ARTICLE IN PRESS
Energy 32 (2007) 1075–1092 www.elsevier.com/locate/energy
Fuel speciﬁcation, energy consumption and CO2 emission in oil reﬁneries Alexandre Szklo, Roberto Schaeffer Energy Planning Program, Graduate School of Engineering, Federal University of Rio de Janeiro, 21.941-972 Cidade Universita´ria, Ilha do Funda˜o, Centro de Tecnologia, Bloco C, Sala C-211, Caixa Postal 68565, Rio de Janeiro, RJ, Brazil Received 17 May 2006
Abstract The more stringent environmental quality speciﬁcations for oil products worldwide are tending to step up energy use and, consequently, CO2 emissions at reﬁneries. In Brazil, for example, the stipulated reduction in the sulfur content of diesel and gasoline between 2002 and 2009 should increase the energy use of Brazil’s reﬁning industry by around 30%, with effects on its CO2 emissions. Thus, the world reﬁning industry must deal with trade-offs between emissions of pollutants with local impacts (due to fuel speciﬁcations) and emissions of pollutants with global impacts (due to the increased energy use at reﬁneries to remove contaminants from oil products). Two promising technology options for reﬁneries could ease this clash in the near-to-mid term: the reduction per se of the energy use at the reﬁnery; and the development of treatment processes using non-hydrogen consuming techniques. For instance, in Brazilian reﬁneries, the expanded energy use resulting from severe hydrotreatment to comply with the more stringent speciﬁcations of oil products may be almost completely offset by energy saving options and alternative desulfurization techniques, if barriers to invest in technological innovations are overcome. r 2006 Elsevier Ltd. All rights reserved. Keywords: Oil reﬁning; Fuel speciﬁcations; Energy use; Treatment processes; Brazil
Corresponding author. Tel.: +55 21 2562 8760; fax: +55 21 2562 8777.
This situation is worsened by the more restrictive environmental quality requirements for oil products2 and the shift towards low-grade crude oils in the world reﬁning industry.3 These two factors increase the energy use at reﬁneries. This increase depends on the type of reﬁnery, in terms of conversion capacity, the feedstock processed and the mix of products obtained. It also varies widely, with a considerable order of magnitude. For instance, it is estimated that the total amount of crude oil processed by US reﬁneries increased, on average, some 5%, merely to comply with the diesel and gasoline stricterspeciﬁcations
E-mail address: [email protected]
(R. Schaeffer). For example, the two largest Dutch reﬁneries in 1995 posted energy use at 7% of the crude oil feedstock . There are reﬁneries in the USA with energy use at almost 15% of the crude oil processed (EIA/DOE ). Actually, in recent years, energy use in US reﬁneries has been virtually stable, ranging between 9% and 10% of the crude oil processed [3,4]. In 2001, total ﬁnal energy use was 3191 PJ, and primary energy use was 3268 PJ . The difference between primary and ﬁnal electricity consumption is relatively low due to the small share of electricity consumption in the reﬁnery and relatively large amount of self-produced electricity.
2 For a detailed discussion of fuel quality speciﬁcations worldwide and in Brazil, see Faiz et al , Guru et al. , Yamaguchi et al. , Junior , Carvalho , Tavares , Plantenga and Leliveld , Song  and Brunet et al. . 3 Reﬁneries are facing many challenges, e.g. increased fuel quality demands, heavier crude feeds and changing product mix, increasing and more volatile energy prices, need to reduce air pollutant emissions, increased pressure on proﬁtability, and increased safety demands. These challenges will affect the industry and technology choice profoundly .
1. Introduction Part of the technological development of the oil reﬁning industry worldwide is justiﬁed by the fact that this is an industrial activity with high fossil fuel consumption and consequently high CO2 emissions. Oil reﬁning processes are energy-intensive, requiring considerable amounts of direct or indirect heat. Between 7% and 15% of the crude oil input is used by the reﬁnery processes.1
0360-5442/$ - see front matter r 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.energy.2006.08.008
ARTICLE IN PRESS 1076
A. Szklo, R. Schaeffer / Energy 32 (2007) 1075–1092
imposed in the USA at the start of the 1990s . This problem is so worthwhile that Shell, for example, assigned the 18 MtC increase in its carbon emissions in 2003 (compared to 2002) speciﬁcally to adapting its reﬁneries for the production of oil products with lower sulfur content . Moreover, stringent environmental quality speciﬁcations stipulated for oil products, particularly diesel and gasoline, affect other speciﬁcations for these oil products, unrelated to the environment, such as the diesel lubricity4 and the gasoline octane number. Actually, FCC gasoline is by far the most important sulfur contributor in gasoline, up to 85–95% . However, FCC gasoline contains also a great quantity of oleﬁns (20–40 wt%), which provides it with a fairly good octane number. Therefore, the challenge is to eliminate a maximum of the sulfur impurities with a minimum oleﬁn saturation . Nevertheless, hydrotreating FCC gasoline through a conventional process would lead to a signiﬁcant octane loss. In addition, stricter environmental quality requirements fragment the oil products market, creating boutique fuels , and reduce supply efﬁciency, associated with the shipment, storage and distribution of oil products (the ability to combine, sequence, and ship batches of similar products together is a key contributor to the reliability of the oil products distribution system). As each reﬁnery is unique in terms of feedstock, products and processes, stricter environmental constraints tend to curtail the supply ﬂexibility of oil products distribution systems. For instance, markets that are heavily dependent on imported oil products become less ﬂexible, such as the US market for gasoline (and diesel, to an increasingly extent), as well as the European diesel market.5 The world reﬁning industry faces challenges associated with the trade-off between pollutant emissions with local impacts and pollutant emissions with global impacts, derived from the production and use of oil products (including oil reﬁnery emissions and oil products combustion). In practical terms, this trade-off consists of the fact that the production of diesel or gasoline with extremely low sulfur contents normally requires more energy. Consequently, the production of ‘‘cleaner’’ gasoline and diesel, in terms of sulfur content, results in higher energy consumption and higher emissions of carbon dioxide (CO2), a greenhouse gas, by reﬁneries. Therefore, more stringent sulfur speciﬁcations impose challenges to oil reﬁners, but also drive technical innovation [11,19, 20]. Deep reduction of gasoline and diesel sulfur must be made without compromising their quality (e.g., gasoline octane number, diesel lubricity); losing diesel and gasoline yields must also be avoided by treating units 4 Additives are often used (fatty acids or esters) to increase the lubricity of diesel with extremely low sulfur contents (or severely hydrotreated, or Ultra Low Sulfur Diesel). These additives contain a polar group attracted by metal surfaces, forming a ﬁne lubricating layer. 5 On this matter, see Hackworth and Shore .
; and the treating units must not increase energy use and CO2 emissions of reﬁneries. In order to deal with these challenges, through supplyside actions,6 there are three promising technological alternatives for reﬁneries: (a) alternatives for saving energy at reﬁneries; (b) less severe or non-conventional treatment process alternatives (replacing severe hydrotreating); and (c) oil gasiﬁcation and the removal of CO2 at the reﬁnery. The ﬁrst two options could be implemented in the near medium terms, while the last option would require further technology development and construction of new facilities before implementation would be possible. There is also a feasible alternative outside oil reﬁning processes, which is the formulation of the oil products with non-oil products components (and/or additives) having low or even nil contaminant content levels. For sulfur and aromatics, this would involve the blending of reﬁning products with biofuel, particularly ethanol for gasoline7 and biodiesel for diesel8 and/or with synthetic fuels produced through the Fischer-Tropsch route—for instance, based on natural gas using the gas-to-liquids (GTL) route. However, this paper emphasizes alternatives directly related to oil reﬁning (or undertaken inside reﬁneries). The analysis focuses also on the two ﬁrst alternatives mentioned above, for the near to medium terms (alternatives (a) and (b)), since a detailed discussion of the alternative (c) may be found in Szklo and Schaeffer . Section 2 of this paper describes the energy use in oil reﬁneries. Section 3 discusses the conventional options for specifying oil products in oil reﬁneries, highlighting the case of diesel and gasoline. Section 4 estimates the increase in energy use and CO2 emissions at the Brazilian reﬁning sector, due to its actual sulfur reduction program for diesel and gasoline. Through the Brazilian example, this estimate shows the need to seek technological alternatives to the trade-off between local emissions of atmospheric pollutants (due to fuel speciﬁcations) and emissions of greenhouse gases by oil reﬁneries (due to the expanded use of energy resulting from the operation of severe hydrotreatment units). Sections 5 (energy savings options) and 6 (alternative desulfurization techniques) propose and estimate the impact on energy savings (and consequently on CO2 emissions) of technical alternatives for saving energy or removing sulfur in oil reﬁneries. This analysis is carried out for the near to medium terms, assuming that there will be 6 From the standpoint of Integrated Energy Planning, demand-side actions (energy policies associated with lower consumption of oil products by the transportation sector, for instance) will probably be effective here, but their analysis is outside the scope of this paper. 7 Or ETBE, based on ethanol. 8 It is also possible to add ethanol to diesel. However, this signiﬁcantly affects important properties of the diesel, such as its cetane number, lubricity, volatility, ignition temperature and stability.
ARTICLE IN PRESS A. Szklo, R. Schaeffer / Energy 32 (2007) 1075–1092
no drastic changes in the ﬁnished products blending and reﬁning processes. Finally, Section 7 presents the ﬁnal remarks of the paper, focusing on the results of the Brazilian case and the barriers faced by reﬁners in Brazil and worldwide to invest in the technical innovations analyzed in this paper. 2. Energy use at oil reﬁneries Heat and electricity are used by reﬁneries, with heat outstripping electricity. Heat is used directly (in the furnaces) or indirectly (as steam). Meanwhile, electricity may be supplied partially by co-generation plants.9 The fuel required to produce steam and electricity comes mainly from process wastes: reﬁnery gas, residual fuel oils (fuel oil, vacuum wastes and asphalt wastes) and FCC coke. However, the composition and quality of the energy sources consumed by a reﬁnery vary considerably. Moreover, the set of fuels used by a reﬁnery is the outcome of the balance between the energy required by the processes, the types of crude processed, the current constraints on emissions and economic analyses. In general, reﬁnery processes are energy-intensive. The most intensive process (in terms of energy demands per barrel processed) is the production of lube-oils, which takes place in only a few reﬁneries (some 1500 MJ/b).10 Other processes with high speciﬁc energy use include etheriﬁcation (production of MTBE and TAME), alkylation and isomerization (Table 1). However, once again, the production of MTBE and the alkylation and isomerization units are found merely at reﬁneries focused on higher-grade gasoline. In other words, these energy-intensive processes (in terms of energy use by processed feed) are not the main units using energy in the world reﬁning industry, because their feed are not signiﬁcant, in terms of volume processed. Thus, in absolute terms, the energy use of a reﬁnery is concentrated in a few processes that are not the most energy-intensive (in terms of energy use per barrel), but instead process large volumes of feedstock. The Vacuum and Atmospheric Distillation plants usually total 35–45% of the energy use of a reﬁnery , because any barrel of oil entering a reﬁnery runs through the topping separation units. This same reasoning explains why, in advanced reﬁning industries having considerable conversion capacity and focused on fuels with low contaminant content levels, hydrotreatment units post considerable energy use, in absolute terms.11 These plants pre-treat catalytic process feeds and enhance the quality of the ﬁnished products. In addition, more complex reﬁneries tend to be larger (processing more crude oil) with more energy-intensive 9 This is the case with some Brazilian reﬁneries such as REDUC, which generated 373 GWh in 2003, with an installed capacity of 63.3 MW, with natural gas accounting for 49% of this power generation, while reﬁnery gas accounted for 15%, fuel oil for 23% and FCC coke for 13% . 10 Particularly in view of the energy use associated with solvent recycling. 11 For example, these plants accounted for 19% of energy use in US reﬁneries in 1996 . In 2001, this proportion rose to 24% .
