From:	 	Lester Wyborny
To:		Docket EPA-HQ-OAR-2011-0135
Date:		April 11, 2013
Subject:	Refinery-by-Refinery Gasoline Sulfur Cost Model - Response to Peer Review Comments
   1. Background 
	EPA developed a cost model which estimates the cost of reducing gasoline sulfur levels on a refinery-by-refinery basis for the Tier 3 notice of proposed rulemaking (NPRM).  To enhance the cost estimation ability of the cost model, EPA utilized detailed, refinery-specific information in the cost model, much of which was confidential business.  The confidential business information included gasoline volume and quality information that refiners report to EPA as part of the compliance requirements.  To allow EPA to better estimate the sulfur levels of each refinery's gasoline, EPA obtained crude sulfur level data for each refinery from the Energy Information Administration (EIA).  EPA also obtained actual refinery unit volumetric throughput information from EIA for the fluidized catalytic cracker (FCC unit), atmospheric crude tower, hydrocrackers, and coker units.  All of this data is considered confidential business information by EIA.  Finally, EPA obtained detailed FCC naphtha desulfurization information from several vendors.  That information was also confidential business information.
	 We contracted RTI to facilitate the independent peer reviews of the refinery-by-refinery cost model.  RTI subcontracted Charles Lieder as one peer reviewer (Reviewer 1).  Charles Lieder was the Technology Task Group Refining Committee Chair for the National Petroleum Council (NPC) when the NPC was assessing the cost of clean fuels, including low sulfur gasoline, prior to Tier 2.  Complying with Tier 3 requires revamping the FCC postreaters installed for Tier 2, or installing scaled down grassroots postreaters like those installed for Tier 2, which is why Charles Lieder's experience with Tier 2, particularly the Tier 2 postreating technologies, is valuable for this project.  The second and third peer reviewers were Martin Tallet and Dan Dunbar of Ensys (Reviewers 2 and 3).  Ensys is a refining industry contractor and uses its World Model, an LP refinery model, to conduct energy and cost analyses for the refining industry, as well as the government.    The refinery-by-refinery cost model that we are using for cost estimation is broadly similar to an LP refinery model like the World Model.  Thus, Martin Tallet's and Dan Dunbar's experience running their World Model provides them good background experience for reviewing the refinery-by-refinery cost model.    
      Rather than deal with all the challenges and complexity associated with providing the contractors with access to all the CBI data discussed above, we discovered that we could remove the EPA fuel quality data from the cost model and still allow a peer review of the model.  The refinery-by-refinery cost model then defaulted to estimating its own gasoline volume for each refinery based on the capacities of the refinery units and production volumes of straight run naphtha as estimated from the volume of crude oil that was refined.  The peer reviews were conducted in early November and early December of 2011.  Draft reports were provided to EPA in December, and the completed peer review reports were provided to EPA in early January of 2012.
	The purpose of this document is to respond to the comments provided by the peer reviewers.  We do not believe the peer reviewers identified any problems with the draft refinery model that warranted changes prior to being used for the NPRM.  However, the peer reviewers did provide valuable and meaningful review and comment, and we intend to take their comments into consideration in conducting the refinery modeling used to support the final rule.  Other than committing to fixing minor calculation errors the peer reviewers discovered in the cost model, this document is not intended to summarize any specific changes that the refinery cost model will undergo for the final rule.  We will assess whether we need to make changes to the cost model once we have received the public comments that are provided on the draft cost analysis.  
   2. Peer Review Comments and EPA Response

  A.  Calculation Errors
      Comments
	As indicated in the comments primarily by Reviewer 1, but also by Reviewers 2 and 3, the refinery-by-refinery cost model contained a small number of errors which, when corrected, has a negligible impact on the overall gasoline desulfurization costs estimated by the model.   The errors identified include:
   * The formula which estimates steam costs solely picked up the variable costs, omitting the capital cost portion of the cost.  Also, the capital cost portion of the equation applied a capital cost amortization factor which was too low (0.04 instead of 0.11).   The peer reviewer estimated that these errors resulted in underestimating the desulfurization cost by 0.02 cents per gallon.
