The amount of hydrogen required for desulfurization is also uncertain, because the industry has no experience with widespread desulfurization at ultra-low levels. One of the primary determinants of cost is hydrogen consumption and the related investment in hydrogen-producing equipment. Hydrogen consumption is the largest operating cost in hydrotreating diesel, and minimizing hydrogen use is a key objective in hydrotreating for sulfur removal. In general, 10 ppm sulfur diesel would require 25 to 45 percent more hydrogen consumption than would 500 ppm diesel, in addition to improved catalysts.⁵⁸ Hydrogen requirements at lower sulfur levels rise in a nonlinear fashion.

In addition to improvements in design and catalysts, other modifications to refinery operations can contribute to the production of ULSD. For example, high-sulfur compounds in both straight runs and cracked stocks lie predominantly in the higher boiling range of the materials. Thus, reducing the final boiling point for the streams and cutting off the heaviest boiling segment can reduce the difficulty of the desulfurization task. If a refiner has hydrocracking capability, the hydrocracker would be an ideal disposition for these streams. Some refiners making both high- and low-sulfur distillate products may be able to allocate the more difficult distillate blend streams to the high-sulfur product; however, the EPA is in the process of promulgating “Tier 3” non-road engine emission limits around 2005 or 2006, which are expected to be linked to sulfur reduction for non-road diesel fuel.⁵⁹

A processing scheme that has been promoted primarily in Asia and Europe employs a combination of partial hydrocracking and FCC to produce very-low-sulfur fuels. In this scheme a partial conversion hydrocracking unit is placed in front of the FCC unit to convert the vacuum gas oil to light products (distillate, kerosene, naphtha, and lighter) and FCC feed. The distillate product is low in sulfur (less than 200 ppm) and has a cetane number of about 50. The cracked stocks produced in the FCC unit are also lower in sulfur and higher in cetane. The relatively greater demand for distillate relative to gasoline demand in Europe and Asia and the higher diesel cetane requirement are more in keeping with the strengths of this process option than is the case for most U.S. refineries.

A few new technologies that may reduce the cost of diesel desulfurization—sulfur adsorption, biodesulfurization, and sulfur oxidation—are in the experimental stages of development (see discussion on "Developing Technologies and Ultra-Low-Sulfur Aternatives"). Although they are being spurred by the EPA rule, they are unlikely to have significant effects on ULSD production in 2006; however, they may affect the market by 2010. In addition, methods have been developed to produce diesel fuel from natural gas and organic fats, but they still are costly.

NEMS Approach to Diesel Desulfurization Technology

The Petroleum Market Module (PMM) in the National Energy Modeling System (NEMS)⁶⁰ projects petroleum product prices, refining activities, and movements of petroleum into the United States and among domestic regions. In addition, the PMM estimates capacity expansion and fuel consumption in the refining industry. The PMM is also revised on a regular basis to incorporate current regulations that may affect the domestic petroleum market.

The PMM optimizes the operation of petroleum refineries in the United States, including the supply and transportation of crude oil to refineries, the regional processing of these raw materials into petroleum products, and the distribution of petroleum products to meet regional demands. The production of natural gas liquids from gas processing plants is also represented. The essential outputs of the model are product prices, a petroleum supply/demand balance, demands for refinery fuel use, and capacity expansion.

The PMM employs a modified two-stage distillate deep desulfurization process based on proven technologies.⁶¹ The first stage consists of a choice of two distinct units, which accept feedstocks of various sulfur contents and desulfurize to a range of 20 to 30 ppm (Table 2). The second stage also includes a choice of two processing units, which further deep desulfurize the first-stage streams to a level below 10 ppm. The purpose of reducing the sulfur level to 20 to 30 ppm in the first stage, rather than the common goal of 250 ppm or less, is to enable a more accurate representation of costs for processing streams.

The PMM retains the option of conventional distillate desulfurization when 500 ppm sulfur diesel can still be produced (before June 2010). Because the PMM models an aggregation of refinery capacities in each of the refinery regions,⁶² the above representation of multiple processing options is possible, although in reality individual refineries may choose one process over the other on the basis of strategic and economic evaluations.

