From: 
Zavoda, Rich <Rich.Zavoda@arcelormittal.com>
Sent:
Sunday, February 9, 2020 6:49 PM
To: 
Jones, DonnaLee <Jones.Donnalee@epa.gov> (Jones.Donnalee@epa.gov) <Jones.Donnalee@epa.gov>; Raymond, Gabrielle <graymond@rti.org>
Cc:
Zavoda, Rich <Rich.Zavoda@arcelormittal.com>; Skolnekovich, Danielle L <Danielle.Skolnekovich@arcelormittal.com>; Benoit, Simonne T <Simonne.Benoit@arcelormittal.com>
Subject: 
FW: September 2019 QA/QC Review of Enclosure 2 Information for EPA 2016 ICR for ArcelorMittal (Excel file) Monessen


Dr. Jones

Thanks for the opportunity to review the draft information.  Attached is the updated Coke ICR Enclosure 2 excel file for ArcelorMittal Monessen. Please note that ArcelorMittal Monessen (AMM) is an area source and the Pushing, Quenching, Battery Stack MACT is not applicable.  We reserve the right to provide additional information in the future as may be appropriate. 

The attached file entitled AM-MonesPA-CokeOvens-Enclosure2-QA-Emissions-Data-Sept-2019-IndustryQA_AM-Monessen-PA_ChangestoEPA.xlsx includes the requested AMM changes to USEPA summary of test data in tab A-1. Stack Air Stack Tests.   The requested changes to the file are summarized below in detail.  USEPA requested that all changes be highlighted in yellow and described in Column BJ.  All changes were highlighted in yellow, but summarizing the changes in Column BJ was not reasonable because there are so many changes.  

A-1. Air Stack Tests  -  Requested Changes to USEPA's Spreadsheet
 General
 Latitude/Longitude does not match what was reported in the ICR Enclosure.  The latitude/longitude was changed to those coordinates submitted in the ICR enclosure to USEPA.     
 USEPA is including non-detectable values in the reportable values as the full value.  If the constituent's concentration was found to be non-detectable, (1/2) the detection limit was used as the concentration.  This was done for all of the emission sources reported.  The resulting concentration was then used to calculate the mass emissions (lb/hr).  
 USEPA noted that the following D/F constituents were not reported: Total OCDD, Total OCDF.  Per the stack testing firm, USEPA requested Tetra through Hepta.  If EPA is asking why we didn't report Total Octa-dd and Octa-df, it was because it wasn't requested by USEPA.     

 Battery Combustion Stack #2
 USEPA is asking for PM2.5 filterable and PM2.5 condensable (highlighted orange in rows (39-44), TSO (Rows 174-176), SO2, CO and HCl (Rows 183-191).  In the test report (page 8 of pdf), it states that USEPA approved previously submitted testing which exempted them from conducting additional testing for TSO, HCl, PM2.5, condensable PM, SO2 and CO at the Battery 2 Combustion Stack.   Therefore, testing was not completed for the aforementioned constituents. 
 Table 6 of ICR report (SW 846 Method 0031 Semi-volatiles), Runs 2 and 3 lb/hr were found to be in error.  The stack test company concurred in writing with this error.  The error was corrected, and the updated values, taking into consideration the non-detectable approach discussed in 1.b above, has been corrected in USEPA's table.

