          FHR and Molex Response to 5/13/22 Conference Call with EPA
                                       

 Bump Tests  -  FHR/Molex reviewed the bump test results from the Mid-crude and m-Xylene pilot studies to determine how many times the responsivity tests passed using the 50% response threshold but would have failed if the threshold had been set at higher levels. 
   
   The bump test results are included in the attached Excel file, Bump Test Data from 2019 Pilot and a summary of the data review is shown in the table below. 

Response Threshold 
50%
55%
60%
65%
70%
Pass Rate
97.3%
91.7%
83.8%
75.2%
63.7%
	
   As should be expected, the pass rate decreases with increasing P/F threshold values. If the P/F threshold value were increased from 50% to 55%, an additional 5.6% of the bump tests conducted during the pilot run would have failed. If the P/F threshold value were increased to 60%, an additional 13.5% of the bump tests would have failed. Each additional failure triggers the need to replace the sensor with a pre-calibrated one. 
   
   It is important to note that all the research projects and pilots conducted with EPA ORD utilized a 50% minimum response threshold based on which  equivalency was demonstrated to the EPA M21 CWP. Increasing the minimum threshold for bump tests would directly result in an increase in the number of sensors used, the number of back-up spares required to be kept on-site, and correspondingly an increase in the ongoing cost to operate the system.

   
 Detection Floor  -  Provide data showing minimum detection floor calculation results for Mid-crude and m-Xylene during the pilots using 2 minute, 5 minute, and 10 minute averages.
   
   The detection floor calculation results are included in the attached Excel file, Detection Floor Data from 2019 Pilot.

   When a sensor displays continuous peak detections of gas, a clean baseline is not available for the calculation of the sensor detection floor per EPA's detection floor definition. FHR and Molex have proposed the sensor detection floor values to be reviewed at the same time each day to confirm each sensor detection floor remains below established threshold of 10 ppbe on a rolling 2-minute average during at least one 2-minute period in the past 72 hours. If a sensor does not pass the detection floor review, then a sensor fault notification shall be issued, and the sensor issue shall be corrected through recalibration, repair, replacement, or another appropriate measure, unless FHR can demonstrate the sensor was continuously experiencing positive detections during this time. 
   
   Using the method described above, each sensor's minimum detection floor values using different time segments (2 min, 5 min, and 10 min) were calculated and are shown in the attached Excel file. Based on the number of sensors and the duration of the pilot tests, 8736 daily checks of the minimum detection floor values in past 72 hour have been conducted. Out of those 8736 checks, the results for the 10-minute averages indicate that 143 checks would not pass the 10ppbe threshold because of a lack of data for the detection floor calculation. That number is 141 and 143 for 5-minute and 2-minute averages respectively. This lack of data is most often due to a sensor being inoperative (e.g. not transmitting data) at the specific time where data is needed for the calculation. Note that those periods will already be included in the required sensor downtime calculations and records. No significant difference was found between the 72-hour look back data when calculating  the detection floor using either the rolling 2 minute data, 5 minute data, or the 10 minute data. Based on this analysis, FHR and Molex propose maintaining the 10 minute data segment specification from the proposed AMEL to allow more data to be included in the detection floor calculation. A larger time segment is not recommended because of potential errors introduced in the calculation from the baseline drift. 
   
   Alternatively, we compared the daily look back at 72-hours of data to a daily look-back of 24-hours of data using the rolling 10 minute data. In this comparison, the number of detection floor check failures increased from 143 to 277, of which the number of failures due to continuous gas exposures increased dramatically from 1 in the 72-hour lookback to 28 in the 24-hour lookback. This data suggests that continuous gas exposure over 24 hours is not uncommon, and that reviewing the detection floor values each day at 0:00UTC and having at least one detection floor value under the 10ppbe for the past 72 hours instead of the past 24 hours is more reasonable and representative.  
   
   FHR therefore agrees with EPA's previous proposal to use 10-minute averages in Sections IV. A (a)(2) and IV. A (e)(3). Proposed verbiage:
    
Section IV. A (a)(2): 
            The sensor must be capable of maintaining a detection floor of less than 10 ppbe on a rolling 10-minute average, when adjusted for the system response to the most recent successful bump test conducted in accordance with IV.A(e)(2). The detection floor is determined at three times the standard deviation of the previous 10 minutes of data excluding excursions related to emissions peaks. 
            