Individual Refinery Analysis Approach to Diesel Desulfurization Technology

To assess the supply situation during the transition to ULSD in 2006, industry-level cost curves were constructed for this study and matched against assumed demand and imports. The cost curves are the result of a refinery-by-refinery analysis of investment requirements and operating costs for refineries in Petroleum Administration for Defense Districts (PADDs) I through IV. The ULSD production costs were estimated for different groups of refineries based on their size, the sulfur content of the feeds, the fraction of cracked stocks in the feed, the boiling range of the feed, and the fraction of highway diesel produced. The capital and operating costs for the different groups were developed for EIA by the staff of the National Energy Technology Laboratory (NETL).

For the study, a semi-empirical model was developed to size and cost new and retrofitted distillate hydrotreating plants for production of ULSD. Sulfur removal was predicted using a kinetic model tuned to match the limited literature data available on deep distillate desulfurization. Correlations were used in the model to relate hydrogen consumption, utility usage, etc., to the three major constituents of the distillate pool: straight-run distillate, cat-cracker light cycle oil, and coker gas oil. (See Appendix D for a discussion of the assumptions used to construct the model.)

Capital costs ranged from $592 to $1,807 per barrel per day, depending on the size of the unit, whether it was new or retrofitted, and the percentage of straight run feedstock (Table 3). A large hydrotreater using only straight-run distillate derived from high-sulfur crude had the least cost for both new and retrofitted units. The most expensive units were small hydrotreaters running 32 percent cracked stocked, about the average proportion of cracked feedstocks in PADD II.

Expected Developments and Cost Improvements

Recent experience indicates that consistent, high-volume production of ULSD is a technologically feasible goal, although many refineries could face major retrofits or new unit construction. The variation in feedstock concerning both sulfur content and the amount of cracked stock may be influential in the choice of process option and the cost of desulfurization, which may also entail a different allocation of streams to products. Although unconventional desulfurization technologies have been promoted recently by various vendors, none has made sufficient progress toward the commercial stage to warrant consideration by most refiners who must start producing ULSD by June 2006.⁶³

The two-stage desulfurization process can be accomplished through revamping existing units, building new units, or a combination of both. Several aspects of unit design are important. Properly designed distribution trays can greatly improve desulfurization efficiency, in that catalyst bypassing can make it virtually impossible to produce ULSD. Because hydrogen sulfide (H₂S) inhibits hydrodesulfurization reactions, scrubbing of recycle gas to remove H₂S will improve desulfurization. New design or revamps will also include gas quench to help control temperature through the reactor. In the design of a two-stage system, there will be a hot stripper between the two reactors where ammonia and H₂S are stripped from the first-stage product.

As more commercial evidence and cost information become available for diesel desulfurization in the next few years, it will be possible to better assess the technology choices—including equipment requirements, operating conditions, and production logistics—that most refiners will have to make in order to meet the new ULSD standards. However, the EPA’s tight compliance timetable for producing ULSD might short-circuit the learning process for refiners to acquire necessary experience to make cost-effective decisions.⁶⁴ The many caveats within current vendors’ statements must be carefully scrutinized, to avoid overestimating the capability or underestimating the costs for new or revamped distillate hydrotreating facilities. Most vendors state that their goal is to use or revamp a client refiner’s current process units whenever possible. In trying to reach a 10 ppm or lower sulfur target, however, many units may be unsuitable or require major capital outlays. Uncertainty about the level of revamp is a major source of uncertainty in estimating the cost of the ULSD Rule.

Further consolidation of the refinery industry may achieve better economies of scale, although some industry analysts have expressed concern that a shortage of diesel supply could materialize in the short term if some economically challenged refineries exit the diesel market. Catalyst improvements are expected to be one of the main factors in reducing operating costs, both in terms of recycle rate and efficient use of hydrogen. Other factors, such as the dependence of the refinery on distillates, access to lower-sulfur crude, level of competition, and ability to upgrade infrastructure, must also be taken into account. The European experience could also provide valuable insights for U.S. refineries.