 Pushing Baghouse
 Method 5/29 testing: Elapsed Hours for Runs 2 and 3 incorrect  -  there is an incorrect formula in the cells. (M513:M548).  ArcelorMittal made the necessary changes to USEPA's table.  
 Mercury  -  USEPA notes that fraction 3A is missing.  Per the stack test company, the lab combined the rinses instead of analyzing separately.  On page 16 of 22 on the lab report, it says "Mercury 1B in Filter + Rinse (M29): Due to lab error this front half fraction contains nitric rinses from the backhalf impinger.  Mercury 1B in Filter + Rinse (M29): Due to lab error this FH fraction contains nitric rinse only from the front half."
 H2S and CO2 concentrations are incorrect in USEPA's table (BB789:BD794).  The necessary changes were made to USEPA's table, taking into consideration non-detects as noted in 1b.   
 CO2 lb/hr (BE792:BE794).  USEPA inputted tons/hr instead of lbs/hr.  The necessary changes were made to USEPA's table.  
 USEPA is asking for PM2.5 filterable and PM2.5 condensable (highlighted orange in rows (549-39-554, TSO (Rows 786-788), SO2, CO and HCl (Rows 795-803).  In the test report (page 8 of pdf), it states that USEPA approved previously submitted testing which exempted them from conducting additional testing for TSO, HCl, PM2.5, condensable PM, SO2 and CO at the Pushing Baghouse Stack.  Therefore, testing was not completed for the aforementioned constituents. 
 Table 6 of ICR report (SW 846 Method 0031 Semi-volatiles), Runs 2 and 3 lb/hr were found to be in error (similar to combustion stack).  The stack test company concurred with this error.  The error was corrected, and the updated values, taking into consideration the non-detectable approach discussed in 1.b above, has been corrected in USEPA's table.
 Dioxins/Furan.  USEPA had different values than what was in the ICR Test Report for the following D/F constituents.  It appears USEPA has re-calculated the D/F calculations in tab called A.3 DF TEQ.
 All lb/hr values were different in USEPA's table than the values in Table 3 of the ICR Test Report. USEPA is incorrect and the test report is correct.  USEPA used 60 min/hr to derive the lb/hr, instead of using the pushing min/hr.  ArcelorMittal made the necessary corrections to USEPA's table, takin taking into consideration the non-detectable approach discussed in 1.b above, has been corrected in USEPA's table.
 The concentrations for 1,2,3,4,7,8,9-Heptachlorodibenzofuran (Run 1) and 1,2,3,4,7,8,9-Heptachlorodibenzofuran TEQ (Run 1) in USEPA's  do not match what is in the ICR Test Report.  Looking at lab report, it appears the ICR test report is correct and EPA is incorrect.  Page 193 of report shows 3.3 pg, while USEPA typed in 3.7 pg in their spreadsheet.  ArcelorMittal made the necessary changes to USEPA's table, taking into consideration the non-detectable approach discussed in 1.b above, has been corrected in USEPA's table.  

 Quench Tower
 Column AP has Coal Charged during test run.  The formula is incorrect  -  USEPA is referencing the dry coal charged per oven (16.2) divided by the test duration.  It should be the dry coal charged per oven x the number of ovens charged per testing divided by the test duration.   The necessary changes were made to USEPA's table.  
 USEPA is denoting Method 5 particulate (Rows 810-812) as PM2.5 Filterable,  However, it includes all filterable particulate, PM2.5 is just a portion of the filterable PM.  In USEPA's table, the pollutant name from PM2.5 Filterable to PM Filterable was changed.  
 Number of quenches in the ICR test are not correct and ArcelorMittal is requesting that the values in Column O be removed.   ArcelorMittal is also requesting that the average numbers of quenches per hour of 3.3 be used for all test runs (Column P).  
 USEPA has calculated a mass per hour (lb/hr) for all pollutants for quenching, based on 60 minutes/hr and full detection level.  ArcelorMittal is requesting that the estimated lb/hr be calculated using 3.3 quenches/hr and 5 minutes/quench, for a total of 16.5 minutes of quenching per hour.  The values using this approach have been incorporated into the USEPA's  
 Dioxins/Furans
 Data input for CO2 for Runs 2/3.  Should be 0.16 and 0.11 instead of 0.016 and 0.011 (Column AY).  The necessary changes were made to USEPA's table.  
 For all D/F concentrations, the concentration values were different in USEPA's table compared to values in Table 4 of the ICR Test Report.  It appears that USEPA copied the weights (pg) from pushing runs in their tab entitled A.3 DF TEQ.  AY176:BB194 (Pushing) had the same values as AY252:BB270 (Quenching).  The stack test company also had errors on Table 4 of the ICR Test Report for 1,2,3,7,8-Pentachlorodibenzofuran TEQ (all runs), 2,3,7,8-Tetrachlorodibenzofuran (all runs), 2,3,7,8-Tetrachlorodibenzofuran TEF (all runs), 1,2,3,7,8-Pentachlorodibenzofuran TEQ (all runs), 2,3,4,7,8-Pentachlorodibenzofuran TEQ (all runs).  ArcelorMittal made the necessary changes to USEPA's table, taking into consideration the non-detectable approach discussed in 1.b above, has been corrected in USEPA's table