   * For a subset group of refineries, the refinery model was not using the maximum FCC naphtha throughput volume as estimated using actual 2009 FCC unit throughput volumes, which slightly underestimates the capital costs for the postreater revamps for those refineries.  
   * There was an error in the desulfurization technology that was applied to one refinery for both for the 10 and 5 ppm cost estimate.
   * For PADD 4 refineries the model was referring to PADD 5 utility costs instead of those for PADD 4, slightly overestimating the desulfurization costs.
   * There was an error in the application of light straight run (LSR) treating costs for 1 refinery.
      Response
These calculation errors will be corrected for the final rule cost analysis.
  B. Capital cost estimates
      Comments
	Both reviewers commented on the capital cost estimate information provided by the vendor companies.  Reviewer 1's comments were two-fold.  First, the commenter suggested obtaining more information from the vendors that would allow us to better assess whether the capital cost estimates provided by the vendors are realistic.  Second, the commenter suggested hiring a hydrotreating capital cost expert to review the cost information to determine if the capital cost information is realistic, and if not what adjustments should be made.  
      Reviewers 2 and 3 also had reservations about the capital costs that we used.  One comment that Reviewers 2 and 3 made was that often vendor cost estimates tend to be biased low and that they should be increased to account for this potential bias.  To account for this potential bias, Reviewers 2 and 3 suggested using a 20% contingency factor.  Reviewers 2 and 3 also commented that the capital cost information provided by one of the vendors was internally inconsistent.  This vendor had supplied different capital and operating cost estimates for revamping an existing FCC naphtha postreater based on whether the FCC naphtha exiting the FCC unit contains 2400, 800 and 200 ppm sulfur before being desulfurized in a Tier 2 gasoline desulfurization unit.  Reviewers 2 and 3 noted that a revamp to an FCC postreater treating FCC naphtha with 2000 ppm sulfur was highest, but that the revamp to a FCC postreater treating FCC naphtha with 200 ppm sulfur was next highest in revamp costs, not an FCC naphtha postreater treating 800 ppm  -  arguing that the revamped capital costs did not seem consistent across the different revamp scenarios.  Reviewers 2 and 3 also observed that the ratio of revamp capital costs to grassroots capital costs were not consistent with the ratio reported by Handework.  In his book [Petroleum Refining Technology and Economics, 4[th] edition, Handewerk], Handework estimates that revamp costs for a unit are 30 percent of the capital cost for a grassroots unit.  Reviewers 2 and 3 also noted that we used the vendor-provided capital cost for a grassroots hydrotreater for revamping an FCC postreater to desulfurize FCC naphtha from 75 ppm to 10 ppm , for processes to desulfurize FCC naphtha from 75 ppm to 25 ppm.   
      Response
	In an attempt to verify the level of capital cost provided by the vendors, we performed two cross checks of the capital costs.  For the grassroots FCC postreater, we compared our FCC postreater grassroots capital cost estimate to two other capital cost estimates to serve as a check on our costs.  One crosscheck was the grassroots capital cost for Tier 3 used by Mathpro for an FCC postreater to reduce their gasoline's sulfur content to 10 ppm from their current Tier 2 sulfur levels.  Mathpro estimated a capital cost of $1830/bbl/day, but Mathpro's cost estimate includes outside battery limits (OSBL) costs.  Since the vendor 1 cost estimate of $1500/bbl/day only includes inside battery limits costs (ISBL), to compare our costs with Mathpro's we needed to apply our ISBL to OSBL cost factor which is 1.25.  When we apply our OSBL cost factor to our grassroots FCC postreater costs, vendor 1 costs increase to $1870/bbl/day, which is slightly higher than Mathpro's capital cost estimate, but because the values are so close, they are essentially consistent with each other.