Deployment of Desulfurization Technologies

The deployment of diesel desulfurization technologies will hinge on several factors, such as the ability and willingness of refiners to invest, the timing of investment and permitting, the ability of manufacturers to provide units for all U.S. refineries at once, and the availability of engineering and construction resources.

One impediment to acquiring desulfurization upgrades may be the willingness and ability of individual refiners to obtain capital. The EPA estimates that average investment for diesel desulfurization will cost $50 million per refinery, slightly more than the estimated $44 million per refinery required to meet the Tier 2 gasoline sulfur requirement. Most refiners will invest in the gasoline sulfur upgrade because gasoline is their major product. Because U.S. refineries typically produce three to four times as much gasoline as highway diesel fuel, the per gallon investment cost of ULSD will be three to four times as high.⁶⁵

In its Regulatory Impact Analysis, the EPA provided an analysis of capital requirements indicating that the combined annual capital investment for gasoline and diesel desulfurization would be $2.15 billion in 2004 and $2.49 billion in 2005.⁶⁶ The EPA analysis spread the diesel investments over a 2-year period (to reflect “a somewhat more sophisticated schedule for the expenditure of capital throughout a project”) and assumed that the gasoline investments would be incurred in the year before a unit came on line. The EPA concluded that this level of investment should be sustainable by the industry because it is roughly two-thirds of the estimated environmental investments incurred during 1992-1994, when the industry was responding to the 500 ppm highway diesel and oxygenated and reformulated gasoline requirements. Other estimates of ULSD investment costs range from $3 billion to $13 billion (see Chapter 7).

Although not discussed in the EPA’s investment analysis, the 1990s was a period of rationalization for the refining industry, marked by refinery sales, mergers, and closures. Between January 1990 and January 1999, 50 of 205 refineries were closed (4 of which were merged wth adjacent refineries).⁶⁷ The NPC attributes the refinery closures to heightened competitiveness. Although the environmental requirements of the 1990s cannot be pointed to as the cause of the closures, they contributed to the inability of some refineries to compete economically. Refiners who chose not to invest in the 500 ppm sulfur limit (required for highway diesel since 1993) found it more economical to shift their existing high-sulfur diesel production to non-road markets.

Some refiners will be more able than others to obtain capital for Tier 2 gasoline and ULSD projects. Assuming that capital is accessible, a refiner’s willingness to invest in ULSD projects will depend on its assessment of the economics of the market. For instance, a refiner would be less likely to invest if it believed it could not compete favorably with others because the investments would result in a higher cost per gallon. History may lead some refiners to be cautious about investment. In the 1990s refinery upgrades for meeting reformulated gasoline requirements resulted in excess gasoline production capacity. As a result, gasoline margins were depressed, making it difficult for refiners to recoup investments.

Profit margins for ULSD could be depressed if refiners build too much capacity, and the fear of overinvestment could lead some refiners to delay investment until more highway diesel production is required. On the other hand, refiners anticipating inadequate supply of ULSD may choose to invest as early as possible to benefit temporarily from higher margins and sell credits to those that do not invest early. The EPA believes that any lack of investment will be compensated for by the temporary compliance options and credit trading provisions of the ULSD Rule.

Another possible hurdle to the timely adoption of desulfurization technologies is the ability of the engineering and construction industries to design and build diesel hydrotreaters in a timely manner. In addition to providing diesel hydrotreaters, the same contractors will be providing gasoline desulfurization units for the Tier 2 gasoline sulfur reduction requirements that will be phased in between 2004 and 2007. Moreover, engineering and construction requirements will also be expanding outside the United States. The Canadian government has committed to harmonizing gasoline and diesel requirements with the United States. In Europe, refiners will be making upgrades to meet tighter gasoline and diesel requirements in 2005 and have may incentives to produce even cleaner fuels for markets in Germany and the United Kingdom (see discussion in Chapter 6).