Section IV. A (e)(2)(ii): 
            (ii) The bump test is successful if the response of the sensor exceeds 50 percent of the nominal value of the standard and the adjusted detection floor does not exceed 10 ppbe. The bump test may be repeated up to two additional times if the first bump test is unsuccessful.

Section IV. A (e)(2)(iv): 
            (iv) FHR must maintain records of the bump test in accordance with IV.C(f) and records of the detection floor must be maintained in accordance with IV.C(g).

Section IV. A (e)(3): 
            IV.A.(e)(3): The health of each sensor must be confirmed for power and data transmission at least once every 15 minutes. Data transmission, which includes data recorded by the sensor every second as noted in IV.A(a)(3), must occur at least once every 15 minutes. Appropriate corrective actions must be taken for any sensors that fail to collect data in accordance with IV.A(a)(2) and (3) and transmit data in accordance with this paragraph to ensure any errors or malfunctions are corrected in a timely manner. Such periods are considered downtime until corrected. If a sensor reset or repair is necessary, FHR must test the responsivity and wireless communication of the sensor through a bump test according to the procedure specified in IV.A(e)(2). FHR must maintain records of sensor health in accordance with IV.C(f). 
            The sensor detection floor shall be reviewed at 00:00 UTC each day to confirm each sensor detection floor remains below established threshold of 10 ppbe on a rolling 10-minute average during at least one 10-minute period in the past 72 hour period. If a sensor does not pass the detection floor review, then a sensor fault notification shall be issued, and the sensor issue shall be corrected through recalibration, repair, replacement, or another appropriate measure, unless FHR can demonstrate the sensor was continuously experiencing positive detections during this time.
Recordkeeping Section IV. C (g): 
            (g) Raw sensor readings. Additionally, for each sensor, the percent of time positive detections were registered during the 72-hour lookback must be recorded each day and the minimum, average, and maximum together with the minimum detection floor.
            
 Threshold for Entire LDSN to be Non-Compliant with AMEL  -  EPA's published proposed AMEL indicated if one or more leaks >18,000 ppm were detected during the Annual Compliance Verification the entire LDSN system was out of compliance. FHR previously proposed that while any identified gaps would be reported and corrected, only if the number of leaks >18,000 was 10% or more of the valves monitored during the Annual Compliance Verification would the entire system be considered non-compliant. EPA communicated their concern with that threshold. Therefore, FHR is proposing a reasonable compromise at 3% of the valves monitored during the Annual Compliance Verification as an indication that the entire LDSN system in that process unit be considered to be out-of-compliance with the AMEL.  The facility should be required to conduct M21 monitoring consistent with the underlying LDAR regulations on the associated AMEL covered components until the LDSN system is redesigned, approved, implemented and tested, per the current proposal. Reverting to conducting M21 monitoring consistent with the underlying regulations and meeting other compliance requirements of such regulations would resolve any non-compliance with the LDAR regulations.  In this type of instance, if reverted to conducting M21 monitoring, other inspection and work practice requirements in the AMEL (e.g., "Non-LDAR components") would not be required during the period considered to be out-of-compliance with the AMEL.
   

   
 Non-LDAR Components  -  During the call on 5/13/22, EPA expressed concerns about not expanding the scope of the AMEL beyond LDAR regulations to include non-LDAR leaks that had been discovered via LDSN/DRF because the non-LDAR leaks would be "hanging around" and negatively impacting the ability of the system to detect other leaks. FHR and Molex consider such an expansion of the scope of the AMEL beyond the scope of any LDAR regulations as overreach and unnecessary but does believe that EPA understands the basis for that viewpoint. However, in the interest of all parties to move the approval process of this important alternative forward for this AMEL request, FHR would agree to a 30-day repair requirement for non-LDAR leaks if appropriate caveats were specified. Those include:
 an inspection and work practice repair requirement to these non-LDAR leaks, and
 a Delay Of Repair (DOR) provision for non-LDAR component repairs requiring a process unit shutdown, and
 if repair requirements for a non-LDAR component are already applicable in another regulation (e.g. BWON, MACT CC cooling water systems, etc.), the AMEL repair requirements will not supersede those requirements, and
 it does not include leak sources in an area outside the AMEL covered area
