mist (36 tpy) however no additional controls are proposed for this pollutant.

One of the purposes of the Prevention of Significant Deterioration (PSD) program is to “preserve, protect, and enhance” the air quality in national parks and wildernesses designated as Class I areas. The Clean Air Act (CAA) gives the Federal Land Manager (FLM) an affirmative responsibility to protect the air quality related values (AQRVs) of the Class I areas. Robinson has the potential to impact air quality and AQRVs at two Class I areas administered by the Monongahela National Forest (MNF), Dolly Sods and Otter Creek Wildernesses. Dolly Sods is located 170 kilometers, and Otter Creek 160 kilometers, to the south-southeast of the proposed facility.

In fulfilling our CAA responsibility we have reviewed the PSD application and draft plan approval to determine whether or not the emissions from Robinson will have an adverse impact on any of the AQRVs in the class I areas. The air quality related values at these areas include visibility, vegetation and water. We have also reviewed the class I increment consumption analysis.

The applicant conducted the modeling analysis, using the CALMET/CALPUFF dispersion-modeling package, to assess potential incremental increases in pollution and impacts to air quality and AQRVs in the class I areas. Several iterations of the class I air quality analyses have been provided to the FLM, but our comments are based on therevised results provided to Mr. Sean Nolan of the PADEP by Mr. Larry Simmons on March 24, which reflect reductions in the 3-hour SO2emission rate to 1115 lb/hr, and 24-hour and annual SO2 emission rates to 0.245 lb/MM Btu (675 lb/hr). We reviewed the modeling and verified that all previous errors have been corrected in these most recent runs.

Class I Increment Consumption

The Robinson PSD increment modeling results are summarized in Table 1. Concentrations above the class I significant impact limits for PSD increment consumption were modeled at both Otter Creek and Dolly Sods. Robinson, alone, exceeds the class I increment significant impact levels for short-term sulfur dioxide (SO2) concentrations (both 3-hour and 24-hour averages). For other pollutants, particulate matter (PM-10) and nitrogen oxides (NOx), and annual mean SO2 concentrations, the maximum impacts are less than their respective significant impact levels. Because the SO2 class I increment significant impact levels were exceeded, the need for a cumulative CALPUFF modeling analysis was triggered.

Table 1. Class I PSD increment modeling results (March 16, 2005) showing maximum impacts reported in micrograms per cubic meter.

The class I increment modeling analysis is very important to protecting AQRVs. When increment is violated, it must be corrected through emission reductions. Therefore, it is most beneficial to the class I area if the analysis accurately identifies all sources contributing to the violation.

We provided PADEP with comments on the cumulative increment analysis for Greene Energy, another PSD project currently in the permitting process, in a letter dated February 18, 2005. The Greene inventory was used as a basis for the Robinson cumulative increment inventory. The February 18 letter and technical support document are attached as Appendix A of this report, and the concerns expressed therein apply to the Robinson cumulative assessment as well. Additionally, the Robinson inventory raises one cumulative inventory issue that we did not address in the February 18 letter. This pertains to the size of the modeling domain used in a class I cumulative increment analysis. It is common that a domain used in single source modeling is smaller than a domain that would be considered appropriate for a cumulative analysis, and yet we have never seen an applicant expand the modeling domain when they are required to conduct a cumulative analysis. This can lead to the omission of sources that should be considered. For example, Robinson omitted some sources (particularly several in eastern Ohio that PADEP added to the Greene inventory) from the inventory because they were outside the established modeling domain. Our suggested guidelines for a cumulative analysis call for inclusion of all appropriate sources within 300 kilometers of the class I areas. We also suggested screening criteria that would eliminate sources from the inventory based on magnitude of emissions and distance from the class I areas, but some large sources could still be included that are 200-300 kilometers out. This may require expansion of the single source modeling domain to accommodate the cumulative inventory.

As a result of FLM comments, PADEP directed the applicant to make several changes to the inventory used in the cumulative analysis and to expand the modeling domain. While the inventory does not match exactly our suggestions in the February 18 letter, it does demonstrate some improvement and we feel that development of a complete increment inventory is now the responsibility of the affected states and EPA. It is unclear to us whether or not the applicant incorporated the changes to the cumulative inventory requested by PADEP in the cumulative increment modeling received March 24, 2005.

The results of the cumulative increment analysis, received March 24, 2005, indicate that that short-term increments are violated at Dolly Sods Wilderness but Robinson is not a significant contributor.

Impact to Air Quality Related Values

Of the many AQRVs FLMs are charged with protecting, visibility is the one AQRV that was specifically mandated by Congress. The assessment of visibility impacts to the Class I areas is therefore a very important one. The applicant’s assessment predicts the maximum daily change in visibility that could be caused by emissions from the Robinson project. The applicant followed the FLAG guidance in conducting the visibility analysis, with the exception of setting relative humidity at 95%, rather than 98%. The FLMs feel this is reasonable modification and agreed to it early in the original modeling protocol.