	For the second crosscheck, we compared our grassroots capital costs to the capital cost estimated by Jacobs Engineering, which we have access to as part of the Haverly LP refinery model.  The Jacobs Engineering data base is provided with the LP refinery cost model to enable the refining costs of using existing and new refinery units to be estimated.  What is important about this is that Jacobs Engineering engineered Tier 2 FCC postreaters for refiners and therefore has first hand access to capital cost information that refiners faced when complying with the 30 ppm Tier 2 gasoline sulfur standard.  Jacobs' ISBL cost for an FCC postreater is $2200/bbl/day.  Jacob's capital cost is higher than the $1500/bbl/day ISBL cost that we are using based on the vendor data, however, the desulfurization scenarios are quite different.  While the grassroots hydrotreater designed for Tier 3 is for desulfurizing FCC naphtha from 100 ppm to 25 or 10 ppm, a FCC postreater designed for Tier 2 was typically designed to desulfurize FCC naphtha from an average of about 800 ppm down to 75 ppm.  Thus, the Tier 2 FCC postreater is required to remove about ten times more sulfur than the Tier 3 treater.   Higher FCC naphtha sulfur levels tends to cause increased recombination reactions which limits the ability of a simple hydrotreater to achieve the necessary desulfurization duty with reasonable levels of octane loss and hydrogen consumption.  To avoid high octane loss and hydrogen consumption, the FCC postreater vendors designed their postreaters with additional subunits to limit the recombination reactions.  The additional subunits varied depending on the vendor.  For Scanfining, a Zeromer or Exomer unit could be added.  For Selectfining, a second stage could be added.  For Axens, a splitter and Selective Hydrogenation Unit (SHU) could be added.  We asked the vendor which supplied the capital cost information that we used for a Tier 3 FCC postreater what the capital costs would be if the unit were designed to desulfurize FCC naphtha with a significantly higher sulfur level (i.e., 800 ppm) for Tier 2 rather than Tier 3.  The vendor responded that one or two sub units would likely be added as well as using a larger reactor.  The vendor estimated the average capital cost for these additions or changes would be about $500/bbl/day, thus increasing the vendor cost for a grassroots FCC postreater designed to treat an 800 ppm FCC naphtha feed to about $2000/bbl/day, within 10 percent of the Jacobs capital cost estimate.  We will further evaluate the capital costs for a grassroots treater to determine whether we need to make any adjustments to our grassroots FCC postreater capital cost estimate for our final rule analysis.
	We were not able to make a similar capital cost comparison for the revamped FCC postreater capital cost.  However, instead of solely using rule-of-thumb type estimates for revamp costs, we obtained specific cost estimates from the vendors, and thus the capital changes are technology-specific and are dependent on the types of investments made for Tier 2.  One advantage of obtaining cost information directly from the vendor companies is that they have information about how the FCC postreaters were configured when they were installed for Tier 2 and how those units are currently operated.  In addition, they know their technology and would know what it would take to revamp those units for complying with Tier 3.  
Reviewers 2 and 3's comments that vendor 1's capital cost for revamping an existing FCC postreater were inconsistent with the unit's severity appear to stem from a lack of understanding of the implications of the Tier 2 baseline and a focus solely on capital costs instead of looking at overall costs.  Based on our request for information, Vendor 1 provided cost information for the scenario where the FCC naphtha exiting the FCC unit is 2000 ppm, 800 ppm or 200 ppm and the existing Tier 2 postreaters must desulfurize the sulfur level of FCC naphtha down to 75 ppm.  Based on the way that refiners in each of these situations invested in Tier 2, the vendor provided one or more ways that refiners in each situation could comply with Tier 3.  Vendor 1 reported that in cases when refiners have only 200 ppm sulfur in their FCC naphtha they had used minimal capital investments to comply with Tier 2, so they would need to make comparatively high investments for Tier 3 as they would need to install a new distillation column in addition to adding catalyst to the reactor for Tier 3 to avoid large octane losses, high hydrogen consumption and greatly shortened catalyst life.  On the other hand, refiners with FCC naphtha sulfur levels at 800 ppm invested much more capital for Tier 2 and therefore their situation for complying with Tier 3 would entail very modest capital cost investments.  We discussed this with Vendor 1 numerous times and adopted it for our analysis.  The result is that capital costs were a function of Tier 2 investments, and not simply a function of FCC naphtha sulfur levels.  Through this process we also discovered that one needs to look at total costs not just capital costs by itself, since when capital costs are low, operating costs tend to be higher, and vice-versa.  Confirming this, however, is very difficult because we don't have access to refinery-specific capital investment information for Tier 2; nor do we have access to actual unit operating information that the vendor companies have access to.  We can, however, do what Reviewers 2 and 3 suggested which is to ask the vendor to review the information provided to us and verify that it correctly represents what the vendor intended.  We intend to do that for the final rule analysis as well as incorporate any other information available to us that might better inform the capital costs.  
