





Date:
January 12, 2012
Subject:
Peer Review of a Refining Industry Cost Model
From:
Sandra Burns, RTI International
To:
Lester Wyborny, US EPA


1.0 Background
An independent, third-party peer review was conducted of a refinery-by-refinery cost model developed by EPA.  The Cost Model Spreadsheet developed by EPA models the cost  for each refinery to  reduce their gasoline sulfur levels. This memo is a summary of the observations and recommendations of Mr. Charles Lieder. The review was facilitated by RTI International.
2.0 Initial Observations
Mr. Lieder noted that the submitted Technical Package (Cost Model Spreadsheet and supporting files) can only be read and understood if the third-party reviewer has almost unlimited access to the EPA Author(s). The Technical Package is a working draft from the desks of the EPA Author(s).  With their instruction, a reviewer can understand the concepts behind the Cost Model Spreadsheet. 
 3.0 Summary Comments
EPA requested that six areas be peer reviewed. These are numbered one through six below, and are followed by Mr. Lieder's comments. 

1.  Review and comment on the overall design of the refinery-by-refinery cost model for estimating the cost of sulfur control, including: (a) The means for modeling each refinery's operations, including their gasoline volume and sulfur level as tools for calibration, (b) the use of FCC naphtha desulfurization equipment as the most cost-effective means for reducing gasoline sulfur levels, and (c) comment on how the refinery-by-refinery cost model could be improved.

 (a) Based on the Nov 9[th] RIA provided to the reviewer and the Model's "refinery-core-calculations" (developed in MSAT / Benzene Regulation modeling), the spreadsheet modeling is appropriate and similar to approaches used in the U.S. refining industry.  EPA's access to confidential refinery-specific operations information from EIA (in the peer review model) and gasoline volumes and quality information available from its own reformulated gasoline data base (removed from the peer review model), provides the model a firm foundation for checking calculations and subsequent calibration.  
      
 (b) EPA selection of a `FCC naphtha desulfurization' strategy is appropriate and should be the cost-effective path.  Recently, a July 2011 API Study, Potential Supply and Cost Impacts of Lower Sulfur, Lower RVP Gasoline with refinery modeling (Baker & O'Brien, 2011) relied heavily on FCC naphtha desulfurization to attain a Gasoline Pool sulfur level of less than 10 ppm. The refinery modeling study done by Mathpro also relied on FCC naptha HT (Mathpro, 2011).
      
 (c) Because the EPA reformulated gasoline quality and volume data which it could use for calibrating the model are confidential, the Peer Review cannot include them or evaluate the original, corrections, refinements, or final EPA selections for calculations.  Therefore, the Peer Review must look at the current spreadsheet and it is acceptable. The main improvement area is in capital cost data for desulfurization of FCC naphtha and Light Straight Run / Natural Gas Liquids.  
 (d) The clarity, understanding and logic flow of the spreadsheet model could be enhanced by descriptive titles and comment fields for key data and calculations.    

The overall design of the refinery-by-refinery cost model is appropriate since it contains both a sound data foundation (assumed as firm on basis of EPA's statements about their source of confidential data and its use in calibration) and an accepted design / methodology for refinery spreadsheet calculations and cost estimations.

2.  Review and comment on the inputs for the various desulfurization technologies provided to EPA to ensure that these are being represented adequately to correctly estimate the desulfurization costs. 

 EPA sought inputs and cost information for the desulfurization technologies which would be predominant in a near-term Tier 3 Fuels Sulfur Program.  Unfortunately, only one source provided detailed, complete engineering information which could be the basis for confident estimation of capital costs for the necessary new desulfurization targets. Thus to move the calculations forward, EPA has currently had to make numerous assumptions and estimations to select and calculate desulfurization costs for the technologies for which they do not have technically complete engineering information. 

