Study updates refinery investment cost curves

April 23, 2007
In this article, we have updated and extended the refinery investment cost curves presented in Gary and Handwerk’s “Petroleum Refining: Technology and Economics.

In this article, we have updated and extended the refinery investment cost curves presented in Gary and Handwerk’s “Petroleum Refining: Technology and Economics.”1 2 This article presents the methodological framework, data sources, and normalization procedures used for cost estimation, along with a discussion of the limitations of analysis.

Cost functions are summarized for the main refinery processes.

Data sources

Many different data sources are available to estimate the construction cost of a refinery unit, including government organizations, private and public companies, commercial databases, trade and academic publications, and press releases for licensors and companies.

Oil & Gas Journal and Hydrocarbon Processing semiannually report planned capacity expansions for refineries (OGJ Apr. 16, 2007, p. 18). The data coverage is similar in both surveys and provides information on refining capacity that will be added at each location by project type (increment of capacity added; total capacity after construction; revamping, modernization or debottlenecking; expansion); licensor, engineering company, and constructor; estimated completion date; estimated cost; and project status (abandoned, engineering, feed, completed, maintenance, planning, under construction).

The data are subject to availability, and because project descriptions are not provided, are of limited use for cost estimation.

The best public sources of information for cost data are technical articles found in OGJ and other trade publications, material presented at such professional conferences as National Petrochemical and Refining Association’s annual meeting, and industry studies.

Robert Meyer’s “Handbook of Petroleum Refining Processes” is an excellent source of process and economic data for a wide compendium of technologies. Maples, Raseev, Sadeghbeigi, and various other texts provide cost estimates for mid-1990 and earlier configurations and technologies.3-6

Commercial databases from Baker & O’Brien Inc.; Purvin & Gertz Inc.; Solomon Associates LLC; Turner, Mason & Co.; and other consultancies such as Independent Project Analysis Inc. are widely used throughout industry for benchmarking and policy studies. Cost information in commercial sources is some of the best available, having been collected from projects and other assignments over an extended period of time and at various levels of detail.

Each agency considers its database and modeling software proprietary, however, and so the quality, reliability, and consistency of commercial sources remain difficult to assess and compare.

Process technologies

Table 1 shows the primary process technologies used in petroleum refining. For each process, one or more subcategories are typically defined, based on technology attributes, operating conditions, or feedstock.

Function specification

Capital costs for refining units are frequently specified as a function of capacity and scaled using the power-law relation as shown in Equation 1 (see accompanying equation box).

The value of x varies with each unit and is frequently assumed to vary between 0.5 and 0.7.7 8 To estimate x, data are collected for units of comparable design and technology, and then the value is determined empirically through regression analysis.

For some units, it is not possible to develop useful curves from its feed capacity, while in other cases, if a large number of factors influence the design parameters, unrealistic uncertainty bounds may result. To manage these issues we frequently cross-correlated cost data, based on two or more process descriptors, and in cases where data were extremely sparse, (unit) cost data on a per-barrel basis are presented.

Analysis limitations

Many engineering estimates are available, but the actual (finished) cost is preferred to avoid estimator bias and to assess actual system performance. Ideally, we would only use actual cost data in our analysis, but the sample sets in most cases would be too sparse. We therefore, found it necessary when building cost curves to base the assessment on both engineering estimates and actual cost records.

Cost curves are meant to represent typical, or average, values and stand as point estimates rather than in terms of intervals or ranges. Cost functions represent an “average” refinery, which of course does not exist as an actual plant but is useful in developing conceptual cost estimates in early stages of assessment and design.

Every cost estimate involves uncertainty due to differing qualities of equipment fabrication, design differences, market conditions, vendor profit, and other considerations. The cost curves can be assumed to have an accuracy limited to ±25%.

Working capital, inventories, start-up expense, the cost of land, site preparation, taxes, licenses, permits and duties are not considered in the estimation. The cost curves that we established are inappropriate for benchmarking studies.

The level of uncertainty in estimation can be reduced through a detailed front-end engineering design based on site-specific information. For definitive economic comparisons and estimations, other factors such as feedstock, production specification, operating conditions, design options, and technology options must be considered.

Utility requirements

Utility requirements for each process are usually presented on a per-barrel unit feed or product basis and correspond to average characteristics associated with the mid-point of the construction cost curve. Wide variability in utility values can result, depending upon the capacity of the unit and other process-specific factors.

Table 2 shows typical utility cost data.

Normalization

The cost data of units of roughly comparable design and technology are normalized with respect to construction requirements, process specifications, location, and time of installation:

  • Dependent variable. The normal basis in computing construction cost is the liquid-volume fraction of the crude that is fed to each process, but in several units (alkylation, polymerization, aromatics manufacture), a better descriptor is the barrels of product rather than the feed.