      Comments
	Reviewers 1, 2 and 3 suggested requesting additional information from the various vendors.  In some cases, the vendor provided information for one scenario, but not another (such as a grassroots FCC postreater desulfurizing FCC naphtha down to 10 ppm, but not to 25 ppm, thus requiring EPA to estimate the information for the other scenario using other available data).  Two vendors did not provide any information for revamping their existing Tier 2 FCC postreaters to comply with Tier 3.  
      Response
We intend to make additional requests of the vendors so that we may be able to obtain that information and use it for the final rule analysis.  
  C.  Catalyst Costs
      Comments
	Reviewers 2 and 3 noticed that Vendor 1 did not provide a catalyst cost estimate for the grassroots FCC postreater.  Reviewers 2 and 3 also observed that for a revamp scenario that Vendor 1 provided information for, that the vendor had estimated a catalyst cost of 25 to 35 percent of the ISBL capital cost for that unit.  
      Response
      Although Vendor 1 did not provide a means to estimate catalyst cost for a grassroots FCC postreater, we in fact estimated its cost and included that cost in our Tier 3 cost estimate.  One input for estimating catalyst costs was to look back at the catalyst costs estimated for Tier 2 and adjusting those costs to current year costs.  We also looked at the data provided by other vendors for their catalyst costs.  Since we received the peer review comments, we requested Vendor 1 to provide catalyst cost information for a grassroots unit for Tier 3.  Vendor 1 estimated that catalyst cost would be about 2  -  4 percent of ISBL cost.  The catalyst cost that we estimated was 4 percent of ISBL capital cost, so our catalyst cost is at the upper end of the range provided by Vendor 1.  We also asked Vendor 1 about the catalyst cost information provided for that specific revamp case which was provided as a range of 25 to 35 percent (this was for the addition of a second stage for the scenario where FCC naphtha is 2400 ppm exiting the FCC unit).  Vendor 1 stated that those percentages represent the increased catalyst volume over current catalyst volume in the Tier 2 reactor, not catalyst cost estimate as a percent of ISBL.  Our catalyst cost estimate is consistent with the way that Vendor 1 intended it.  We will further review our catalyst cost estimate for the final rule cost analysis and make adjustments as necessary.
  D.  Natural Gas Prices
      Comments
	Reviewers 2 and 3 compared the natural gas prices that we used in the refinery-by-refinery cost model with their assessment of natural gas prices.   The following table contains the comparison of natural gas prices that Reviewers 2 and 3 provided in their report (the natural gas prices that we used are found in the column labeled "EPA 2017," and the those estimated by Reviewers 2 and 3 are found in the column labeled "Revised 2017").  
               Natural Gas Prices Comparison ($2009/million Btu)

EPA Base 2009
EIA 2010/2011
EPA 2017
Revised 2017
PADD1
4.87
9.96
7.88
10.30
PADD2
4.86
7.44
7.86
7.87
PADD3
3.49
5.21
5.64
5.64
PADD4
4.36
6.34
7.06
6.77
PADD5 ex CA
4.60
8.33
7.44
8.76
PADD5 CA
n.a.