 Quench Tower Test Invalid Air Flow Data and Mass Per Time Calculations
      The Montrose stack test report for the ArcelorMittal Monessen Quench Tower test conducted October 25-November 2, 2016 stated, "USEPA was pre-notified and agreed to non-isokinetic testing to be conducted on one-half of the quench tower using twelve total traverse points from four equal distance ports located only on one side of the quench tower per USEPA September 13, 2016 "Frequently Asked Questions (FAQs) and Response for Coke Ovens Information Collection Request (ICR)," as noted in questions 13, 17 and 26. A copy of the USEPA September 13, 2016 FAQs and Responses that includes the questions and responses to 13, 17, and 26 is attached to this report.  A temporary platform had to be erected on the west side of the quench tower to add four ports to conduct the ICR testing.  USEPA ICR testing was required to be conducted above the last baffle, which did not allow for a suitable test location that meets minimum USEPA Method 1 upstream and downstream disturbance dimension requirements. USEPA was pre-notified of the unique circumstances associated with quench towers that commonly have particulate removing baffles installed near the top of the quench tower.  USEPA agreed to the ICR testing using non-standard test methods to be conducted at twelve points, recognizing the testing location did not meet minimum upstream and downstream requirements of USEPA Method 1. 
      
      Additionally, isokinetic testing is not believed to be possible at a Quench Tower due to the flow dynamics of the process. When steam is first sprayed onto the hot coke car, and the initial (high) velocity point was measured and recorded with pitot tube static pressure readings as required.  As the coke is cooled during the quench cycle duration of five minutes, the non-steady steam flow velocity decreases from the initial high value to zero. Following the standard USEPA Method 2 pressure reading frequency requirements, the normal quench cycle duration was not long enough to collect subsequent readings at the standard frequency that would have recorded lower values than the initial reading.  Applying the initial high velocity reading to the entire quench cycle, per the test methods, would be technically flawed and result in very high biased and inaccurate flow calculations that do not take into account the observed decreasing velocity over the entire quench cycle that lasted five minutes. Unlike other sources that have a steady fan induced air flow, the non-steady air flow from the quench tower originates from the thermal buoyancy from the convection heat transfer from the water being applied to the hot coke which decreases from the start to the end of the coke quenching process. 
      
      Based on the inability to use standard USEPA testing methodologies for the quench tower testing that provided non-standard results from non-isokinetic sampling, known high bias flows and the use of a non-standard USEPA Method 1 upstream and downstream flow measurement testing methods, as disclosed to and accepted by USEPA in advance of testing, this report includes the individual constituent results expressed in concentrations only, since accurate mass calculations are not possible due to inaccurate flow calculations.
      
      A cyclonic flow check could not be performed due to safety considerations and was acknowledged by USEPA September 13, 2016 FAQs and Responses to question 13.  Personnel are not allowed to be on top of the quench tower during a quench due to safety concerns of steam and scalding risks, and no flow occurs in between quench cycles when personnel have safe access to the top of the tower."
      
      Based on the facts listed above in the Montrose quench tower test report, mass per time emission calculations would be inaccurate and greatly overstate emissions. Although we are providing comments and corrections to these quench tower test report excel calculations provided by USEPA, ArcelorMittal agrees with the Montrose report conclusion and statements that did not attempt to estimate mass per time calculations due to the USEPA approved non-standard testing methodology and flawed flow rate data results.