    In other units (isomerization, hydrotreating, catalytic reforming, hydrogen production), it is necessary to cross-correlate the cost with other factors, while in gas processing and sulfur manufacture, the liquid-volume basis needs to be replaced with the measures cubic feet (gas) and long tons (sulfur).
  • Project type. Projects are classified according to new capacity, expansion of existing capacity, or a revamp or modernization of existing facilities. Cost estimates in this article pertain exclusively to grassroots construction, limited to equipment inside the battery limits (ISBL) of each process, and including materials and labor; design, engineering, and contractor’s fees; overheads; and expense allowance.
  • Off-site expenses. These include the cost and site preparation of land, power generation, electrical substations, off-site tankage, or marine terminals. Off-site costs vary widely with the location and existing infrastructure at the site and depend on the process unit. ISBL costs do not include off-site expenses.
  • Location. It is common practice to state cost estimates relative to the US Gulf Coast because this location has favorable construction costs relative to other domestic and international markets. Design requirements, climate, regulations, codes, taxes, and availability and productivity of labor all influence the cost of construction, and to some extent, the operating cost of a facility.
  • Time. The purchase cost of processing equipment in refining is generally obtained from charts, equations, or quotes from vendors at a particular date, usually month and year. Factors such as regulatory requirements, heavy feedstock, and the cost of materials may increase costs and inflate refinery investment over time, while other factors may act to lower cost, such as improvements due to a technological and process nature like improved catalyst, control and instrumentation, and materials technology.

Time captures both long-term dynamics such as improvements in technology and operational efficiency, as well as local effects such as the cost of steel, permit requirements, and pollution control.

Nelson-Farrar cost indexes

An estimate of the purchase cost at a given time t2 is obtained by multiplying the original (quoted) cost at time t1 by a ratio of cost indexes (Equation 2).

The Nelson-Farrar (NF) construction cost index normalizes cost during the time required to construct a process unit. The NF cost index is unsuitable for determining the cost for refineries or process units that are more than 3-5 years old. The NF cost index also does not account for productivity attained in design, construction, or management skills.

The NF construction cost index (Fig. 1) is published in OGJ in the first issue of each month.

The NF operating cost indices compare operating costs over time (Table 3). Unlike the construction index, the operating cost indices are normalized for the productivity of labor, changes in the amounts and kinds of fuel used, productivity in the design and construction of refineries, and the amounts and kinds of chemicals and catalysts used (OGJ, Dec. 30, 1985, p. 145; OGJ, Oct. 2, 1989, p. 90)

Comparisons of operating indices can be made for any two periods of time.

Unit complexity

The Energy Information Administration division of the US Department of Energy publishes data on US refineries that are similar in format to those published annually by OGJ.9

Crude distillation, vacuum distillation, coking, thermal processes, catalytic cracking, catalytic reforming, catalytic hydrocracking and catalytic hydrocracking are described in terms of charge capacity, which describes the input (feed) capacity of the facilities. Production capacity represents the maximum amount of product that can be produced and are presented for alkylation, polymerization/dimerization, aromatics, isomerization, lubricants, oxygenates, hydrogen, coke, sulfur, and asphalt facilities.

Wilbur Nelson introduced the concept of complexity factor to quantify the relative cost of components that make up a refinery (OGJ, Nov. 29, 1976, p. 68; OGJ, Jan. 10, 1977, p. 86). Nelson assigned a complexity factor of 1 to the atmospheric distillation unit and expressed the cost of all other units in terms of their cost relative to distillation.

For example, if a crude distillation unit of 100,000 b/d capacity cost $10 million to build, then the unit cost/daily barrel of throughput would be $100/b/d. If a 20,000 b/d catalytic reforming unit cost $10 million to construct, then the unit cost is $500/b/d of throughput and the “complexity” of the catalytic reforming unit would be 500/100 = 5.

The complexity factor of process unit Ui of capacity Qi is defined in Equation 3.

Various methodological issues limit the use of complexity factors in estimating cost. For example, complexity factors do not account for the impact of capacity on cost because the complexity factor is capacity-invariant, and trends in complexity factors change slowly (or not at all) over time (Table 4), making their application suspect.

Refinery complexity

A refinery’s complexity indicates how complex it is in relation to a refinery that performs only crude distillation. The complexity index of a given refinery, R, is determined by the complexity of each individual unit weighted by its percentage of distillation capacity (Equation 4).

A simple refinery is typically defined by γ(R)<5; a complex refinery by 5≤γ(R)≤15; and a very complex refinery by γ(R)>15. Refinery complexity is an often-cited industry statistic and is a useful tool in comparative analysis, being frequently used as a correlative or descriptive variable in marketing and valuation studies (OGJ, Sept. 19, 2005, p. 43).

Refining example

Tables 5 and 6 show ExxonMobil Corp.’s charge and production capacity of its Louisiana refineries. The complexity index of the Baton Rouge refinery is computed in Table 7 as 13.4, a complex refinery.