7.04
7.44
7.49






      Response
	Based on the increased natural gas prices in PADDs 1 and 5 outside of California, Reviewers 2 and 3 estimated that our estimated costs for sulfur control would increase by 0.02 cents per gallon if we were to use their revised natural gas prices.  However, Reviewers 2 and 3 also suggested that we update our natural gas prices to those estimated in AEO 2011.  Reviewers 2 and 3 are correct in that we inadvertently based our natural gas prices on the 2010 AEO.  The 2011 AEO estimates that natural gas prices in 2017 will be about 25% lower than those in the AEO 2010 for the year 2017.  However, for the final rule analysis we plan on using natural gas prices as estimated in the most recent AEO. The early release version of the 2013 AEO projects natural gas prices will be slightly higher in 2017 than those projected in the 2010 AEO, so if this projection continues in the final version of the 2013 AEO, the updated natural gas prices will tend to very slightly increase our estimated desulfurization costs.   
  E.  Hydrogen Demand and Costs
      Comment
	Because the hydrogen demand for some of the desulfurization scenarios provided by the vendors seemed low, Reviewers 2 and 3 expressed some concern that the hydrogen consumption values provided by the vendors reflect stoichiometric values, not actual hydrogen consumption.  Reviewers 2 and 3 stated that if the values reflect stoichiometric hydrogen consumption, then the hydrogen consumption values should be adjusted to reflect actual hydrogen consumption.   The comments by Reviewers 2 and 3 were made in reference to information provided by Vendor 1.
	Reviewers 2 and 3 reviewed our hydrogen cost estimates and concluded that our estimated hydrogen costs are in line with other estimates, and are appropriate given the natural gas prices that we are using.  
      Response
      We contacted Vendor 1 and Vendor 3 since we learned of these comments by Reviewers 2 and 3, and   both vendors confirmed that the hydrogen consumption values provided to us represent actual hydrogen consumption including any losses, not stoichiometric hydrogen consumption only.  For the revamp cases in particular, this makes sense.  
      We took a step back to review hydrogen consumption by hydrotreaters.  In addition to the hydrogen consumed for desulfurization and the olefin saturation side reactions, there are also losses.  One loss of hydrogen during hydrotreating is solution losses.  Solution losses are the losses of hydrogen because the hydrogen is dissolved in the FCC naphtha  -  the FCC naphtha enters the hydrotreater with no hydrogen dissolved in the FCC naphtha, but it exits the hydrotreater with hydrogen dissolved in the FCC naphtha.  However, for Tier 3, most refiners will be revamping an existing hydrotreater and the FCC naphtha will already have been treated in the Tier 2 hydrotreater and therefore already contain hydrogen in solution.  Thus, there should be little or no increased hydrogen loss in solution for the revamped hydrotreaters.  
      Another category of hydrogen losses during hydrotreating is purge losses.  Purge losses are required to prevent the accumulation of hydrogen sulfide gas and other gasses in the recycle hydrogen.  However, the very small additional sulfur reduction required for Tier 3 (reducing FCC naphtha from an average of 75 ppm down to 25 ppm) is not likely to require any appreciable increase in the purging of recycle gas.  
      The last category of hydrogen losses is mechanical losses.  In this case, the hydrogen under pressure leaks out of joints in the pipes and reactors which contain the hydrogen.  Because gasoline desulfurization occurs at modest pressures, mechanical losses of hydrogen would be modest in this situation.   Furthermore, in the case of revamps, there is little to no additional chance for hydrogen to leak because in many cases the Tier 3 revamp relies on the same hydrogen piping and reactor arrangement of the Tier 2 postreater.   
