DATE:	August 2, 2021
FROM:	Donna Lee Jones and Kevin McGinn, U.S. Environmental Protection Agency
TO:		Coke Ovens Residual Risk and Technology Review (RTR) Project File
SUBJECT:	Revised Equation to Estimate Coke Oven Emissions from Oven Doors
_________________________________________________________________________________________

INTRODUCTION

The emission standards for Coke Ovens, 40 CFR part 63, subpart L were promulgated as National Emission Standards for Hazardous Air Pollutants (NESHAP) on October 27, 1993 (58 FR 57898). This NESHAP (subpart L) set maximum achievable control technology (MACT) standards for leaks of coke oven emissions (COE) from oven doors, lids, and offtakes, and other coke oven sources. COE are a separately-listed HAP consisting of approximately 71 chemicals, including volatile and semi-volatile organic HAP and HAP emitted as particulate matter (PM). 
      There are two types of coke plants. One type produces coke and sends the exhaust COE to a chemical recovery plant that is on the facility site and recovers chemicals from the COE exhaust for sale. This type of coke plant is called a coke by-product plant. The other type of coke plant does not recover COE and is called a nonrecovery plant. Most of these nonrecovery coke plants, however, recover heat from the coke exhaust, which is used to generate steam on site and to generate electric power that is sold. Coke by-product ovens operate under positive pressure, whereas nonrecovery coke plants operate under negative pressure. For the by-product coke plants, MACT limits for subpart L sources were set based on the percent of oven doors, offtake gas ducts and oven charging lids that leak visible emissions, as measured using EPA Method 303. Because nonrecovery coke ovens operate under negative pressure, there are zero leaks from their doors, lids, and offtakes. Emissions of COE leaks from by-product coke ovens have historically been estimated using AP-42 emission factor equations[1] that include the number of leaking doors, lids, and offtakes, and the seconds of visible emissions measured with EPA Method 303. 

This memorandum discusses a new method to estimate emissions from oven doors that better reflects the technological improvements in today's coke industry.
BACKGROUND
In 2005, the EPA promulgated the residual risk and technology review (RTR) for subpart L (72 FR 19992). In the RTR for subpart L, COE from leaking oven doors, were estimated using the following equation taken from the estimating procedures in AP-42. The nonvisible emissions in the AP-42 equation were considered to be zero in the RTR, but were acknowledged to contribute to the uncertainty in the estimates, where actual emissions were stated as potentially higher than estimated with an acknowledgment that they also could be lower (70 FR 20007).
      
COE-doors (lb/hr) = ND x (PLDyard/100) x (0.04 lb/hr) + ND x (PLDbench/100) x (0.023 lb/hr)
                                       
where:
ND		=	number of doors
PLD		=	percent leaking doors
bench		=	walking platform running next to the ovens (and doors)
yard		=	50 to 100 feet from the oven doors
PLDyard 	= 	percent of doors with visible leaks visible from the yard 
PLDbench	=	percent of doors with visible leaks only visible from the bench, assumed to be 6%.  

Because of safety concerns, actual observations are not typically taken from bench and, therefore, this equation has historically included a default value of 6% for leaking doors visible only from the bench, as shown below:

COE-doors (lb/hr) = ND x (PLDyard/100) x (0.04 lb/hr) + ND x (6%/100) x (0.023 lb/hr)
 
All emissions estimates used in the initial draft 2021 RTR baseline risk assessment were based on application of the 6% default value for bench-only leaks from doors.

RE-ASSESSMENT OF DEFAULT DOOR LEAKS VISIBLE ONLY FROM BENCH

      The value of 6% leaking doors visible only from the bench originates from the July 2008 update to AP-42,[2] where data from 1981 was used as the basis for the 6% value, as described in the 2008 AP-42 background document. In this document, it is stated that the data used to develop the 6% leaking doors that are visible only from the bench were derived from measurements where the percent leaking doors from the yard was 6.4% and the total percent leaking doors visible from the bench was 12.4%, which includes both leaks visible from the yard and leaks visible only from the bench.[3] The difference between 12.4% and 6.4%, at 6%, was taken to represent the percent leaking doors visible only from the bench. The data used to support these estimates are explained in the AP-42 background document,[5] as follows:

      "In August 1981, EPA conducted bench and yard observations of door leaks at the US Steel Clairton Works (Batteries 7, 8, and 9). The program was designed to obtain accurate data and to minimize the bias created by process and observer variance. There were four observers divided into two teams. Each team traversed the same side of the battery simultaneously, one from the bench and one from the yard. At the end of the traverse, the teams switched positions and conducted another traverse. Two traverses of the same side of the battery conducted in this manner are considered a set. A total of 25 sets of observations were made, and in all cases, more leaks were observed from the bench than from the yard. The average was 6.4 PLD from the yard and 12.4 PLD from the bench, which gives an average difference of 6 PLD. The difference ranged from 0.8 PLD to 14.5 PLD."
      