Generalized complexity

The complexity index can be generalized across any level of aggregation, such as a company, state, country, or region (Equation 5).

Process cost functions

The first steps in every refining process are desalting and the subsequent separation of crude oil into fractions by atmospheric and vacuum distillation.

Table 8 summarizes cost functions for different types of refining processes in terms of the function parameters α and β. For example, Cost (desalter) = 0.44•Capacity0.555, where α = 0.44 and β = 0.555.

Delayed coking, visbreaking, fluid coking, and flexicoking are common thermal cracking units.

Fig. 1 shows a typical cost curve, in this case, crude distillation.

Catalytic cracking is accomplished through the use of a catalytic agent and is an effective process for increasing the yield of gasoline from crude oil. Fluid catalytic cracking cost curves for distillate and resid feed are given by Cost(Catalytic cracking; distillate feed) = 24.67•Capacity0.461 and Cost(Catalytic cracking; resid feed) = 32.98•Capacity0.510.

Table 9 also shows:

  • Catalytic hydrocracking capital investment costs at 1,000 and 3,000 scf/bbl hydrogen consumption.
  • Hydrodesulfurization capital costs processing a naphtha feedstock, a distillate feedstock, and a residual feedstock.
  • Semiregenerative and continuous-regeneration catalytic reforming costs. Continuous regeneration has been used in the majority of the new catalytic reformers built in the US, although fixed-bed units are still constructed depending upon product requirements.
  • Paraffin isomerization unit capital costs for butane and pentane-hexane feed for once-though and recycle processes.
  • Investment cost of alkylation units expressed in terms of alkylate production. Capital costs for sulfuric and hydrofluoric alkylation units are generally comparable, and therefore the investment curve can be applied for either unit. Composition of the costs, however, varies with the technology. The sulfuric acid process has an expensive reactor section and requires refrigeration, whereas HF alkylation uses feed driers, product treating, regeneration equipment, and more exotic metallurgy.
  • Investment cost curves for hydrogen-production units that use steam reforming of natural gas, catalytic reforming, and partial oxidation.
  • Gas processing and amine-treatment capital investment costs.
  • Investment costs for sulfur-removal technologies including the Claus process, which uses both thermal and catalytic-conversion reactions and the S-zorb process.
  • Solvent dewaxing capital costs.
  • Capital costs and utility consumption for methyl tertiary butyl ether and tertiary amyl methyl ether production based on a one-stage design. The capital costs are expressed in terms of total hydrocarbon feed (excluding alcohol), and therefore the iso-olefin content of the feedstock must be known for cost estimation.

References

  1. Gary, J.H., and Handwerk, G.E., Petroleum Refining: Technology and Economics, 4th ed., New York: Marcel Dekker Inc., 2001.
  2. Gary, J.H., Handwerk, G.E., and Kaiser, M.J., Petroleum Refining: Technology and Economics, 5th ed., Boca Raton, Fla.: CRC Press, 2007.
  3. Meyers, R.A., Handbook of Petroleum Refining Processes, 3rd ed., New York: McGraw-Hill Cos., 2004.
  4. Maples, R.E., Petroleum Refining Process Economics, 2nd ed., Tulsa: PennWell, 2000.
  5. Raseev, S., Thermal and Catalytic Processes in Petroleum Refining, New York: Marcel Dekker, 2003.
  6. Sadeghbeigi, R., Fluid Catalytic Cracking Handbook, 2nd ed., Houston: Gulf Publishing Co., 2000.
  7. Peters, M.S., Timmerhaus, K.D., and West, R.E., Plant Design and Economics for Chemical Engineers, 5th ed., Boston: McGraw Hill Cos., 2003.
  8. Seider, W.D., Seader, J.D., and Lewin, D.R., Product & Process Design Principles, New York: John Wiley & Sons, 2004.
  9. Annual Energy Review 2004, DOE/EIA-0384 (2004), US Department of Energy, Energy Information Administration, Washington DC, August 2005.
  10. Nelson, W.L., Guide to Refinery Operating Cost (Process Costimating), 3rd ed., Tulsa: Petroleum Publishing, 1976.

The authors

Click here to enlarge image

Mark J. Kaiser ([email protected]) is a research professor at the Center for Energy Studies at Louisiana State University, Baton Rouge. His primary research interests are related to policy issues, modeling, and econometric studies in the energy industry. Before joining LSU in 2001, he held appointments at Auburn University, the American University of Armenia, and Wichita State University. Kaiser holds a PhD in engineering from Purdue University.

Click here to enlarge image

James H. Gary is professor emeritus in the department of chemical engineering at the Colorado School of Mines, Golden, Colo. He has also served as an engineer at Standard Oil Co. (BP Oil Co.). Gary holds a BS and MS in chemical engineering from Virginia Polytechnic Institute and a PhD from the University of Florida. He is a member of AIChE and the American Association for the Advancement of Science. Gary is a registered professional engineer in Colorado and Ohio.