	One reason why Reviewers 2 and 3 suggested that the hydrogen consumption values might represent stoichiometric hydrogen demand, not total hydrogen demand, is because of one of Vendor 1's hydrogen consumption estimates which seemed low - only 6 standard cubic feet of hydrogen per barrel of feed (SCF/bbl).  A second hydrogen consumption estimate, of 19 SCF/bbl was also lower than the other revamp scenarios, which ranged from 30 to 100 SCF/bbl.  However, in the context of the revamps which were studied by Vendor 1, the very low hydrogen consumption values make sense.  In both those cases, the 6 SCF/bbl and 19 SCF/bbl, Vendor 1 indicated that refiners relied on limited FCC postreater investments for Tier 2, choosing to have higher hydrogen consumption to minimize the capital investments needed.  However, when faced with a more stringent gasoline sulfur standard, refiners would likely choose to make a modest capital investment in lieu of being faced with even larger hydrogen consumption demands.  In the case of 6 SCF/bbl, the feed to the Tier 2 FCC postreater would only contain about 200 ppm sulfur.  As discussed above, most refiners in this situation apparently avoided the capital cost of installing a splitter to separate out the olefin-rich light FCC naphtha from the heavy FCC naphtha, choosing instead to feed the entire FCC naphtha stream to the postreater.  This resulted in somewhat higher hydrogen consumption for their Tier 2 treater.  However, instead of avoiding further capital investments for Tier 3 resulting in even higher hydrogen consumption levels (and octane losses and much shorter catalyst life), refiners would likely install a splitter to separate out the light FCC naphtha and not run that volume through the hydrotreating reactor.  If the refiner were to add the splitter to its Tier 2 FCC postreater but still comply with 30 ppm Tier 2 standard, the refiner would likely reduce the hydrogen consumption for complying with Tier 2 (i.e.,  20 SCF/bbl).  Then by increasing the severity to the Tier 2 treater with splitter to then comply with Tier 3, the FCC postreater might demand an additional 26 SCF/bbl, however, the net hydrogen increase over their Tier 2 FCC postreater without the splitter would only be 6 SCF/bbl.  A similar story applies for the 19 SCF/bbl revamp case, except that in this case the capital investment that could be made to result in incrementally low hydrogen consumption is a second stage.  Additional discussion about the investments being made for Tier 3 is provided in Chapter 4 of the draft RIA.     
  F.  Electricity Prices
      Comment
      Reviewers 2 and 3 commented that EPA had used electricity prices which are low.  Reviewers 2 and 3 stated that EPA's electricity prices fell in the range of 4.7 to 8.4 c/kw-hr. However the AEO reports an average electricity price of 6.8 c/kw-hr for 2009.  
      Response
      Relative to 2009's electricity costs for industrial users, our electricity prices may be low.  However, in AEO 2011, EIA estimated that average electricity prices in 2017 for industrial users will average 6.0 c/kw-hr, which is in the middle of our range.  For our final rule analysis, we will use the EIA's most recent AEO for estimating electricity prices to refiners.  The early release of the 2013 AEO shows a higher average electricity price for 2017 of 7.0 cents per kilowatt-hour, and if the latest AEO maintains this electricity price, it will slightly increase our desulfurization cost estimates.  As Reviewers 2 and 3 acknowledged though, small differences in electricity prices would have a negligible impact on the estimated desulfurization costs.  
  G.  Octane Costs
      Comment
      Reviewers 2 and 3 provided some context for octane costs. Reviewers 2 and 3 note that the regular grade-premium grade price differential between 2008 and 2010 averaged $5.29/gal, or $0.90/bbl per octane number.  Furthermore, as Reviewers 2 and 3 point out, the historical methods for estimating the cost of making up lost octane, such as the regular-premium price differential, do not account for increased blending of ethanol into the future.  Reviewer 2 and 3 stated that its future year octane costs based on its World model are similar to ours.  They estimated costs for octane which ranged from 0.25 to 0.45 $/octane-bbl for PADD 3 and PADD 1, respectively (which is 0.59 to 1.07 cents/octane-gallon).  