Table 1 Energy use by reﬁning process units (MJ/b)
Atmospheric distillation Vacuum distillation Visbreaking Delayed coking Fluidized catalytic cracking (FCC) Hydrocracking (HCC) Hydrotreating Catalytic reform Alkylation by H2SO4 Alkylation by HF Etheriﬁcation Isomerization through isobutane Isomerization through isopentane Isomerization through Isobutylene Lube-oils production
90 50 100 120 50 170 60 220 350 430 310 360 100 480 1500
200 120 150 250 180 340 180 360 360 430 600 360 250 480 1500
Notes: Average values based on Energetics . This consumption includes power generation losses, based on a heat rate of 10,500 BTU/ kWh (or First Law Efﬁciency of 32.5%, which is compatible with the cogeneration power plants found at reﬁneries).
units. However, there are small low-complexity reﬁneries with high CO2 emissions, because they operate with less-efﬁcient processes in terms of energy use and selectiveness. An analysis of European reﬁneries is quite explanatory, showing that there are reﬁneries with the same Nelson Index,12 but with very different speciﬁc CO2 emissions (emissions per feedstock processed). For instance, the speciﬁc CO2 emissions of European reﬁneries with low conversion capacities (Nelson Index equal to 2) vary from 250 to 550 ktCO2/Mt of crude, while, for reﬁneries with higher conversion capacities (Nelson Index equal to 8), there is a range between 200 and 350 ktCO2/Mt of crude . Nevertheless, it is possible to infer from the European reﬁning data that the more complex reﬁneries have higher speciﬁc CO2 emissions on average than the less complex ones. However, the standard deviation in the speciﬁc CO2 emissions by less complex reﬁneries is higher. Thus, the increase in energy use (and consequently speciﬁc CO2 emissions) by more complex reﬁneries is due solely to their higher conversion and treatment capacities. For the less complex reﬁneries, higher energy use and CO2 emissions may also be associated with inefﬁciencies in their production processes.
12 The Nelson Complexity Index is a measure of secondary conversion capacity in comparison to the primary distillation capacity of any reﬁnery. This index was originally developed by Wilbur L. Nelson in 1960 to quantify the relative costs of the components that constitute the reﬁnery. Nelson assigned a factor of one to the primary distillation unit. All other units are rated in terms of their costs relative to the primary distillation unit also known as the atmospheric distillation unit.
ARTICLE IN PRESS 1078
A. Szklo, R. Schaeffer / Energy 32 (2007) 1075–1092
3. Hydrotreatment and blending for diesel and gasoline speciﬁcations The quality speciﬁcations for oil products at a reﬁnery are conventionally achieved through hydrotreatment (HDT) units for intermediate and ﬁnished products and through blending for ﬁnished products. The severity levels of the HDT process determine the ﬁnal product speciﬁcation. Mild HDT is normally used to remove sulfur and oleﬁns. More severe HDT additionally removes nitrogen compounds, higher sulfur compound content and aromatic rings. The HDT process catalysts are selective for removing sulfur and nitrogen compounds, and metals and or other contaminants . For example, diesel hydrotreatment involves the desulfurization and hydrogenation of insaturates (oleﬁns and aromatics), in order to increase the diesel cetane number . Severe hydrodesulfurization of the diesel aims at obtaining low sulfur content levels (8 ppm), in addition to fuel stabilization (operation at low pressures of around 45 bar). The blending operation is complex. It is designed to comply with technical and economic reﬁning constraints, which encompass different and not always harmonious ﬁnished products speciﬁcations. Actually, the diesel and gasoline speciﬁcations13 have different attributes, whose values are nationally or regionally deﬁned, according to the regulations stipulated for each marketplace, but that tend to converge to some extent in major consumer markets (particularly Western Europe and North America). This convergence affects the global oil products market, because Western Europe is the main diesel import zone and USA is the main import zone for gasoline and gasoline blending components . Thus, diesel and gasoline speciﬁcations tend to converge in the global reﬁning segment as follows: 1. Impose constraints on the sulfur content of both oil products: This initially involves the challenge of boosting desulfurization capacity (HDS), which steps up the energy use at reﬁneries. Depending on its severity, the hydrotreatment process energy use varies between 60 and 180 MJ/b . These processes also require hydrogen, between 50 and 350 Nm3/m3, depending on the HDS feed , which results in a higher hydrogen production at the reﬁnery, and consequently another energy use increase. For example, steam reform of natural gas to produce hydrogen requires a consumption of around 50 GJ/tH2 . An alternative (or sometimes complementary) option to HDS is to blend the feedstock entering the reﬁnery, favoring sweeter crudes. However, these crudes have a price-premium compared to the sourer crudes, requiring a balance between the ﬁxed costs (represented by the increase in the HDS capacity) and the variable costs (represented by the sweet oil pricepremium), in order to deﬁne the best possible strategy for reducing the sulfur content of the oil products. 13 This Section focuses on the analysis of gasoline and diesel, which are the main products of the world reﬁning industry .
2. Increase gasoline octane number: The technological options most commonly used include boosting the capacity of the following processing units: catalytic reform (however, to the detriment of increasing the aromatics contents of the gasoline), FCC, alkylation and isomerization. Additives may also be used, such as ethers (for example, MTBE) and organic alcohols (ethanol). 3. Reduce the Reid vapor pressure (RVP) of the gasoline (reduction of evaporative emissions): The usual procedure is to adjust the ﬁnal gasoline blending through its components and additives. For example, the debutanization of the gasoline reduces its vapor pressure. However, the removal of butane from the gasoline also lowers its octane number and steps up its sulfur content (per barrel o gasoline). The addition of ethanol to gasoline affects its Reid vapor pressure, compared to the increase that occurs with MTBE. This may impose constraints on the amount of ethanol that may be added to gasoline . 4. Increase the cetane number of the diesel: The cetane number of the diesel may be increased through boosting the HDT capacity, including HDA, and the hydrocracking (HCC) capacity, depending on the characteristics of the feedstock. There are additives that boost the cetane number . These compounds, frequently organic nitrates, improve cold-start performance, reduce combustion noise, and may reduce particulate matter (PM) emissions. However, at the same time, cetane boosters increase the ﬂammability of the fuel, are potentially more hazardous due to ultraﬁne particle emissions, and degrade the storage stability of the fuel . Finally, the cetane number may also be increased through redeﬁning the diesel blending, for instance, by removing the FCC cycle oil from it (mainly due to its high aromatic content). However, this poses a trade-off between the diesel reﬁning yield (quantity) and the speciﬁcations of this oil product (quality). Ceteris paribus, by limiting the diesel pool to higher-grade components, the reﬁnery will produce smaller amounts of diesel. 5. Reduce diesel smoke and the emission of particulates: This reduction may be achieved through careful blending of the diesel, for instance by lowering the T9514 and removing the FCC cycle oil from the diesel pool. Once again, this requires a trade-off between the quantitative output and the grade of the diesel produced by the reﬁnery. 6. Lower the aromatics content of gasoline and diesel: Normally, it involves the reduction in the use of catalytic reform units , increasing, for example, the use of the alkylation process to produce high-octane gasoline. For diesel, and even for gasoline, this may also be achieved through wider use of HCC units, which, in the case of diesel, also increase the cetane number, while lowering the octane number of gasoline. For treatment, the hydrodearomatization units (HDA) are particularly noteworthy. In sum, the use of HCC and HDT units and modiﬁcations of the gasoline and diesel blends are the most 14
Temperature at which 95% of diesel is distilled.
ARTICLE IN PRESS A. Szklo, R. Schaeffer / Energy 32 (2007) 1075–1092
Table 2 Gasoline components and quality speciﬁcations High octane rating? (RON+MON)/2489.5
Low RVP? (o7)
Low oleﬁns content? (o14% vol)
Low benzene? (o1% vol)
Low aromatics? (o35%vol)
Low sulfur content? (o150 ppm)
Basic blend Butanes Alkylates Isopentane Isomers Light FCC Naphtha Heavy FCC Naphtha Reformed HCC Naphtha Straight-run gasoline
YES YES YES NO YES NO YES NO NO
NO YES NO YES NO YES YES NO DEPENDS
YES YES YES YES NO DEPENDS YES YES YES
YES YES YES YES NO YES NO YES YES
YES YES YES YES NO NO NO YES YES
YES YES YES YES DEPENDS(2) DEPENDS YES YES DEPENDS
Additives MTBE Ethanol
Notes: (1) the effect of the ethanol on the gasoline RVP is greater than that of the MTBE. (2) ‘‘DEPENDS’’ means that it ‘‘DEPENDS on the feedstock of the unit. Source: Based on Yamaguchi et al. .
Table 3 Diesel components and quality speciﬁcations
HCC diesel Light cycle oil (LCO) Straight-run diesel Diesel HDS mild Severe HDS Diesel HDA diesel (1)
High cetane number? (450)
Low sulfur content? (o0.05% weight)
Ultra low sulfur diesel? (o0.005% weight)
Low aromatics? (o20% vol)
YES NO DEPENDS on the load YES YES YES
YES YES YES
DEPENDS on the load NORMALLY YES
Notes: (1) Hydrodearomatization (HDA) is a high-energy-intensive process that, in addition to removing sulfur from low-reactive compounds, saturates the aromatics and consequently increases the diesel cetane number. Source: Based on Yamaguchi et al. .
conventional ways for meeting oil products speciﬁcations. For the blending process, the gasoline pool has more optional components and additives than the diesel pool (Tables 2 and 3). Nevertheless, since FCC gasoline represents 30–40% of the total gasoline pool, and is by far the most important sulfur contributor in gasoline, up to 85–95% ,15 merely altering the blend of the gasoline pool does not solve the issue of deeply removing sulfur from it.16 15 The most important class of sulfur compounds present in FCC gasoline is made of thiophene and its light alkyl derivatives in addition to benzothiophene. Most of the sulfur impurities present in FCC gasoline (mainly thiophene and short alkyl chain thiophenes) are not present in the FCC feedstocks. Therefore, several possibilities (in fact two main routes a priori) have to be considered to account for the formation of the ‘‘organic sulfur compounds’’ of the gasoline boiling range during the FCC process. 16 Commercial gasoline is made up of different fractions coming from reforming, isomerization and FCC units. Those coming from the reforming and isomerization units are produced from distillation cuts, and consequently contain little or no sulfur because the sulfur containing compounds present in crude petroleum have high boiling points and the feedstocks used in the isomerization and reforming units are hydrotreated. On the opposite, the atmospheric residues or the vacuum distillates which constitute FCC feedstocks contain signiﬁcant amounts of sulfur, 0.5–1.5 wt% .