In its 2000 study, the NPC provided an analysis of the number of construction projects required for U.S. refiners to provide both gasoline and diesel fuel meeting a 30 ppm sulfur cap. The analysis concluded that “if a diesel sulfur reduction is required for 2006, implementation would overlap significantly with the Tier 2 Rule gasoline sulfur reduction, and engineering and construction resources will likely be inadequate, resulting in higher costs, implementation delays, and failure to meet the regulatory timelines.” The study also concluded that if a 15 ppm diesel standard is required, further investments in new units will be required and there will be a significant risk of inadequate diesel supplies.

The NPC estimated that 89 refineries will require gasoline hydrodesulfurization units by 2004 and that 89 refineries (presumably the same ones) would make upgrades for new highway diesel standards and concluded that if the diesel standard were required within 12 months of completion of Tier 2 gasoline projects, construction labor shortages could occur. The analysis provided peak monthly engineering and construction personnel requirements for five scenarios with different assumptions about the timing and overlap of Tier 2 gasoline and ULSD requirements (Table 4). The scenarios ranged from a “balanced implementation” case, in which one-fourth of the required projects would begin in each quarter of the first year (Scenario A), to highly front-end loaded cases (Scenarios D and E), in which three-fourths of the projects would begin in the first quarter of the first year. Scenarios B and C assumed that refiners would start projects as late as possible.

In the Regulatory Impact Analysis for the ULSD Rule, the EPA conducted its own analysis of the personnel requirements for design and construction services related to the overlapping requirements of the Tier 2 gasoline and ULSD requirements. The analysis provided monthly estimates for each personnel category, assuming that in a given year 25 percent of the projects would be completed per quarter. The monthly estimates were used to develop estimates of the maximum number of personnel required in any given month for the Tier 2 gasoline program alone and for the gasoline and ULSD programs together, both with and without a temporary compliance option. The estimates of the two programs taken together without the temporary compliance option were about double the employment estimates for the Tier 2 gasoline program only, in all three job categories. When the temporary compliance option is taken into account, personnel requirements for the two programs are only about 30 percent higher than for the Tier 2 gasoline program alone.

Because the largest impact is expected to occur in front-end design, where 30 percent of available U.S. personnel are required, the EPA believes that the engineering and construction workforce can provide the equipment necessary for compliance. It appears that the EPA’s criterion for the adequacy of engineering and construction personnel lies somewhere between 30 percent and 50 percent over the personnel requirements of the Tier 2 requirements alone.

The EPA’s estimates without a temporary compliance option are most consistent with the timing assumptions of NPC’s Scenario A. EPA’s analysis indicates that engineering and construction requirements will be lower given the temporary compliance option of the ULSD Rule; however, NPC Scenarios D and E demonstrate that different assumptions about project timing lead to very different estimates for personnel. The range of personnel estimates shown in Table 4 highlights the uncertainty of the estimates.

The EPA’s analysis assumed that a total of 97 units would be added to make Tier 2 gasoline and that 121 diesel desulfurization units would be added for ULSD (Table 5). The expected startup dates for the gasoline and diesel desulfurization units indicate an overlap of 26 gasoline units and 63 diesel units in 2006. The 2006 overlap in gasoline and diesel startups is noteworthy because it is significantly greater than it would have been with ULSD implementation in any other year except 2004.

Another possible hurdle to implementing technology for the ULSD Rule raised by the NPC is the ability of manufacturers to provide critical equipment. As mentioned earlier, the NPC analysis assumed that a sulfur requirement below 30 ppm would require new deep hydrotreaters with reactor pressures in the range of 1,100 to 1,200 psig, requiring thick-walled reactors. As compared with other reactors, the delivery time for thick-walled reactors is longer and the number of suppliers is more limited. Only one or two U.S. companies produce thick-walled reactors, whereas four to six can supply reactors with more typical wall widths. Outside the United States, 10 to 12 companies are able to supply reactors regardless of wall width. This view is at odds with the EPA analysis, which was based on vendor estimates, with reactor pressures in the range of 650 to 900 psig.