       Response
      We agree with the comments by Reviewers 2 and 3 about the disadvantage of using historical regular grade-premium grade price differential method for estimating octane cost because it doesn't consider future use of ethanol, and also we believe is that it likely includes some of the profit earned by the much higher premium gasoline prices.  Also, part of the price premium for premium gasoline could be due to the need to segregate and distribute a much smaller volume of product.  In our case, we are solely interested in the cost of making up losses in gasoline octane within the refinery, not the octane cost difference of high octane blends at retail.  For this case, LP refinery models are better suited for estimating octane costs because they solely reflect the cost for recovering the lost octane.  
	Reviewer 2 and 3 stated that future year octane costs based on its World model are similar to ours.  They estimated costs for octane which ranged from 0.25 to 0.45 $/octane-bbl for PADD 3 and PADD 1, respectively (which is 0.59 to 1.07 cents/octane-gallon).  Our estimated octane cost developed using the Haverly LP refinery cost model is 0.32 $/octane-bbl (0.76 cents/octane-gallon) is right in the range of costs estimated by Reviewers 2 and 3.  Reviewer 2 and 3's refinery model runs were run in 2015 while our refinery modeling runs were made in 2017.   Because Reviewers 2 and 3 refer to E10 runs, their runs were likely conducted with about 100% E10 in the gasoline pool.  Our runs assuming 2017 renewable fuels volumes resulted in about half E10 and half E15 in the gasoline pool.  Reviewers 2 and 3 also found that the more ethanol that is being blended into the gasoline pool, the lower the marginal cost for making up for lost octane, and our analysis agrees with that statement.  We chose, however, to assess the cost of recovering lost octane on the first year of the proposed Tier 3 gasoline desulfurization program, which is 2017, instead of later years when projected octane costs would likely be lower, probably making our octane cost projections, based on the first year of the Tier 3 program, somewhat conservative.  Reviewers 2 and 3 commented that our octane costs were not lowest in PADD 3 (Gulf Coast), which typically is the lowest refining cost PADD of refineries.  Because of this concern about the relative projected octane recovery costs in the different PADDs, we will review and possibly rerun our octane cost cases for the final rulemaking (FRM).
  H.        Light Straight Run (LSR) Treating
      Comment
	Reviewers 2 and 3 commented that the refinery-by-refinery cost model did not appear to apply an overdesign factor to LSR treaters.  
      Response
      Initially a dialog box was created that contained a 1.15 overdesign factor to apply for LSR treating.  However, the overdesign factor was subsequently integrated directly into the cost equation for LSR treating, hence the 1.15 factor contained in the dialog box was never utilized, although an overdesign factor was included.
  I.  Other Cost Inputs
      Comment
	Reviewers 2 and 3 commented that the cost model did not appear to consider cost factors outside of FCC naphtha and LSR desulfurization units, such as acid gas and sulfur recovery plants.  
      Response
      The cost for operating sulfur recovery plants was estimated and the very low cost impact was included in the refinery cost model.
  J.  Future Refining Industry Changes Impacting Desulfurization Costs
      Comments
	Reviewers 2 and 3 commented that over time changes in the US refining industry should be accounted for because of their likely impact on gasoline desulfurization costs.  These include 
   * Accounting for major projects at US refineries particularly those which will allow refineries to process volumes of heavy Canadian crudes (tar sands), including the potential impact of the Keystone pipeline if it were built.
   * Potential shifts over time in FCC feedstocks and operating modes, notably: shift towards more residual fuel in FCC feed as VGO is increasingly "pulled away" to hydro-cracking units as demand for distillates grows, FCC catalyst/operating shifts to lower gasoline higher distillate (LCO) and also higher propane yields; potential resulting impacts on FCC gasoline qualities, volumes and thus concentration in the U.S. gasoline pool
   * Potentially substantial changes in U.S. domestic oil liquids supply, both light sweet crudes from the Bakken, Eagle Ford, Niobrara, Utica shale and other emerging domestic crude oil supplies and the potential impacts of rising NGL's supply, and
   * Extent to which current trends for growth in U.S. refined products exports will continue and their potential impacts on U.S. refining.