Therefore, Ceteris paribus, there is an even balance between the oil product grade and quantity at the reﬁnery. A more tightly speciﬁed pool of an oil product is made up of fewer reﬁnery streams. For example, if the purpose of the reﬁner is to produce lower grade diesel (in larger quantities) the pool of this oil product may be equal to the sum of the following intermediate streams: straight-run diesel, HCC diesel (if HCC installed), HDT diesel (mild or severe), FCC LCO,17 and a fraction of the kerosene cut. However, if the purpose is instead to produce ultra low sulfur diesel (ULSD),18 the pool should be based on HCC diesel, diesel severely hydrotreated, and a kerosene fraction speciﬁed for the diesel pool and hydrotreated . 4. Trade-off between sulfur content and CO2 emissions— Brazil’s reﬁning case study Having analyzed the energy use by reﬁneries (Section 2) and the conventional speciﬁcation alternatives for key oil products (diesel and gasoline) in Section 3, it is possible now to move on to the problem of the trade-off between 17
Light cycle oil. Diesel with a sulfur content of below 15 ppm.
ARTICLE IN PRESS A. Szklo, R. Schaeffer / Energy 32 (2007) 1075–1092
Table 4 Quality speciﬁcations requirements for Brazilian diesel (2002–2009) 2002
Density 20/4 1C Sulfur content (ppm) Centane number %T360 (ASTM D86)
0.820–0.865 2000 42 85
0.820–0.880 3500 42 85
o0.860 500 43 90
o0.870 2000 42 85
o0.845 50 45 95
the quality speciﬁcations of oil products and the energy use by reﬁneries (CO2 emissions) in a more objective manner. This means the quantiﬁcation of this trade off for a case study. The case study presented here is that for which more detailed data was available for our estimates: the Brazilian reﬁning industry. We do recognize that the Brazilian reﬁning industry is not subject to explicit greenhouse gases (GHG) emission reduction targets, as Brazil is not part of Annex 1 countries of the Kyoto Protocol . In addition, the country’s regulations have deﬁned rising diesel and gasoline quality requirements. Nevertheless, these requirements are still less restrictive than the targets set for the USA and Western Europe over the same period. But even though the purpose of the Brazilian reﬁning industry is not to formulate a ULSD diesel or a premium gasoline, and even though Brazil has no GHG emissions targets, the case of Brazil per se clearly exempliﬁes the trade-off between fuel speciﬁcation and energy use (CO2 emissions) by reﬁneries. For Brazil, as of today, this tradeoff is perhaps more easily perceived by the reﬁners as a balance between energy use and oil product speciﬁcations. As listed in Table 4, there are currently two diesel speciﬁcations in Brazil (metropolitan diesel and rural diesel), depending on the area of consumption (metropolitan areas with stricter speciﬁcation). However, by 2009, there will be only a single diesel speciﬁcation, with more stringent requirements especially for sulfur content. Thus, between 2002 and 2009, Brazil’s reﬁning industry should strive to lower the sulfur content of its diesel by around 3000 ppm, due to the proportion between rural and metropolitan diesel on the Brazilian market (at around 70% and 30% of the diesel market, respectively). Additionally, the Brazilian reﬁning industry should comply with a reduction in the gasoline sulfur content, currently speciﬁed at 1000 ppm and dropping to 80 ppm in 2009, according to the proposal presented by Brazilian reﬁners —i.e., a reduction of 920 ppm. In order to comply with these speciﬁcation targets and process increasingly heavier crude oils from the Brazilian oil production, with no loss of light and medium products yields,19 Petrobras allocated 43% of some US$ 6 19 For the type of crude oil produced and processed in Brazil, see Tavares et al.  and Szklo et al. .
Table 5 Expansion of hydrotreating (HDT) and hydrocracking (HCC) units in the Brazilian Reﬁning Segment—2004–2011 Expansion
1000 barrels per calendar-day
Hydrotreating 704.4 Diesel 317.6 Gasoline 320.8 Naphta, petcoke and other oil products 66.0 Hydrocracking Lube oils Diesel (fuels)
99.0 35.0 64.0
Source: Assayag  for HDT and Petrobras  for HCC.
billion through its Reﬁning Investment Plan between 2004 and 2008 to upgrading diesel and gasoline quality. The rest of these investments are channeled to conversion (30%); maintenance (11%); expansion (6%); and others. Basically, these investments are earmarked for the HDT and HCC units (Table 5) that will boost HDT capacity from 222,000 bpd (12.5% of Atmospheric Distillation) to 926,000 bpd (51.1% of Atmospheric Distillation), and HCC from zero to 99,000 bpd (5.5% of Atmospheric Distillation) between 2002 and 2009 [32, 33]. These units will operate in compliance with the typical conditions presented in Table 6. Particularly for diesel, compared to HDT, HCC will result in high-grade products (except for the naphtha octane number), with high homogeneity (Tables 7 and 8). However, although versatile, an HCC plant has high costs, particularly when processing Brazilian crude oils to produce lubricants. It is estimated that a plant with a capacity of 5500 m3/year of HDT for diesel costs some US$ 200 million, while an HCC unit with the same capacity producing medium distillates costs some US$ 500 million, and ﬁnally an HCC unit with the same capacity focused on lube oils costs US$ 650 million . Based on these data and the average speciﬁc consumption of the HDT and HCC units under consideration, a huge increase in energy use may be foreseen for the Brazilian reﬁning industry, owning to the introduction of HDT and HCC units and the production of H2 that they
ARTICLE IN PRESS A. Szklo, R. Schaeffer / Energy 32 (2007) 1075–1092 Table 6 Typical HDT and HCC operating conditions tested for Brazilian crudes
Purpose Rated capacity (m3/d) Operating pressure (bar) Operating temperature (oC) Catalyst volume (m3) Number of reactors Wall thickness (mm)
Fuel 5500 165 380 825 3 244
85 365 235 2 144
200 390 980 4 3444
Source: Petrobras .
Table 7 HCC units yields tested for Brazilian crudes Product
Naphtha (wt%)—yield Sulfur (ppm) MON/RON
23 o10 55/65
11 o10 55/65
Jet fuel (wt%)—yield Sulfur (ppm) Nitrogen (ppm)
26 o10 o1
18 o10 o1
Diesel (wt%)—yield Sulfur (ppm) Nitrogen (ppm) IC
46 o10 o1 55
27 o10 o1 53
4 o5 o1
43 o5 o1
Fuel oil residues (wt%)—yield Sulfur (ppm) Nitrogen (ppm) Source: Petrobras .
Table 8 Diesel HDT—crude oil feedstock: Cabiu´nas Load
Density at 20/4 1C T90 (1C) Total sulfur (% mass) Total nitrogen (mg/kg) Cetane number
50% vol: heavy straight-run diesel 29% light cycle oil (LCO) 21% light coke gas oil Load
0.8988 395 0.570 1760 36.1
0.8772 388 0.005 80 42.0
Source: Petrobras .
require (Table 9).20 This increase tops 30%, according to our estimates, which assumed:
A usage factor of 97% for the treatment and hydroconversion plants, according to the HDS, HDA and
20 The estimate does not include HCC for lube-oils, as this is not a plant producing fuels. For H2 production, the average speciﬁc consumption was used for a steam reform plant, which is the technology mainly adopted for the expansions planned by Petrobras .
HDT units already in operation in the country . The average distillation capacity usage factor of 90% was based on Tavares et al. . The following average speciﬁc consumption levels for the HDT and HCC units under analysis, which are given in MJ/b:21 Diesel HDT (100); Gasoline HDT (150); naphtha HDT (135); total conversion HCC (240). The hydrogen demands from the units listed above (in Nm3/b): Diesel HDT (28); Gasoline HDT (15); naphtha HDT (15); total conversion HCC (58). The energy use of a typical H2 production unit based on natural gas steam reform,22 at 57 GJ of fuel/t H2 (Rostrop-Nielsen ;23 EIPCCB ). The main energy source for the HDT and HCC units consist of process wastes and residual fuel oils, in addition to petcoke. The data from Brazilian reﬁneries such as REDUC justiﬁes this assumption (Schaeffer et al. ). According to MCT , the CO2 emission factor of the petcoke is 27.5 tC/TJ. For the steam reform to produce H2, by deﬁnition the energy source is natural gas (emission factor of 15.3 tC/TJ). For the total CO2 emissions derived from the production and use of energy in Brazil, the methodology of the Ministry of Science and Technology  was applied by Cima  to 2002.
As presented in Table 10, stricter diesel and gasoline quality speciﬁcations in 2009 increase CO2 emissions in Brazil by around 1.5 MtC. Overall, the units intended to meet the diesel and gasoline sulfur speciﬁcations boost energy use of Brazil’s reﬁning industry by around 30%. These units account for some 2% of the total CO2 emissions derived from the Brazilian energy system. This estimate should not lead to the conclusion that stricter oil product quality speciﬁcations in Brazil are not appropriate. The positive impacts of these speciﬁcations are signiﬁcant, for sulfur emissions as well as emissions of other pollutants such as CO and NOx, whose catalyst removal system is poisoned by sulfur. What is presented here is the trade-off between the quality speciﬁcations for the oil products and the increase in energy use (and CO2 emissions) by reﬁneries, in order to comply with these speciﬁcations. This trade-off has objective implications for the strategies of the countries covered by Annex I of the Kyoto Protocol but is also signiﬁcant for countries such as Brazil, China and India, in view of current discussions over the pertinence of establishing greenhouse gases reduction 21 This consumption is based on data from Brazilian plants compared to the values range for similar units installed in European reﬁneries [25, 36, 37] and in the USA [23, 28]. See also Dusse et al.  and Almeida et al. . 22 This has been the typical production process for producing ultra-pure H2 (99.999%) in Brazilian reﬁneries, although the solid wastes gasiﬁcation process may also be analyzed for a period beyond 2009 (Szklo & Schaeffer, 2005). 23 According to Rostrop-Nielsen , the thermodynamic yield of a typical H2 production unit is around 3.5 Gcal/1000 Nm3.
ARTICLE IN PRESS 1082
A. Szklo, R. Schaeffer / Energy 32 (2007) 1075–1092
Table 9 Estimated energy consumption by the Brazilian reﬁning segment after the introduction of new HDT and HCC units Total energy use—Brazilian Reﬁning Segment—2002 (1)—(TJ) Energy use by new HDT, HCC and H2 production plants (TJ) Increase in energy use due to more stringent fuel speciﬁcations (%)(2) Total energy use of HDT/HCC (TJ)(3) Proportion of HDT/HCC in total energy use (%)
228,141 70,438 31.0 82,050 27.5
Note: (1) Based on the speciﬁc consumption data for Brazilian reﬁneries . In 2002, it is estimated that energy use represented 6.6% of the feedstock crude-oil entering the Brazilian reﬁning park. (2) After the introduction of the HDT, HCC and H2 production plants, energy use accounted for 8.6% of the feedstock crude oil entering the Brazilian reﬁning park. The increase of 31% in ﬁnal energy consumption is due only to the above-mentioned plants. This analysis is carried out on a Ceteris paribus basis—i.e., possible energy conservation measures were not taken into consideration, which might be implemented by the Brazilian reﬁning industry, and no marked changes were assumed in the average feedstock of the reﬁneries (API, sulfur content, acidity, etc.). (3) Units existing in 2002 and installed after 2002, in order to comply with the new quality speciﬁcations. Consumption by all the existing and new HDT and HCC units and the H2 production plants accounts for 27.5% of the ﬁnal energy consumption of the Brazilian reﬁning segments. Interestingly, this is compatible with the ﬁgure presented by the US reﬁning industry .
Table 10 Estimated CO2 emissions by the Brazilian reﬁning segment after the introduction of the treatment plants
remarks of this paper highlight some of them, while this section focuses mainly on the technical alternatives.