EPA must clearly communicate this situation to potential users of the results of the Cost Model Spreadsheet (or users of spreadsheet). Mr. Lieder recommends that EPA make another `request for information' of a detailed engineering nature (citing the type of information that was received from one desulfurization vendor). 
                                       
3.  Review and comment on the sulfur levels assigned to each gasoline blendstock.

The Nov 9[th] RIA describes the logic of Tier 2 and potential Tier 3 sulfur levels of refinery blendstocks. The document provides an appropriate display of technically-sound blendstock levels.

4.  Review and comment on the assumption that some refiners will need to desulfurize their LSR to comply with lower sulfur gasoline.

Assumption of the need to desulfurize LSR and NGL is technically sound.  Other studies note the desulfurization of `straight-run naphthas'; and details of use in their appendices (Baker & O'Brien, 2011; Mathpro, 2011). EPA's approach to premise desulfurization of all LSR / NGL with either existing,  revamped, or new units is a technically sound way to estimate the potential impact of a Tier 3 Fuel Sulfur Program.

5.   Review and comment on our means for estimating the blendstock volumes for each refinery.

The  Nov 9[th] RIA provided a  well-documented set of analyses that document how the blendstock volumes were estimated in the cost model. It is sound, transparent and supported by substantial data and references.

6.   Ensure the integrity of the refinery-by-refinery cost model by working through the equations present throughout the spreadsheet. Check the equations throughout the spreadsheet with sufficient frequency (i.e., one refinery in each PADD) to ensure that the refinery model formulas refer to the appropriate cells. Report any errors.

For production of 10 ppmS Gasoline Pool, 13 refineries (out of the ~101 refineries)  were checked  for the "Costs 10ppm" worksheet calculation  and the background calculations which covered data and calculations from almost every worksheet that was reviewed; especially T3,, Proc 1  -  Proc4, Crude Regression, Vendor Cost Info, 09 ref cap data, Data for 2009, and Utility & Capital Costs.  The methodology of the cost calculation was sound for these 13 refineries. For one unique technology implementation, a calculation error was identified and communicated to EPA
             
For production of 5 ppmS gasoline pool, 8 of the above 13 refineries were checked for the `Costs 5ppm' worksheet calculation and the background calculations which covered almost every worksheet. Again, the methodology of the cost calculation was sound for these 8 refineries (for same unique technology implementation, a calculation error was identified) .
4.0 Detailed Comments
EPA provided several cost-estimates from various sources; these were the basis for EPA's analysis of sulfur-reduction technologies and generation of capital cost estimates to be used by refineries in the main Cost Model spreadsheet. 
4.1 Cost Model Calculation Methodology 
Basically, the methodology and flow-of-calculation is technically-sound. Note that that the review will subsequently comment on the soundness of the data (such as capital costs) that are used in this methodology and the exact decision points for technology choices in sulfur reduction strategies for a refinery. One key worksheet in the model pulls together the cost model data and methodology to produce the estimated cost to comply with a 10 ppmS gasoline pool requirement on a refinery-by-refinery basis. Mr. Lieder conducted an analysis of 13 refineries which permitted an in-depth understanding of the EPA approach. The following key aspects were noted.

   * Appropriately addressing the status of and need for additional FCC naphtha desulfurization. methodology, calculations and logic flow is defensible. 

   * Addressing the comparison between maximum-estimated FCC Gasoline and  actual 2009 FCC gasoline: about a dozen refineries have larger actual 2009 than the maximum-estimated; however, the spreadsheet always used maximum-estimated in subsequent calculation.  EPA needs to decide which MAX FCC gasoline volume is used for sizing FCC post treaters.  

   * Methodology uses extensive reference data from the Utility & Capital Costs worksheet. The data and use calculations are appropriate. However, Mr. Lieder found an error in the refinery/PADD association; EPA should make certain that refineries are associated with correct PADD.                   

   * For the appropriate FCC desulfurization technology and sulfur level for a particular refinery this fuel program cost calculation relies on many data inputs. Again, data from the utility & capital costs worksheet is used extensively; some EPA assumptions are embedded in the Utility & Capital Costs worksheet (such as return on investment (ROI) of 7% and fixed operating costs as 6.7% of capital cost). More documentation as to where this data is used would be useful.
     