      Response 
      Reviewers 2 and 3 raised some important points about how future US refining industry changes could impact the Tier 3 modeling effort.  For the NPRM cost analysis, we initially identified that three US refineries were to undergo major crude oil throughput expansions.  To obtain information about the expansions, we reviewed the unit capacity information from EIA for 2011 (our base year was 2009 so we originally relied on 2009 refinery capacity data from EIA and the Oil and Gas Journal) and used the later data which included the increased crude and other unit throughput increases due to the expansions.   Reviewers 2 and 3 provided information about the refinery throughput expansion or projects that would allow refineries to process Canadian tar sands or other heavy crude oil.  We intend to incorporate the refinery expansion/modification information provided by Reviewers 2 and 3 and any we can find from other sources in our final rule cost analysis to the extent possible within the constraints of a refinery-by-refinery model.
      K.	Changes in Crude Quality
      Comment
      Reviewers 2 and 3 also pointed out that crude quality will change over time which would impact gasoline desulfurization costs. 
      Response 
      Very sour (high sulfur) Canadian bitumen or tar sands, as well as sweet (low sulfur) crude oils from shale are increasingly being processed by US refineries.  Because of the very different nature of Canadian tar sands, refineries planning on processing it typically undergo significant capital modifications and we therefore know fairly confidently which refineries will be processing tar sands in the future, which is helpful when running a refinery-by-refinery cost model.  However, predicting how refineries will process the tar sands provides an additional challenge.  We will attempt to figure out how to integrate the increased use of tar sands in our cost analysis.  The development of shale resources can yield both shale oil and light hydrocarbons such as natural gas liquids.  Shale oil is typically refined, and natural gas liquids are typically blended by the refineries which are located nearby the location where these hydrocarbon sources are being developed.  We will assess how to include these expanding sources of hydrocarbons in our cost analysis for the final rule within the constraints of a refinery-by-refinery model.
      L.	Increased Dieselization 
      Comments
	Reviewers 2 and 3 suggested that we account for the change in refinery operations caused by the shifting product market which increasingly favors diesel fuel over gasoline.  Revieweres 2 and 3 pointed out that to respond to these changing product demands, refiners could shift vacuum gas oil, the typical feedstock for FCC units, to hydrocrackers resulting in heavier, more sour feedstocks for FCC units which could increase desulfurization costs.
      Response
      Since Tier 3 will take effect in the future, our cost analysis should take into account, as much as possible, future trends in refining.  As diesel demand increases while gasoline demand lags, refiners are expected to add hydrocracking capacity which favors diesel fuel production.  While refiners may make the feedstock changes to FCC units that Reviewers 2 and 3 suggest, this change to the FCC would be offset by the increased production of low sulfur hydrocrackate which would displace some of the volume of FCC naphtha blended into the gasoline pool, thus tending to lower the cost of desulfurization.  We will consider ways to incorporate these changes for the final rule.  However, it is challenging to make such adjustments in a refinery-specific model since it is difficult to predict the refineries that will add hydrocracker capacity.  Another likely and important change of dieselization is the undercutting of FCC naphtha into the distillate pool.  If refiners undercut their FCC naphtha into distillate, it would reduce the volume of FCC naphtha needing to be hydrotreated thus decrease the volumetric demand on the existing Tier 2 postreater.  However, more importantly, undercutting FCC naphtha can reduce by half the sulfur exiting the FCC unit.  Overall, these changes would reduce the sulfur and volume content of the feed to the existing postreaters significantly easing compliance with Tier 3 even if refiners reserve the dirtiest of feeds to the FCC unit.   
      M.	Refinery-by-Refinery Model Documentation and Organization
      Comments
	Reviewers 1, 2 and 3 commented that the refinery-by-refinery cost model lacked documentation, needed simplified formula structure and was poorly laid out which made navigating the refinery cost model difficult.  Despite the difficulty navigating the refinery cost model, both reviewers indicated that they had adequate time to review the refinery-by-refinery cost model.  
      Response
Our plan is to heed the Reviewers' advice by organizing and better documenting the refinery-by-refinery cost model. 