79.2 Energy system emissions – 2002 (MtC)—A(1) Emissions by the HDT, HCC and H2 production plants (MtC)—B 1.8 Expansion to comply with speciﬁcations in 2009 (MtC) 1.5 Existing plants in 2002 (MtC) 0.3 (B/A) % 2.2
5.1. Improvement of heat integration and waste heat recovery
Notes: (1) Emissions from fossil fuel consumption in Brazil, according to Cima .
targets for non-Annex I countries, after the period stipulated in the Kyoto Protocol . In this case, the problem of carbon emissions may become signiﬁcant in Brazil, in the near future. In brief, technological alternatives should be sought for quality speciﬁcations with no considerable increase in energy use by reﬁneries. The next sections of this paper discuss some alternatives.
5. Energy savings potential in reﬁneries According to Petrick and Pelegrino , in the medium to long terms a target reduction of 15–20% in energy use (and consequently in CO2 emissions) from US reﬁning sector is achievable. Waste heat recovery seems to be one of the most important options in the short to medium terms, while fouling mitigation, and new reﬁning processes are promising technologies in the medium to long terms.24 As this section will discuss, a similar energy savings potential can be achieved in Brazilian reﬁneries, offsetting a least 50% of the increase in energy use estimated in the previous section. Clearly, there remain barriers to invest in the set of energy savings options identiﬁed in this section. The ﬁnal
In the short to medium terms, the improvement of heat integration and waste heat recovery appears as one of the main options for saving fuel in Brazilian reﬁneries. There is no substantial R&D development related to this option, and chemical plants in Brazil and worldwide are already aware about the techniques that might be applied for implementing this option. Moreover, chemical plants  and oil reﬁneries  in Brazil have experiences in optimizing heat networks for saving fuels. Especially for oil reﬁneries, large temperature differences indicate the possibility of managing heat and cold ﬂows to reduce energy use. For instance, heat integration of process systems can ensure that a substantial proportion of the heat required will be provided by exchanging heat between streams to be heated and streams to be cooled, and not by external sources for heating and cooling. A positive sideeffect of that is the simultaneous reduction of waste water (used for quench) and make-up water (used for steam generation). In brief, to improve heat management and waste heat recovery in reﬁneries includes:
use of waste heat in absorption refrigeration systems ; use of waste heat to pre-heat feeds (e.g., through the installation of waste heat boilers or heat recovery steam generators);25 heat and/or mass (water and hydrogen) integration using basically Pinch techniques [47, 48]; improvement of furnaces efﬁciencies combined with computer controlled combustion ; direct feed of ‘‘intermediary products’’ to processes without cooling and storage, aiming at recovering the
API  indicates, however, fouling mitigation as an alternative to the near-to-mid term. Brazilian experiences on this issue lead to the same conclusion .
25 For instance, waste heat boilers can recuperate some of the heat produced during the coking process.
ARTICLE IN PRESS A. Szklo, R. Schaeffer / Energy 32 (2007) 1075–1092
waste heat of these hot products. For instance, the waste heat of the products of the crude distillation unit can be recovered by later feeding them directly to the downstream units, rather then cooling them for storage and later feeding the downstream units from tankage ; use of heat pumps ; decreased ﬁlm temperature and increased turbulence on heat transfer surfaces; insulation of buildings and process units ; adoption of steam management . Steam used for stripping,26 vacuum generation, atomization and tracing is usually lost to waste water and to the atmosphere. In order to reduce the sour water loads to strippers and, as such, reduce chemical treatments in the overhead system, and option could be to strip sidedraw products, particularly the lighter cuts, using reboiled sidestrippers instead of steamed strippers .
Overall, through the use of Pinch techniques,27 reductions in energy consumption around 20% have been reported by Petrick and Pellegrino . This ﬁgure agrees with estimates from EIPCCB . However, according to Hallalle  and CTEC , typical savings resulting from energy pinch hover between 10% and 25% in oil reﬁneries (as a percentage of total fuel consumption). Finally, Alsema  estimates that 2% savings on fuel can be realized by improved waste heat recovery; and, in the case of process integration and energy pinch, the fuel saving potential ranges between 2% and 6%. Beer  agrees with these numbers, estimating an average fuel saving of 5% in Dutch reﬁneries, at low costs (less than 10$/GJ). In turn, two studies performed at the Brazilian reﬁnery Replan28 [53,46], and one study performed at the Brazilian reﬁnery Reduc29 , analyzed the potential application of water and energy pinch in these plants. According to them, making better use of residual heat in the process streams is a major option for saving fuels in Brazilian reﬁneries.
26 Normally, stripping steam is used to meet ﬂash point speciﬁcation and to improve front-end fractionation and yield distribution. 27 Energy optimization, through heat exchanger networks using the Pinch Analysis, seeks integration between cold streams (that need warming) and hot streams (that need cooling), considering the temperature and enthalpy of each stream. As mentioned before, simultaneously, by reducing external heating and cooling demands, Pinch Energy Analysis does not result only in fuel savings but also water savings (used as steam in industrial processes or as work ﬂuid in cooling towers). For more details about this technique, see Hallale  and Linnhoff [50,51]. 28 Replan is the largest Brazilian reﬁnery, in terms of topping units capacity (352 million barrels per calendar-day). It is located in the state of Sao Paulo, and it focuses mainly on fuels, particularly on diesel production (around 40–45% of its historical oil products yield). It also produces signiﬁcant volumes of naphtha for petrochemicals production. 29 Reduc is the Brazilian reﬁnery with the largest Nelson Index of complexity, being focused on fuels (mainly, diesel, jet fuel, premium gasoline, and LPG), petrochemicals and lube oils (80% of total lube oils production in Brazil). Its primary capacity is 239 million barrels per calendar-day. A detailed analysis of Brazilian oil reﬁneries can be found in Szklo .
However, the analysis of the Brazilian reﬁnery Replan shows that not all hot streams may be used by an energy integration network. Initially, volatile products that must be cooled rapidly through direct contact with water (quench) cannot be used. Neither can intermittent streams . Processes with high speciﬁcity levels (such as HDT) should also be discarded, as well as hot streams containing solids in suspension (such as catalyst streams). Finally, high exergy streams (such as FCC outlet gases) are hard to recycle, as they are found in inaccessible parts of the reﬁnery . Therefore, the high-energy use of the atmospheric distillation column and the fact that this column processes large volumes of feedstock make it the ﬁrst option for heat integration in oil reﬁneries. For instance, at the distillation column of the Brazilian reﬁnery Replan, the variation of temperature is from 124 to 35 1C for diesel (mass ﬂow of 80.5 kg/s), from 165 to 35 1C for jet fuel (mass ﬂow of 16.6 kg/ s), and from 304 to 35 1C for light gasoil (mass ﬂow of 8.3 kg/ s).30 The ﬁrst two oil products leave the atmospheric column to be hydrotreated, while the latter one follows to a FCC unit. Therefore, to optimize heat recovery from Replan’s reﬁnery atmospheric distillation column, two or three reﬂux streams could be kept in continuous circulation at several points. In addition, besides the introduction of optimized heat networks in Brazilian reﬁneries, the use of waste heat to provide cooling is also an attractive option. Again, the analysis of Replan is elucidative. Given the variations of temperature of the streams leaving its distillation column, the regenerator of the absorption refrigeration system (single-stage absorption system) can be easily kept at 100 1C, with the coefﬁcient of performance (COP) of 0.80. In this case, the cold stream produced can be used as the last stage of the vacuum production system of the vacuum distillation column. As such, it reduces the temperature of the water used by this system, lowering also the pressure of the column, and, as such, increasing its performance.31 In sum, considering mainly heat integration and waste heat recovery for Brazilian reﬁneries and according to the options described before and tested in two major Brazilian reﬁneries, this study estimates a fuel savings potential of around 10% (percentage of total fuel consumption). 5.2. Fouling mitigation The deﬁnition of the Pinch Point32 is highly affected by the control of fouling. In heat networks where fouling readily occurs, this may reach 40 1C , when the typical 30 These variations of temperature are generated by the heat rejected to the water refrigeration system, after the distillation column. 31 Petrick and Pellegrino  describe the application of an absorption refrigeration unit, using waste heat, to recover additional LPG from a reformer reactor. The implementation of this alternative at a reﬁnery in Denver, Colorado, had a payback period of 1.5 years . Petrick and Pellegrino  also report on using a similar technology for the CDU, estimating energy savings ranging from 10% to 20%. 32 The Pinch Point is determined as the minimum temperature difference that is accepted by the heat exchangers in the heat network.
ARTICLE IN PRESS 1084
A. Szklo, R. Schaeffer / Energy 32 (2007) 1075–1092
ﬁgures range between 10 and 20 1C. Fouling reduces thermal efﬁciency and heat transfer capacity, resulting in increases in energy use. It is difﬁcult to prevent, as the mechanisms, which lead to fouling, are not yet well understood . Therefore, an important measure for lowering energy use by reﬁneries is the control of fouling in heat exchangers, which, in addition to lessening the heat exchange area, also cause maintenance halts and accident risks at industrial plants.33 In addition, the desalter unit plays an important role in reducing energy use by the reﬁnery, through removing salts and contaminants from its feedstock. Interestingly enough, this may be a win–win situation, because better heat exchanges have positive effects on the desalter efﬁciency, as this efﬁciency is associated with operations within a restricted optimum temperature range. Heat exchangers with losses of heat exchange areas due to fouling cannot guarantee that this optimum range will be reached, lessening the salts and metals removal capacity of the desalter . This results in oil losses that pollute the liquid wastes discharged by the plant and relatively high contaminant levels in the feedstock. It was estimated that a typical reﬁnery, ca. 1980, with a primary processing capacity of 100 Mbpd, could decrease the energy use by its Atmospheric Distillation unit by 30% by controlling heat exchanger fouling . However, a more recent study indicated a lower energy potential through fouling control, which remains signiﬁcant at 10% . According to Bailey , the oil reﬁning industry in the US alone spends US$ 2 billion a year on crusting problems. However, the steadily increasing diversity of the crude oils processed in the world reﬁning industry hampers the development of fouling prevention techniques. Heat stability studies are needed, as well as analyses of the solubility of asphaltenes and naphthenic acids, in addition to the development of anti-fouling chemical compounds that also remove scale without adversely affecting the quality of the reﬁnery products. The challenges are explained by the fact that fouling is the effect of several process variables and heat exchanger design. Fouling may follow the combination of different mechanisms . In sum, fouling of heat exchangers is a bottleneck in the application of heat recovery schemes. Fouling prediction and mitigation can therefore improve energy efﬁciency. The fuel savings of 2% of total fuel use obtained by Petrick and Pellegrino  to US reﬁneries agree with the results of studies performed in Brazilian reﬁneries . However, the higher ﬁgure obtained by Panchal and Huangfu  indicates the need for new studies. These authors analyzed fouling effects of a 100,000 bbl/day crude distillation unit and found an additional heating load of 13.0 MJ/barrel
processes (or around 3.4% of the average speciﬁc energy consumption of Brazilian reﬁneries). 5.3. Advanced process control Advanced process control based on computer models and extensive use of sensors might result in improved production reliability and thus increased product yield. Indeed, modern control systems are often not solely designed for energy efﬁciency, but rather to improve productivity, product quality and efﬁciency of a production line. Control systems result in reduced down time, reduced maintenance costs, reduced processing time, and increased resource and energy efﬁciency, as well as improved emissions control . Large potentials remain to implement control systems . For instance, Timmons et al.  combined online optimizer with existing control systems to improve the operation of the FCC unit at the CITGO reﬁnery in Corpus Christi, Texas. They reported savings of $0.05/barrel. According to Alsema , energy savings can be estimated at 2–4% of fuel consumption. However, Worrell and Galitsky  indicate a savings potential varying from 2% to 18% for US reﬁneries, using moisture, oxygen, air ﬂow and temperature controls based on fuzzy logic or rulebased systems.34 There is no estimate or study performed for Brazilian reﬁneries, precisely, on this subject. Therefore, we suggest as a conservative ﬁgure the lower value obtained by Alsema  to Dutch reﬁneries (2% of fuel savings). In addition, no cost estimates are available for Brazilian reﬁneries as well, but ﬁxed costs are probable relatively high because of the requirement to install several sensors and to customize the control software to each speciﬁc unit or plant. Actually, process control systems depend on information of many stages of the processes. A separate but related and important area is the development of sensors that are inexpensive to install, reliable, and analyze in real-time. Development aims at the use systems, which should be resistant to aggressive environments (e.g. oxidizing environments in furnace or chemicals in chemical processes) and withstand high temperatures. Nevertheless, it is expected that advanced control systems will be progressively implemented in Brazilian reﬁneries thanks to their various beneﬁts, not only related to energy efﬁciency. According to Katzer et al. , the reﬁnery of the future will look more like an automated chemical plant. 5.4. Replacement of the conventional atmospheric and vacuum distillation units As mentioned in the introduction, the atmospheric and vacuum distillation columns are large consumers of heat. According to Worrell and Galitsky , atmospheric and
An interesting analysis of the effects of fouling on the performance of heat exchangers in Brazilian reﬁneries was carried out by Negrao et al. .