   * Also, for the appropriate FCC desulfurization technology and sulfur level for a particular refinery,  the crucial inputs are the capital costs and associated operating costs.  Current input data required extensive EPA analysis and calculation as well as capital selection (see extensive discussion in RIA). EPA needs to re-request detailed engineering input on capital costs (for new or revamped or expanded FCC post treaters) from the public. 
      
   * For one refinery the total cost section needs to be reviewed. EPA needs to review equations in the LSR Addition / Expansion section for total capital employed, as an error was found and there is a chance that several equations need to be revised.  This was communicated to EPA, and Mr. Lieder believes it is likely that this is the only refinery with this specific calculation error.
4.2 Vendor Cost input Information and Use by EPA   
EPA received only limited technical information from vendors. Thus, for the refinery that must meet potential Tier 3 Fuel Program requirements, EPA had to assess the level and range of sulfur reductions, and propose various technologies which can be added to current refineries and estimate the capital and operating cost of those various technologies. Because such limited data was made available, EPA had to modify the information provided, extrapolate for several cases and estimate substantial areas of cost and operating factors. Thus, EPA's cost estimates do not have broad input from engineering and construction firms. If feasible, EPA should seek additional capital cost estimates and/or input from an expert in capital cost estimation.  

Mr. Lieder reviewed the worksheet containing capital cost information. The following key aspects were noted.
      
   *       A detailed description of the logic flow and results is needed. EPA needs to critically review the results.  Because of the lack of detailed interpretation of data in their submittal, EPA had to take a circuitous route to the presented numbers. The capital costs of Vendor 1 were not well documented and therefore the reviewer was unable to comment critically on the validity of EPA's capital cost estimate.

   *       Although possibly futile, EPA should consider other routes to obtaining additional capital cost estimates such as another Request For Information; contracting with capital cost experts or a refining consultant .
4.3 Methodology for Individual Refinery Selection of Sulfur-reduction Technology for Tier 3 Sulfur Targets 
The methodology can be defended; but is open to a rational technical discussion of decision points and criteria of technology choices from others who may have slightly different technical criteria.
The methodology for individual refinery decisions on FCC Naphtha and LSR/NGL desulfurization is described in the RIA and is a useful background.  
4.4 Calculating EPA's Fuel Program Costs with Cost Model
1. Refinery Capacity methodology:  EPA has obtained data from EIA; the experts on such U.S. data.  Best source of U.S. data collected under DOE-EIA legal imperative.

2. Blendstock Production Volumes methodology: The methodology was peer- reviewed for the benzene reduction rulemaking / model spreadsheet. A quick review of the methodology shows that the approach is detailed and in-line with other technical approaches. This explanation of "Estimating Individual Refinery Gasoline Blendstock Volumes" in the Nov 9[th] RIA is more than eight pages of tables and analysis. This well-documented set of analyses is sound, transparent and supported by substantial data and references.

3. Blendstock Sulfur Levels & methodology:  The extensive RIA narrative explains the logic of Tier 2 and potential Tier 3 sulfur levels of refinery Blendstocks, and displays technically-sound blendstock levels. 

4.  Utility & Capital Costs: Well-referenced information is presented for use as factors in subsequent spreadsheet calculations.  A small amount of documentation as to the use of the data in this worksheet would be useful. 

5.0 References

Baker & Obrien, 2011. Potential Supply and Cost Impacts of Lower Sulfur, Lower RVP Gasoline. Prepared for the American Petroleum Institute. Available online at http://www.api.org/Newsroom/upload/110715_LowerSulfur_LowerRVP_Final.pdf 

Mathpro, 2011. National Association of Clean Air Agencies Report's Refining Economics. Available online at http://www.theicct.org/pubs/ICCT04_Tier3_Report_Final_v4_All.pdf  