 Toluene Soluble Organic (TSO) Lab Contamination
      ArcelorMittal received the 2016 Coke ICR from USEPA on April 4, 2016.  The 2016 Coke ICR included the use of Method 315 (Particulate and Methylene Chloride Extraction Matter (MCEM), except that USEPA specifically required that the standard method extraction solvent of methylene chloride be substituted with the use of toluene to determine emissions of Toluene Soluble Organics (TSO).  The attached March 24, 2017 Maxxam laboratory evaluation (pages 35-39) concluded that the use of toluene (instead of methylene chloride, as the extraction solvent of the filter and probe rinse, resulted in significantly elevated toluene filter extraction blanks and toluene laboratory processed blank results that cannot be related to the sampled sources and must be considered when attempting to report the actual Method 315 values.  See the previously submitted individual stack test reports for additional TSO lab contamination details, including the Quench Tower Test report which is attached for your convenience. 

Rich Zavoda | Regional Manager, Air Quality Program
ArcelorMittal USA
Environmental Affairs| 4020 Kinross Lakes Parkway, Richfield, Ohio 44286-9000
T +1 330 659 9163 | F +1 330 659 7434 | www.arcelormittal.com

<< ArcelorMittal Monessen Quench Tower ICR Report 4.13.17 pgs 1_42.pdf >>
<< AM-MonesPA-CokeOvens-Enclosure2-QA-Emissions-Data-Sept-2019-IndustryQA_AM-Monessen-PA_ChangestoEPA.xlsx >>

From: Jones, DonnaLee <Jones.Donnalee@epa.gov> 
Sent: Wednesday, September 4, 2019 5:05 PM
To: Zavoda, Rich <Rich.Zavoda@arcelormittal.com>
Cc: GABRIELLE RAYMOND <graymond@rti.org>
Subject: September 2019 QA/QC Review of Enclosure 2 Information for EPA 2016 ICR for ArcelorMittal (Excel file)

Rich  -  Thank you again for your efforts in providing your facility information. Now is the time to correct, update, or add to your responses from 2016 for Enclosure 2. Attached is your facilities' information in one Excel file each.

Please refer to the `Review Instructions' worksheet of the attached spreadsheet for specific review instructions.
The Enclosure 2 data are organized into three groups within the spreadsheet:
A. Air Stack Tests
B. Process Data
C. Non-Stack Test Data (e.g., Opacity, Water Tests, Material Tests, COG Tests, etc.)

Resources worksheets (tabs R-1 through R-4) also are available for your reference and use. 
Please review all data compiled by the EPA Contractor. If your data is not correct, please revise or add it to the applicable worksheet. Highlight (in yellow) any changes made with any cell and describe the change in the "Facility Review Comments" column of the applicable worksheet.
If any of the currently missing information is considered "confidential business information" (CBI), please note "CBI" in the empty data cell. Then please provide this CBI information using the CBI procedures described in the original ICR letter sent by EPA in 2016. We can provide again if needed.
If you have any questions, please first send an email to BOTH Gabrielle Raymond, the EPA contractor from RTI, and me at the email addresses below. If we need to speak on the phone, we can follow-up with a call, but we need an email first to ensure contact.
graymond@rti.org
jones.donnalee@epa.gov  

We will be entering your Enclosure 2 data into the Coke oven RTR subpart CCCCC Modelling file soon after the first of the year 2020, so we need to get all the QA/QC work done during the Fall 2019. 
Note that we will be summarizing CBI data in Enclosure 1 separately at some not yet determined point in the future. 
Thank you in advance for your work, time and expense helping us get good quality data. 
_____________________________________________________________
Regards,
Donna Lee Jones, Ph.D.
Senior Technical Advisor, Metals Sector
U. S. Environmental Protection Agency
Office of Air Quality Planning and Standards
Sector Policies and Programs Division / Metals & Inorganic Chemicals Gr (D 243-02)
Research Triangle Park, NC  27711  Tele:  (919)  541-5251  Fax  (919)  541-3207
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
"Reasonableness never fails to be appreciated."  - anon.