34 An example of the use of fuzzy control in process units can be found in Aprea .
ARTICLE IN PRESS A. Szklo, R. Schaeffer / Energy 32 (2007) 1075–1092
vacuum distillation units accounted for 31% of the total energy use in US reﬁneries in 2001. For the Brazilian reﬁneries in 2002, this share can be estimated at 46%, considering the average speciﬁc energy consumption of distillation units in the country, and the crude oil processed by each reﬁnery. The higher share found in Brazilian reﬁneries in 2002 can be explained by the relatively lower conversion and treatment capabilities of these reﬁneries when compared to US plants.35 Therefore, the large dependence of all reﬁneries on physical and chemical separation processes, presenting low thermodynamic yields, underscores the need for R&D of alternative separation technologies. Actually, leap-frog technology is needed to reduce the large amount of energy used in distillation through the reﬁnery complex . As described before, in distillation, the main short-term developments are in improved integration using heat recovery technology and integration of different distillation units (i.e. CDU and VDU). In the long term, the major developments are the integration of different distillation columns into one reactor (e.g. dividing-wall column) or the development of alternative processing routes allowing for combination of conversion and distillation (e.g. reactive distillation) . Replacing distillations units by cracking ones is also seen as a promising option .
Thermal Cracking Process: Distillation is used to separate crude oil into its various fractions based on differences in boiling points. An alternative to primary separation could be the use of controlled thermal cracking, which separate crude oil into fractions by cracking large hydrocarbon molecules into smaller ones, thus lowering their boiling points. In addition, a controlled operation (low-residence time unit) could achieve primary separation while also removing major fractions of contaminants, particularly sulfur. Therefore, this latter effect indirectly reduces the energy consumption of hydrotreating ﬁnished and non-ﬁnished oil products. Petrick and Pellegrino  estimated a potential net energy savings of 65 MJ per barrel of oil processed. Brazilian oil reﬁneries consumed 228,141 TJ in 2002 or presented an speciﬁc energy consumption of 383 MJ per barrel of oil processed. Therefore, it is possible to calculate a net energy savings potential of 17% for the Brazilian reﬁning sector, if all crude and vacuum distillation towers were replaced by thermal cracking.36 However, that full replacement of distillation towers in existing plants is not likely, and it is a high-cost option. Reﬁners worldwide (and in Brazil) tend to be very cautious in investing in this kind of drastic technological change, as the ﬁnal remarks of this paper will emphasize.
35 The total installed capacity in conversion units at Brazilian reﬁneries in 2003 was around 40% of the total primary capacity (this proportion totaled 89.3% in US reﬁneries, according to Nakamura ). For hydrotreating units this proportion was only 12.5% , while in US reﬁneries it was 67.7%, and 88.8% in Californian reﬁneries . 36 Interestingly, the estimate of Alsema  for Dutch reﬁneries was 18%.
Progressive distillation unit: A progressive distillation unit integrates the atmospheric and vacuum distillation columns, saving up to 30% on total energy use for these units . This technology includes atmospheric distillation (topping), vacuum distillation, gasoline fractionation, naphtha stabilizer if required and gas plant . It is the extreme of heat integration between atmospheric and vacuum distillation, avoiding also superheating of light cuts to temperatures higher than strictly necessary for their separation. This option is merely applicable to distillation units to be constructed. Therefore, it is not a suitable option for existing reﬁneries in Brazil. Dividing-wall distillation: The ﬁrst commercial application of the dividing-wall distillation column dates back to the early 1990s (Hallale ). A dividing-wall column integrates two conventional distillation columns into one, increasing heat transfer. Dividing-wall columns (DWC) can save up to 30% in energy costs, while providing lower capital costs compared to conventional columns . Various companies (Kellog Brown & Root, and UOP) have developed DWC-concepts for the separation of products. However, further development of DWC for the major distillation processes in the petroleum reﬁning industry is still needed [3,49]. Reactive distillation: By combining the chemical reaction and separation in one reactor, capital costs are reduced and energy efﬁciency is improved through better integration of these process steps . Various research institutes and technology developers aim at developing new applications of reactive distillation. A new development includes the use of monolithic structures that contain the catalysts , reducing catalyst loss and the pressure drop. The most promising commercial application is not applicable for replacing distillation units, however. It is the catalytic distillation for replacing severe hydrotreating. This technology will be analyzed in the next section.
5.5. Membrane separation technology Membranes are an attractive technology for hydrogen recovery at reﬁneries .37 Membranes have been demonstrated for recovery of hydrogen from hydrocracker offgases. Various suppliers offer membrane technologies for hydrogen recovery in the reﬁning industry, including Air Liquide, Air Products and UOP. The hydrogen content has to be at least 25% for economic recovery of the hydrogen, with a recovery yield of 85–95% and a purity of 95% [3,49]. However, membrane technology represents the lowest cost option merely for low product rates, but not for high ﬂow rates. Thus, development of low-cost and efﬁcient membranes is still needed, in order to improve 37
Hydrogen recovery is an important technology development area to improve the efﬁciency of hydrogen recovery, reduce the costs of hydrogen recovery and increase the purity of the resulting hydrogen ﬂow.
ARTICLE IN PRESS 1086
A. Szklo, R. Schaeffer / Energy 32 (2007) 1075–1092
Table 11 Energy saving potential in Brazilian reﬁneries Energy saving option
Estimated fuel saving potential (percentage of total fuel consumption)
Heat integration and waste heat recovery(1) Fouling mitigation Advanced process control Replacement of topping units by Thermal cracking Progressive distillation Dividing-wall distillation Pumps and advanced motors with variable speed Vacuum pumps and surface condensers
10% 2% 2%
Fully applicable Fully applicable Fully applicable Only for new reﬁneries
17% 15% 15% 1%(2) —
Application restricted by steam recovery and heat integration Non commercial for high ﬂow rates
Note: (1) These options include: use of waste heat in absorption refrigeration systems; use of waste heat to pre-heat feeds; heat and/or mass (water and hydrogen) integration using Pinch techniques; improvement of furnaces efﬁciencies combined with computer controlled combustion; direct feed of ‘‘intermediary products’’ to processes without cooling and storage; use of heat pumps; decreased ﬁlm temperature and increased turbulence on heat transfer surfaces; insulation of buildings and process units; and adoption of steam management. (2) Percentage of the electricity consumption.
the cost-effectiveness of hydrogen recovery, and enable the recovery of hydrogen from gas streams with lower concentrations. As of today, a broader application of membrane technology to separate products in reﬁneries does not seem very promising, and actual savings are unclear. Petrick and Pellegrino  do not even mention this option. According to Worrell and Galitsky , further research is needed to develop appropriate membrane materials that can withstand the environment found in petroleum reﬁning processes. In addition, membrane technology should be evaluated as an integrated part of the speciﬁc process for which it is being implemented to warrant the full energy savings potential. For this reason, this study did not considered here this option for estimating the energy savings potential in Brazilian reﬁneries. 5.6. Use of vacuum pumps and surface condensers Vacuum pumps and surface condensers can largely replace barometric condensers in many reﬁneries. Replacing the steam ejector by vacuum pump reduces the sour water ﬂow and increase energy efﬁciency . Precisely, the replacement of the steam ejectors by vacuum pumps increases the electricity consumption, while reduces the heat consumption, the cooling water consumption, and the consumption of agents used for conditioning cooling water. The net energy saving is positive. All Brazilian reﬁneries are able to adopt this option, but since there are many processes in it where surplus steam is recovered and used for the production of vacuum, a suitable energy management might undermine the usefulness of this option. Improving energy integration and steam recovery and management in Brazilian reﬁneries tend to restrict the application of this option.
5.7. Use of pumps with variable speed Pumps at reﬁneries are usually operated at variable conditions . Normally, the ﬂow control is based on valves, resulting in energy losses. Therefore, an option studied by Olim et al  for the Brazilian reﬁnery Replan was the use of frequency inverters for electric motors drive. For instance, the atmospheric distillation column of Replan has ﬁve pumps operating at ﬁxed speeds, with capacities ranging from 125 hp to 200 hp. The analysis indicated an electricity savings potential of 2.6 GWh per year, with an attractive internal rate of return (between 20 and 55% per year, depending on the size of the pump). Therefore, summing up the energy savings alternatives that can be implemented together in Brazilian reﬁneries in the near to medium terms, a net energy savings potential hovering between 10% and 20% can be estimated (Table 11).38 Clearly, these energy savings options can be applied to oil reﬁneries, whether or not alternative treatment processes are implemented. This study considered the possibility of adopting, simultaneously, energy savings options and alternative treatment units. Therefore, the net beneﬁt of these two measures depends on the analysis undertaken in the next section. 6. Alternative treatment processes As discussed before, hydrotreating processes have been gaining ground in reﬁning parks supplying markets with more stringent fuel speciﬁcations [12,13]. As noted, sulfur is a key contaminant for oil products speciﬁcations. The sulfur compounds to be removed during hydrotreatment, 38
The near to-medium term is in accordance with the time horizon of the estimates performed in Section 4 (expansion of HDT units in Brazil and impacts on energy use and CO2 emissions).