The 6% leaking doors visible only from the bench were presented as an absolute value and not set in relation to the percent leaking doors from the yard, which is the only measured value related to door emissions. Support for an alternative equation can be found in the response to comments included in the July 2008 background document, where the EPA stated: "It is not clear that the difference should be a constant rather than a ratio, i.e., the difference between yard and bench leaks may change as the total number of leaks decreases."

      In the current coke industry, the percent leaking doors measured from the yard are now much lower, at 2.5% or less, based on 2016 source tests performed for an EPA information collection request. The facility that was used to establish the 6% leaking doors visible only from the bench in 1981 was U.S. Steel Clairton-PA, which had 6.4% leaking doors visible from the yard in 1981, but now has a facility average of 0.5% leaking doors visible from the yard. Therefore, we believe the default fixed value of 6% leaking doors visible only from the bench should be reevaluated so that the total emissions from doors are not over estimated and are in relation to the measured percent leaking doors from the yard and not a fixed value. 
      
      We are proposing that a bench-to-yard "adjustment ratio" be used instead of the 6% fixed value for bench-only visible door emissions. The adjustment ratio would be based on the historic values measured in 1981, as the ratio of the percent leaking doors visible from the bench[3] to the percent leaking doors visible from the yard. This adjustment ratio would be used with the measured percent leaking doors from the yard (with EPA Method 303) to estimate percent leaking doors visible only from the bench. The ratio of bench-only visible emissions to yard emissions from 1981 is ((12.4-6.4)/6.4), equal to 6.0/6.4 or 0.94. This adjustment ratio (0.94) would be multiplied by data for percent leaking doors measured from the yard to estimate the bench-only component of door emissions in the equation for COE for doors from above, and shown below with the adjustment factor:
      
COE-doors (lb/hr) = ND x (PLDyard/100) x (0.04 lb/hr) + 
                                    ND x ((PLDyard x 0.94PLDbench/PLDyard)/100) x (0.023 lb/hr)

where:

ND	=	number of doors
PLD	=	percent leaking doors
yard	=	50 to 100 feet from the oven doors
bench	=	walking platform running next to the ovens (and doors)
0.94	=	ratio of PLDbench to PLDyard

In this new equation, the percent leaking doors visible only from the bench is still represented by the second term, which is now set equal to PLDyard. x 0.94.

COE-doors (lb/hr) = ND x (PLDyard/100) x (0.04 lb/hr) + ND x ((PLDyard x 0.94)/100) x (0.023 lb/hr)


CONCLUSION

      This revised equation is expected to provide more accurate estimates of total COE from doors that reflects advancements in preventing and reducing door leaks that have occurred since 1981. The term used to estimate unmeasured leaking doors visible only from the bench in the revised equation still incorporates the historic measurements but also includes measured values at the facility. Table 1 shows the door emissions before and after this correction to the equation.
      

         Table 1. Door Emissions Before and After Equation Correction
                                   Facility
                               Coke Capacity TPY
                                Number of Ovens
                               Facility Average 
                            % Leaking Doors (yard)
                                Door Emissions 
                                   (COE) TPY
                                       
                                       
                                       
                                       
                               Before Correction
                               After Correction
ABC-Tarrant-AL
                                                                        699,967
                                      132
                                     0.82%
                                      2.2
                                     0.95
AKS-Follansbee-WV
                                                                      1,346,000
                                      224
                                     0.44%
                                      3.2
                                     0.71
AKS-Middletown-OH
                                                                        456,000
                                      76
                                     1.70%
                                      1.4
                                     0.70
AM-BurnsHarbor-IN
                                                                      1,900,000
                                      164
                                     2.30%
                                      3.3
                                      2.0
AM-Monessen-PA
                                                                        372,581
                                      56
                                     0.75%
                                     0.81
                                     0.21
AM-Warren-OH
                                                                        549,000
                                      85
                                     1.70%
                                      1.5
                                     0.80
BLU-Birmingham-AL
                                                                        596,775
                                      120
                                     1.40%
                                      2.0
                                     0.87
EES-RiverRouge-MI
                                                                      1,050,000
                                      85
                                     0.02%
                                      1.0
                                    0.0084
USS-Clairton-PA
                                                                      6,422,997
                                      708
                                     0.54%
                                      9.8
                                      2.0