Emissions from Robinson are predicted to result in a maximum increase in light extinction at Dolly Sods of 12.58%, with 2 days in 3 modeled years showing greater than 5% change, and 1 day in 3 modeled yearsshowing greater than 10% change. At Otter Creek, the maximum increase in light extinction was predicted to be 11.71%, with 10 days over 3 years showing over 5% change, and 1 day in 3 modeled years showing over 10% change. A change of 10% and greater is considered to be perceptible by the general public, and changes above 5% are perceptible to a smaller portion of the population.

Section 160 of the CAAA states the purposes for the PSD program; and one of the five is to “preserve, protect and enhance the air quality” in “national parks” and “national wilderness areas”. Further, the FLM is directed by Congress “in cases of doubt” to “err on the side of protecting the air quality related values for future generations.” (Senate Report No. 95-127, 95th Congress, 1st Session, 1977). Fortunately, we know that emission reductions will result in visibility improvement, as we have seen some positive changes associated with the sulfur dioxide emission reductions resulting from implementation of the 1990 Clean Air Act Amendments (CAAA). However we are far from the national goal for visibility specified in the CAAA and despite improving trends at Dolly Sods and Otter Creek, visibility is still considered impaired. In an effort to achieve the purposes of the PSD program, we must seriously evaluate modeled impacts that show a perceptible degradation in visibility, particularly in class I areas where visibility is already impaired. Therefore, we anticipate that Robinson will cause an adverse impact to visibility at Dolly Sods and Otter Creek Wildernesses.

Table 2. Class I visibility modeling results (March 17, 2005) showing percent change in light extinction from Robinson using RHMAX=95%.

Two additional AQRVs, water and vegetation, could be affected by Robinson’s emissions. It is well known that stream water in many parts of West Virginia, but especially Dolly Sods and Otter Creek, is acidic. The current situation is the result of a combination of underlying geology that lacks buffering capacity and historically high sulfur deposition patterns. These same factors are also affecting soils. As soils acidify several changes take place; among these are the potential loss of nutrients needed by vegetation and the mobilization of aluminum that is toxic to roots and soil biota. Leachates from the soil then move into stream water and affect the aquatic ecosystem. Obviously the effects of acid deposition on streams, soil and vegetation within a watershed are linked and affect one another.

Annual emission rates were modeled to predict the impact of Robinson’s emission on sulfur and nitrogen deposition at the class I areas (Table 3.). We found that annual sulfur deposition from Robinson would not cause measurable decreases in the acid neutralizing capacity of the most sensitive streams in both Otter Creek and Dolly Sods Wildernesses. Based on these results we do not anticipate that emissions from Robinson will cause adverse impacts to the aquatic and terrestrial ecosystems at Otter Creek and Dolly Sods Wildernesses.

Table 3. Maximum annual deposition of sulfur and nitrogen from Robinson’s emission at the class I areas in West Virginia.

Comments Specific to Plan Approval Conditions

Plan approval condition 8.1. PADEP has proposed condition 8.l which states “The owner/operator shall demonstrate, to the Federal Land Manager, that the emissions from this source, do not adversely affect any Class I area's air quality related value (AQRV). [40 CFR Part 52]”. As written, the current condition does not meet our expectations. The condition does not describe the mitigation plan or how it will be determined. It does not indicate how the mitigation will be evaluated and it does not provide any avenue to enforce the mitigation agreed to. PADEP has rewritten the condition and it now adequately covers any mitigation plan that may be needed. The new version of condition 8.1 should be put in the final plan approval.

Emission rates: Emission rates in the conditions should be at least equal to, or less than, the modeled emission rates. There have been a number of changes to the emission rates which need to be reflected in the final plan approval.

Monitoring: We recommend that PADEP add a condition to the final plan approval requiring installation of a PM continuous emission monitor (CEM) upon start up. Continuous emission monitors are an important tool for monitoring compliance, and performance specifications for PM CEMs were promulgated by EPA on December 12, 2004.

Summary and Recommendations

• Modeling results indicate adverse impacts to visibility.
• Modeling results indicate no adverse impacts to aquatic and terrestrial ecosystems.
• Ensure that BACT for SO2, NOx, and PM10is applied, because this would reduce impacts to class I increments and air quality related values.
• PADEP should ensure that emission rates in the final plan approval are at least equal to, or less than, the modeling emission rates for SO2, NOx, and PM10.
• Recommend a change to permit condition 8.1 which addresses mitigation plans.
• Recommend addition of a Continuous Emissions Monitor for particulate matter.

1