ARTICLE IN PRESS A. Szklo, R. Schaeffer / Energy 32 (2007) 1075–1092
more speciﬁcally during hydrodesulfurization, include mercaptans, sulﬁdes, thiophenes and benzothiophenes (BTs). Stricter speciﬁcations on the sulfur content of fuels are forcing reﬁneries to invest in severe hydrodesulfurization units or seek alternative desulfurization processes. But, as stressed before, deep hydrodesulfurization presents two major problems when used in gasoline streams: 1. It affects the gasoline quality (octane rating), by lowering its oleﬁns content. This fact refers mainly to gasoline from FCC, which accounts for 85–95% of the sulfur content of the product,39 and simultaneously contains a great quantity of oleﬁns (20–40 wt%), which provides it with a fairly good octane number .40 In other words, to remove sulfur from gasoline means to remove it mainly from FCC gasoline, aiming at not compromising the quality of the ﬁnished gasoline. 2. As detailed in the previous sections, it increases the energy use of the reﬁnery, resulting in higher operating costs and also more severe environmental impacts caused by reﬁning activities (for example, higher emissions of GHGs such as CO2, tightening up the trade-off between local and global pollution referred to before in this paper). For instance, the Brazilian case study indicated that, overall, the units intended speciﬁcally to meet the diesel and gasoline speciﬁcations, particularly for sulfur removal, boost the energy use of the reﬁning industry by around 30%, accounting for 2% of the total CO2 emissions derived from the Brazilian energy system. Alternative treatment processes are designed speciﬁcally to mitigate these problems. Organic sulfur compounds are present in almost all oil cuts leaving the distillation tower (straight-run streams). Cuts with higher boiling points (or higher cut-off temperatures) contain relatively higher sulfur levels and their sulfur compounds have heavier molecular weights. There are also differences in the sulfur compound reactivity, affecting the efﬁciency and efﬁcacy of their removal during the hydrotreatment process. Lighter fractions from Atmospheric Distillation normally contain aliphatic compounds (such as mercaptans) that are highly reactive and can be withdrawn easily through conventional HDS and even other processes such as Merox Treatment.41 For heavier fractions, such as heavy naphtha and diesel leaving the distillation column, FCC naphtha, delayed coking naphtha and coking diesel, the predominant sulfur 39 Usually, FCC gasoline constitutes 40–60% of the gasoline pool [69,13]. 40 In Brazil, light naphtha straight-run has the average octane rating of 60–65, while FCC gasoline presents the average ranging from 80 to 90 . 41 Mercaptans are slightly acidic organic sulfur compounds. They can be withdrawn from light ends mixtures by caustic washing in a Merox extraction system. This system uses organometallic catalysts to accelerate the oxidation of mercaptans to organic disulﬁdes at or near ambient temperature and pressure, in an alkaline environment .
compounds are less reactive: thiophenes, benzothiophenes, dibenzothiophenes, and other polyaromatics compounds with sulfur, particularly alkylbenzothiophenes.42 Deep desulfurization of products at the reﬁnery should remove these less-reactive compounds. One possibility is to step up the severity of the HDS. However, this measure leads to undesirable reactions (e.g. oleﬁn saturation of FCC gasoline). Additionally, the higher temperatures required for severe HDS also step up coke formation and the de-activation of the hydrotreatment catalyst—i.e., they shorten the useful life of the catalyst, which increases the operating costs of the reﬁnery. Finally, as discussed previously, severe HDS boosts energy use at the reﬁnery. Thus, in response to the problems caused by severe HDS, alternative treatment processes are being proposed. One focus of research is the development of more selective catalysts, which lower the probability of the occurrence of undesired parallel reactions during HDS [11,12,70]. Another focus is the development of advanced reactors that also include special supports for the catalysts . An interesting alternative is the combination of the treatment process with other processes in order to ensure desulfurization and the production of high-grade fuels. 6.1. ISAL process ISAL43 process combines conventional deep HDS (using molybdenum and cobalt oxide catalysts on alumina44) with isomerization reactions of parafﬁns, to subsequent octane recovery. One of the drawbacks of these processes can be some yield loss in gasoline due to cracking into light products. Another disadvantage with respect to selective processes is the hydrogen consumption needed for the quasi-total oleﬁn saturation . Nevertheless, ISAL is not per se an option for saving energy and reducing CO2 emissions. It still depends on deep HDS and presents high H2 requirements. Thus, its main advantage is the fact that octane loss during HDS is lowered, and, as such, the quality of the gasoline pool is kept unchanged without the need for adding additives to gasoline or changing the pool (e.g., adding alkylates and isomerates). 6.2. Olefin alkylation of thiophenic sulfur (OATS) process The Oleﬁn Alkylation of Thiophenic Sulfur (OATS) process steps up the boiling point of the sulfur compounds in the gasoline through acid-catalyzed alkylation reactions . This means that the less-reactive sulfur compounds become heavier and are concentrated in the reﬁnery bottom streams. For instance, the alkylation reaction 42 The reactivity of the sulfur compounds in the heavy fractions at the HDS units normally complies with the following order (from more to less reactive): thiophene, alkylated thiophene, benzothiophene, dibenzothiophene and alkylated dibenzothiophene ). 43 The name of the technology comes from ‘isomerization’ and ‘Salazar’ - name of the technology inventor . 44 CoMo–P/Al2O3 associated to a Ga–Cr/HZSM-5 zeolite .
ARTICLE IN PRESS 1088
A. Szklo, R. Schaeffer / Energy 32 (2007) 1075–1092
between thiophene and oleﬁns (usually hexane) steps up the thiophene boiling point from around 86 1C to some 250 1C , which allows it to be separated from the gasoline through simple distillation. This reaction is very simple and can take place under rather mild conditions . It also hardly inﬂuences the octane number of gasoline . Finally, the HDS process is no longer necessary, reducing hydrogen and energy use by the reﬁnery. The heavy sulfur compound removed through simple distillation may be added to the diesel or heavy gasoil pool at the reﬁnery. However, this technique was invented by BP in 1999 , and still must be tested. BP had built a small-scale test set up , but there is no data about energy use in a future commercial unit. In addition, besides thiophene, in gasoline there were many thiophene derivatives such as methylthiophene, dimethylthiophene, ethylthiophene, and so on. Since these derivatives already have one or two side chains, their alkylation reactions and the reaction extent could be changed. These questions need to be answered by further studies. Zekai et al.  also found that in OATS there are two parallel alkylation reactions: the aromatics alkylation and the alkene alkylation (oligomerization). Both compete with the thiophene alkylation. Apparently, their mechanism is the same as that of the thiophene alkylation, especially in the case of aromatics alkylation. Moreover, although its reaction thermodynamics is slightly less favorable than that of aromatics alkylation, the highly concentrated alkenes in FCC gasoline facilitated the alkene oligomerization. Once again, these questions need to be answered by further studies. 6.3. The oxidative desulfurization process (ODP) process The oxidative desulfurization (ODP) process is applied under mild conditions (room temperature and atmospheric pressure) . As the OATS process, it is a non-consuming hydrogen technique, which is based on the oxidation of organic sulfur compounds, followed by the extraction of reaction products (through simple distillation, extraction by solvent, adsorption, etc.). This extraction is possible because when the organic sulfur compounds are oxidized, they form sulfonic compounds (more polar), whose physical and chemical properties differ considerably from those of the hydrocarbons.45 Oxidants such as hydrogen peroxide and formic acid are being tested . This process is still under development, with good prospects for diesel , but not so promising for gasoline, due to competitive oleﬁn epoxidation reactions . In addition, some studies recommend the use of the oxidation extraction technique as an additional process to the HDS to enable the reﬁneries to meet the future environmental sulfur regulations : conventional HDS is used to lower the sulfur content to 45
Thus, they may be withdrawn from the diesel by solvent extraction using water-soluble polar solvents, such as NMP, DMF, DMSO and MeOH .
few hundreds parts per million. Then, the oxidation/ extraction approach is used to go for ultra deep diesel desulfurization. There is no data for energy use in commercial units. Actually, this alternative technique is still under the development phase, aiming, for example, at improving the catalyst system . However, the combined approach of less-severe HDS and ODP, to diesel desulfurization, would imply in energy savings of 40%, solely considering the replacement of the severe HDS units assessed for Brazil in Section 4. Clearly, this is a maximum savings potential only for diesel treating, as it does not consider the energy use in the ODP process unit. Nevertheless, considering that ODP combined to less-severe HDS is close to commerciality in the mid term , 40% was used as a ﬁrst proxy for estimating the energy savings potential (and CO2 emissions reduction, similarly) for treating diesel in Brazil. As such, when replacing the conventional diesel HDS, the ODP process avoids 23.3% of the total expanded energy use forecasted in Brazil.46 6.4. Catalytic distillation (CD) process The catalytic distillation (CD) process avoids a drop in the FCC gasoline octane number by adapting the severity of the HDS process to each component in the gasoline. In general terms, the gasoline consists of a ‘‘basket’’ of light and medium hydrocarbons. This ‘‘basket’’ may be fractioned by distillation before desulfurization, and each fraction may be treated in compliance with its prevailing sulfur compound reactivity. Therefore, CD makes it possible to treat separately various fractions of FCC gasoline under the most appropriate conditions for each of them in a single operation. Lighter fractions with more reactive compounds are treated under less severe conditions.47 This reduces the saturation reactions of the oleﬁns, which are light components of the gasoline. Simultaneously, the heavier fractions of FCC gasoline containing more refractory sulfur compounds undergo desulfurization at higher temperatures at the bottom of the CD column . Catalyst distillation takes place in a single piece of equipment where the FCC gasoline is fractioned and these fractions are desulfurized. Thus, CD combines separation of FCC gasoline through distillation and catalytic HDS in a same process using in its simplest ﬁtting up a single reactor or vessel .48 46 Diesel HDS accounted for 39% of the total expanded energy use estimated in Section 4. 47 The severity of the desulfurization conditions is controlled by the boiling temperature range of the fraction itself. 48 A variation of this process, which improves its efﬁciency, is based on the use of two columns instead of one. In this design, two distillation columns (CDHydro for the light fractions and CDHDS for the heavy fractions) are packed with desulfurization catalysts . In this variation, the ﬁrst column (CDHydro process) treats light catalytic naphtha to remove mercaptans and dioleﬁns while increasing octane rating. It
ARTICLE IN PRESS A. Szklo, R. Schaeffer / Energy 32 (2007) 1075–1092
This process was developed by CDTech and has been applied in some reﬁneries worldwide, such as Irving St. John in Canada, Motiva Port Arthur in Texas (US) and Pembroke Chevron-Texaco in Wales (UK).49 Therefore, differently from the other alternative desulfurization techniques, which were analyzed before in this paper, CD is already a commercial option.50 Moreover, data on efﬁciency of sulfur removal and on energy use are available. For instance, EIPCCB  reported that a reduction of 95% of sulfur content in FCC gasoline containing 1800 ppm has been achieved in units installed in European reﬁneries. A similar performance was reported by Peninger et al.  in Port Arthur reﬁnery in Texas, while Song  suggests an average performance of 92%. At Chevron Texaco reﬁnery in Pembroke Wales the distillation catalytic unit has been successfully applied, since early 2002, to reduce the sulfur content of FCC naphthas from 2800 to 50 ppm . The experience also showed that the catalyst cycle at the CD unit is long (5 years of guaranteed catalyst life), meaning that it reduces the shutdowns during the FCC cycle. CD eliminates catalyst fouling because fractionation removes heavy coke precursors from the catalyst zone before coke can form and foul the catalyst bed . Finally, the commercial experience with CD shows that octane loss was less than 1% [74,76], and gasoline yield loss was null . In terms of energy use, CD also uses less energy, as it is severe only for the heavier fractions of the gasoline. For a unit with a processing capacity of 30,000 barrels per day, and costing around 20 million dollars,51 Hagiwara  indicates the hydrogen consumption of 18 m3/m3, the power consumption of 3 kWh/m3, the steam use of 70 kg/ m3 and the fuel use of 5.3 kg/m3. Compared to the severe gasoline HDS units to be installed in Brazil in the near-to-mid term (see Section 4), CD reduces the hydrogen consumption by 81% and saves 52% of the energy use in HDS. As such, replacing HDS by catalytic distillation in Brazil can save 62% of the energy use for treating gasoline, including the energy use in the HDS process and the energy use in the hydrogen production. Clearly, the reduction in CO2 emissions is also 62%. Moreover, since gasoline HDS accounts for 37% of the expanded energy use in Brazilian reﬁneries due to hydrotreating and hydrocracking, CD can avoid 22.8% of this increase. (footnote continued) eliminates the need for separate caustic treating and selective hydrogenation units. Coupled with the CDHDS process, it enables reﬁners to produce ultra low sulfur gasoline at high reliability . 49 For further details, see Peninger et al. , Gardner et al.  and Reedy . 50 As of today, there are seven CDHDS commercial units in operation: Motiva (Port Arthur), Irving Oil (Saint John), Chevron Texaco Pembroke (UK), Motiva (Convent), Sopus (Puget Sound) and Valero. 51 These ﬁgures agree with those proposed by Song , who also provided operational costs of US$ 0.03 per gallon (including utilities, catalysts, and hydrogen).
Finally, a promising alternative that is still in the research and development stages is biodesulfurization. The commerciality of the biodesulfurization process would be a complete breakthrough in process development. It would offer mild processing conditions and reduce the need for hydrogen makeup. Both would lead to high energysavings (or CO2 emissions reductions) in the reﬁnery. It has been estimated that biodesulfurization can decrease CO2 emissions (or energy use) in reﬁneries by 70–80% compared to conventional hydrodesulfurization .52 Moreover, the speciﬁcity of the biochemical reactions is greater than that of a conventional HDS, particularly for less-reactive compounds such as dibenzothiophene [78,79,80]. For instance, aerobically grown strains, such as Rhodococcus erythropolis and related species (IGTS8), remove the sulfur from compounds such as dibenzothiophene (DBT) without degrading the carbon skeleton of this molecule . In addition, DOE  anticipates that biodesulfurization units will achieve, in the mid-to long term, 50% lower capital costs and 15–25% lower operating costs than conventional HDS. Biodesulfurization offers potential cost savings, not only because the process operates at ambient temperature and pressure, but also because it produces a non-toxic by-product, eliminating the need for collateral processing of hydrogen sulﬁde . However, research is still needed to study the biological mechanisms of the biocatalysts,53 including methods to control their activity and selectiveness, and to reduce their cost . Biocatalysts for desulfurization are usually obtained by culturing speciﬁc species of bacteria in media with dibenzothiophene as the sole sulfur source . As stressed by Cui-Qing et al. , at present, there is not yet any economically suitable method for large-scale preparation of biocatalysts due to the high cost of dibenzothiophene. In addition, the water needs of microbial cells require the creation of a two-phase biodesulfurization system with high interfacial areas through energy-intensive mixing and/ or addition of a surfactant, with a post-desulfurization deemulsiﬁcation step . Therefore, designing a costeffective two-phase bioreactor system coupled to oil-water separation and product recovery is also a key challenge to the viability of biodesulfurization processes. A possibility is to design a multi-staged air-lift reactor to achieve continuous growth and regeneration of the biocatalyst in the same system rather than in a separate reactor . 52 Alsema  estimates a similar potential, presenting an average energy savings of 300 MJ/b due to biodesulfurization. For the Brazilian reﬁning sector, this savings represents 78% of its speciﬁc energy use (estimated at 383 MJ per barrel of feedstock—see Sections 4 and 5). Clearly, a similar result is achieved for the CO2 emissions reduction potential. 53 The preferable pathway is the one in which dibenzothiphene is oxidized to DBTO. DBTO is then transformed to DBT sulfone (DBTO2) and to sulﬁnate (HPBS), followed by hydrolytic cleavage and subsequent release of sulﬁte or sulfate . As such, the sulfur is selectively withdrawn from the oil product without lowering its heat value.
ARTICLE IN PRESS 1090
A. Szklo, R. Schaeffer / Energy 32 (2007) 1075–1092
Then, the tight emulsions, formed by good oil-cell-water contact and mixing, can be separated continuously with hydrocyclones, to obtain relatively clean oil and water .54 In sum, biodesulfurization has promising prospects in the mid-to-long term . Biodesulfurization can potentially provide a solution to the need for expanded fuel upgrading, because bioprocesses do not require hydrogen and produce far less CO2 than thermochemical processes . However, despite considerable progress, challenges for commercial application remain . As of today, the technology has not yet progressed beyond laboratory-scale testing . Critical aspects include cost and speciﬁc activity of biocatalysts,55 reactor design and oil–water separation. Finally, the widely industrial application of biodesulfurization processes has to overcome the fact that biodesulfurization is unable to work on highly alkylated compounds . Because the composition of organosulfur compounds in oil products during the life of a reﬁnery varies signiﬁcantly, biodesulfurization cannot yet provide the level of reliability required by industrial plants . Therefore, for the near-to-mid term, ODP process for diesel treating and CD process for gasoline treating appear as the most promising desulfurization alternatives. According to the estimates of this study, while conventional HDS for treating ﬁnished oil products potentially expand energy use by 30% in Brazilian reﬁneries, CD and ODP can partially avoid the hydrotreating and reduce by 46% this increase in energy use.
7. Final remarks This paper discusses some challenges faced by the world reﬁning industry, which are prompted by stricter environmental speciﬁcations for oil products. These challenges are driving innovation in oil reﬁneries. For instance, in Brazil, the reduction in the sulfur content of diesel and gasoline between 2002 and 2009 might expand the energy use in reﬁneries by some 30%, with similar implications for CO2 emissions. Precisely, this study forecasted the expanded energy use in Brazilian reﬁneries as 70.4 PJ (expanded CO2 emissions of 1.52 MtC). On the opposite, energy savings options can reduce energy use by 35 PJ (and CO2 emissions by 0.75 MtC);56 and alternative desulfurization techniques, which are applicable in the near-to-mid term, can reduce this energy use by 32 PJ (and CO2 emissions by 0.57 MtC). 54 A promising alternative could be the use of enzymes active in nonaqueous media [86,87]. 55 For instance, the current available biocatalysts require an increase in desulfurization rate of about 500-fold . 56 This ﬁgure considers an energy savings potential of 15%, deriving from measures applied simultaneously with alternative treatment processes. Thus, energy savings measures can potentially reduce energy use in Brazilian reﬁneries that simultaneously lower their fuel consumption by adopting alternative treatment processes.
Therefore, the expanded energy use (or CO2 emissions) in Brazilian reﬁneries resulting from severe hydrotreatment to comply with more stringent speciﬁcations of oil products may be almost completely offset by energy savings options and alternative desulfurization techniques, if barriers to invest in technological innovation are overcome. Actually, the petroleum industry is competitive and the domestic downstream business has, on average, ﬁnancial results below other segments of the oil business. The low returns in the reﬁning and marketing segment result from the competitive nature of the business as well as the signiﬁcant amount of regulatory driven investments that tend to capture little to no return in the market . In Brazil, uncertainties about the domestic oil market regulation (price and quality) are also worthwhile. Therefore, decisions by reﬁners to invest in expanding capacity are complex and depend on expectations of return on investment. Reﬁnery assets are long-lived and capital needs in reﬁning compete with capital needs in other petroleum segments. As such, usually, reﬁners tend to present risk aversion for investing in drastic technological innovations, whose return depends on the uncertain premium price of highly speciﬁed oil products. Moreover, spending on mandatory environmental projects can detract from investments in the core business, which provide capacity growth, yield ﬂexibility and reliability improvements. Finally, it is worthwhile to stress that the decreased sulfur content of diesel and gasoline affects not only the energy use and the CO2 emissions by reﬁneries worldwide, but also presents impacts on the fuel efﬁciency of engines (cars and trucks). As this paper discussed before, lowering the sulfur content of the gasoline pool, through hydrotreating units, reduces its octane number or the thermodynamic performance of the Otto engine fuelled with it. On the other hand, lowering sulfur in gasoline improves the performance of the three-way catalytic converter, reducing NOx and CO emissions by the engine. In the case of diesel, lowering its sulfur content affects negatively its lubricity, but positively its cetane number, or the thermodynamic performance of the diesel engine. In addition, near-zero sulfur diesel allows advanced post-engine exhaust cleanup . Overall, the complete life-cycle energy efﬁciency impact would probably more than offset the expanded energy use in the reﬁnery, which was estimated by this paper. However, as mentioned before, this paper emphasizes alternatives directly related to oil reﬁning (or undertaken inside reﬁneries). Reﬁners must deal with the expansion of in-house CO2 emissions and energy use, as the ﬁrst is an increasing issue for industrial activities, and the latter represents an important share of reﬁners’ operational costs.
References  Alsema, E. Icarus-4. Sector Study for the Reﬁneries. Report no. NWS E 2001 26. Utrecht University, 2001.
ARTICLE IN PRESS A. Szklo, R. Schaeffer / Energy 32 (2007) 1075–1092  EIA/DOE. Manufacturing Energy Consumption Survey 1994. Energy Information Administration. Washington: DOE, 1997.  Worrell E, Galitsky C. Proﬁle of the Petroleum Reﬁning Industry in California. California Industries of the Future Program. Energy Analysis Department. Environmental Energy Technologies Department, Berkeley, March 2004.  Nakamura D. Reﬁning capacity creeps higher in 2003. Oil and Gas Journal 2003;212:64–9.  Faiz A, Weaver C, Walsh M. Air pollution from motor vehicles. Washington: World Bank; 1996.  Gu¨ru¨ M, Karakaya U, Altiparmak D, Alicilar A. Improvement of diesel fuel properties by using additives. Energy Conversion and Management 2002;43:1021–5.  Yamaguchi N, Fridley D, Xiaoming K. Improving transport fuel quality in China: implications for the reﬁning sector. Final Report August 2002. LBNL-50681 Lawrence Berkeley National Laboratory, Berkeley, CA, 2002.  Junior HJ. Especiﬁcac- a˜o de Derivados Tendeˆncias. Comissa˜o de Energia e Meio Ambiente Associac- a˜o Nacional dos Fabricantes de Veı´ culos Automotores. Technical presentation at Semina´rio IBP: Futuro do Reﬁno de Petro´leo no Brasil. November 23, 2004 [in Portuguese].  Carvalho PA. Especiﬁcac- o˜es futuras da gasolina e do o´leo diesel. Superintendeˆncia de Qualidade de Produtos—ANP. Technical presentation at Semina´rio IBP: Futuro do Reﬁno de Petro´leo no Brasil, November 23, 2004 [in Portuguese].  Tavares M. Ana´lise do Reﬁno no Brasil: estado e perspectivas—uma ana´lise cross-section. D.Sc. Thesis. Energy Planning Program of UFRJ, 2005. See also: /www.ppe.ufrj.brS [in Portuguese].  Plantenga F, Leliveld R. Sulfur in Fuels: more stringent sulfur speciﬁcations for fuels are driving innovation. Applied Catalysis A: General 2003;248:1–7.  Song C. An overview of new approaches to deep desulfurization for ultra-clean gasoline, diesel fuel and jet fuel. Catalysis Today 2003; 86:211–63.  Brunet S, Mey D, Pe´rot G, Bouchy C, Diehl F. On the hydrodesulfurization of FCC gasoline: a review. Applied Catalysis A: General 2005;278:143–72.  Swaty T. Global reﬁning industry trends: the present and future. Hydrocarbon Processing 2005;84:35–46.  Petrick M, Pellegrino J. The potential for reducing energy consumption in the reﬁning industry. Report no. ANL/ESD/TM-158. Argonne National Laboratory, 1999.  SHELL. Reducing our own emissions. 2006. See also: /http:// www.shell.com/home/Framework?siteId=royal-en&FC2=/royal-en/ html/iwgen/environment_and_society/key_issues_and_topics/issues/ climate_change/zzz_lhn.html&FC3=/royal-en/html/iwgen/environment_ and_society/key_issues_and_topics/issues/climate_change/reducing_ our_own_emissions.htmlS.  NPC [National Petroleum Council]. Observations on petroleum product supply. Washington: NPC Press; 2004.  Hackworth J, Shore J. Distillate in depth—the supply, demand, and PRICE picture. Winter fuels conference, October 12, 2005.  Lu¨ H, Gao J, Jiang Z, Jing F, Yang Y, Wang G, et al. Ultra-deep desulfurization of diesel by selective oxidation with [C18H37N(CH3)3]4[H2NaPW10O36] catalyst assembled in emulsion droplets. Journal of Catalysis 2006;239:369–75.  Jeyagowry T, Xiao H, Dou J, Nah T, Rong X, Kwan W. A novel oxidative desulfurization process to remove refractory sulfur compounds from diesel fuel. Applied Catalysis B: Environmental 2006;63:85–93.  Szklo A, Schaeffer R. Alternative energy sources or integrated alternative energy systems? Oil as a modern lance of Peleus for the energy transition. Energy—the International Journal 2006;31:2177–86.  Szklo A. Fundamentos do Reﬁno de Petro´leo. Rio de Janeiro: Editora Intercieˆncias, 2005 [in Portuguese].  Energetics. Energy and Environmental Proﬁle of the US Petroleum Reﬁning Industry December 1998 Prepared by Energetics Incorpo-
       
rated Columbia, Maryland Prepared for US Department of Energy Ofﬁce of Industrial Technologies, 1998. API. Technological roadmap. Houston: American Petroleum Institute; 2000. EIPCCB. Reference document for BAT for mineral reﬁneries. Integrated pollution and control. Belgium: European Comission; 2001. Guchhait S, Biswas D, Bhattacharya P, Chowdhury R. Biodesulfurization of model organo-sulfur compounds and hydrotreated diesel—experiments and modeling. Chemical Engineering Journal 2005;112:145–51. Santana R, Do P, Santikunaporn M, Alvarez W, Taylor J, Sughrue E, et al. Evaluation of different reaction strategies for the improvement of cetane number in diesel fuels. Fuel 2006;85:643–56. Meyers R. Handbook of petroleum reﬁning processes. 2nd ed. New York, NY: McGraw-Hill; 1997. Rostrop-Nielsen T. Manufacture of hydrogen. In: Gas-Fuel 05 seminar, Brugges, Belgium, 15–17 November 2005. EIA/DOE. Supply Impacts of an MTBE Ban September 2002 Energy Information Administration. Washington: DOE; 2002. UNFCCC, 2006. See also: /www.unfccc.intS. Assayag M. Engenharia ba´sica da Petrobra´s. Technical report. Rio de Janeiro: CENPES; 2003 [in Portuguese]. Petrobras. Desenvolvimento e aplicac- a˜o de tecnologia Petrobras de HIDRORREFINO. Rio de Janeiro: Petrobras; 2003 [in Portuguese]. Tavares M, Szklo A, Machado G, Schaeffer R, Mariano J, Sala J. Oil reﬁning expansion criteria for Brazil. Energy Policy 2006;34:3027–40. Szklo A, Machado G, Schaeffer R, Simo˜es A, Mariano J. Placing Brazil’s heavy acid oils on international markets. Energy Policy 2006;34:692–705. VROM. Dutch notes on BAT for the mineral oil reﬁneries. Ministry of Housing, Spatial Planning and the Environment, 1999. SEMA. Technical note on the best available technologies to reduce emissions of pollutants into the air from the reﬁning industry. Application of articles 7 and 13 of Directive 84/360 EEC. Report for the European Comission, 1991. Dusse C, Fernandes N, Pinto N, Arau´jo S. Otimizac- a˜o de diesel na REGAP em 2003: melhorias na unidade de hidrotratamento de diesel e no acompanhamento do pool de diesel da REGAP. Boletim te´cnico Petrobra´s 2003, 46 [in Portuguese]. Almeida R, Guirardello R, Souza G. Cine´tica da remoc- a˜o de enxofre na hidroconversa˜o de resı´ duo de va´cuo de petro´leo Marlim. Boletim te´cnico da Petrobra´s 1999. p. 42 [in Portuguese]. Schaeffer R, Szklo A, Machado G, Neto V. Balanc- o Energe´tico do Estado do Rio de Janeiro. Rio de Janeiro: SEINPE 2004 [in Portuguese]. MCT [Ministry of Science and Technology of Brazil]. Emisso˜es de Gases de Efeito Estufa por Queima de Combustı´ veis. Abordagem Bottom-Up, 2002. See also /ww.mct.gov.brS [in Portuguese]. Cima F. Utilizac- a˜o de indicadores energe´ticos no planejamento energe´tico integrado. M.Sc. Thesis. Federal University of Rio de Janeiro. Rio de Janeiro: COPPE/UFRJ; 2006. See also /www.ppe.ufrj.brS [in Portuguese]. Najam A, Huq S, Sokona Y. Climate negotiations beyond Kyoto: developing countries concerns and interests. Climate Policy 2003;3:221–31. Negrao C, Madi M, Massoqueti R. Simulac- a˜o do desempenho de redes de trocadores de calor sujeitas a` incrustac- a˜o. Technical paper presented at Rio Oil & Gas Expo and Conference 2004, 4–7 October, Rio de Janeiro, 2004 [in Portuguese]. Szklo A, Soares J, Tolmasquim M. Economic potential od natural gas-ﬁred cogeneration—analysis Brazil’s chemical sector. Energy Policy 2004;32:1415–28. Olim J, Haddad J, Viana A, Martins A. Conservac- a˜o de energia em reﬁnarias de petro´leo. Tecnical paper presented at 21 Congresso Brasileiro de P&D em Petro´leo e Ga´s, October 2002 [in Portuguese]. Hallale N. Burning bright: trends in process integration. Chemical Engineering Progress 2001;7:30–41.
ARTICLE IN PRESS 1092
A. Szklo, R. Schaeffer / Energy 32 (2007) 1075–1092
 CTEC. L’analyse Pinch: pour une utilisation efﬁcace de l’e´nergie, de l’eau et de l’ hydroge`ne. Canada´: CETC-Varennes; 2003 [in French].  Worrell E, Galitsky C. Energy efﬁciency improvement and cost saving opportunities for petroleum reﬁneries. Environmental Energy Technologies Division, Berkeley, February 2005.  Linnhoff B, Townsend D, Boland D, Hewitt G, Thomas B, Guy A, et al. A user guide on process integration for the efﬁcient use of energy (1992 edition). Rugby, UK: Institution of Chemical Engineers; 1992.  Linnhoff B. Use pinch analysis to knock down capital costs and emissions. Chemical Engineering Progress 1994;90:33–57.  Beer J. de. Potential for industrial energy efﬁciency improvement in the long term. Ph.D. Thesis, Utrecht, ISBN 90-3939-1998-7, 1998.  Amorim R. Abastecimento de A´gua de uma Reﬁnaria de Petro´leo: Caso Replan. M.Sc. Thesis. Engineering School. Rio de Janeiro: Fluminense Federal University (UFF); 2005 [in Portuguese].  Schor A. Water supply risks and alternatives in an oil reﬁnery. Case Study: Reﬁnaria Duque de Caxias—REDUC /RJ. M.Sc. Thesis, Civil Engineering, COPPE/UFRJ, 2006 [in Portuguese].  Jacobs Consultancy INC. Water pollution prevention opportunities in petroleum reﬁneries prepared for the Washington State Department of Ecology. Texas Ecology Publication No. 02-07-017, November 2002.  EXXON. Economic penalties associated with the fouling of reﬁnery heat transfer equipment. In: Sommerscales and Knudsen, editors. Fouling of heat transfer equipment. New York: McGraw-Hill; 1981.  ANL. Effects of fouling mitigation on the energy efﬁciency of crude oil distillation. Presented at the AIChE National Spring Meeting, New Orleans, 1998.  Bailey K. Optimize heat exchanger operations by minimizing fouling. Hydrocarbon Processing 1999;78:113–6.  Bott T. To foul or not to foul, that is the question. Chemical Engineering Progress 2001;11(November):30–6.  Panchal C, Huangfu E. Effects of mitigating fouling on the energy efﬁciency of crude-oil distillation. Heat Transfer Engineering 2000;3:21.  Hydrocarbon Processing (HCP). Advanced Control and Information Systems 2001. Hydrocarbon Processing 2001;9(September):73–159.  Timmons C, Jackson J, White D. Distinguishing online optimization beneﬁts from those of advanced controls. Hydrocarbon Processing 2000;6:69–77.  Aprea R, Mastrullo C, Renno C. Fuzzy control of the compressor speed in a refrigeration plant. International Journal of Refrigeration 2004;27:639–48.  Katzer J, Ramage M, Sapre A. Petroleum reﬁning: poised for profound changes. Chemical Engineering Progress 2000;7(July):41–51.  Hydrocarbon Processing (HCP). Reﬁning processes 1998 and HP innovations. Hydrocarbon Processing 1998, November 1998. p. 53–120.  Schultz M, Stewart D, Harris J, Rosenblum S, Shakur S, O’Brien D. Reduce costs with dividing-wall columns. Chemical Engineering Progress 2002;5(May):64–71.  Babich I, Moulijn J. Science and technology of novel processes for deep desulfurization of oil reﬁnery streams: a review. Review Article. Fuel 2003;82:607–31.  Baker R, Lokhandwala K, Jacobs M, Gottschlich D. Recover feedstock and product from reactor vent streams. Chemical Engineering Progress 2000;12:51–7.  Reedy G, Schwarz E, Dolan D. Chevron Texaco CDHydro s/ CDHDSs startup. ERTC, November 18–20, Paris, 2002.  Zekai Z, Hui J, Shenglin L, Wang Q, Xu L. Alkylation performance of tiophene and its derivatives during oleﬁnic alkylation of tio-
phenic sulfur in gasoline. Chines Journal of Catalysts 2006;27(4): 309–13. Alexander BD, Huff GA, Pradhan VR, Reagan WJ, Cayton RH. US Patent 6,024,865 2000. Zapata B, Pedraza F, Valenzuela M. Catalyst screening for oxidative desulfurization using hydrogen peroxide. Catalyst Today 2005;106: 219–21. Ali M, Al-Malki A, El-Ali B, Martinie G, Siddiqui M. Deep de sulfurization of gasoline and diesel fuels using non-hydrogen consuming techniques. Fuel 2006;85:1354–63. Peninger R, Demeter J, Judzis A, Schwarz E, Rock K. Start-up pf ﬁrst CDHDSs Unit at Motiva’s Port Arthur, Texas Reﬁnery. NPRA, March 18–20, New Orleans, 2001. Gardner R, Schwarz E, Rock K. Start-up pf ﬁrst CDHydros/ CDHDSs Unit at Irving Oil’s Saint John, New Brunswick Reﬁnery. NPRA, March 18–20, New Orleans, 2001. Hagiwara T. Gasoline production technology and methods, and an evaluation of their economic viability. 01 Survey 4-2-1, Technology Department, Petroleum Energy Center (PEC), 2001 [in Japan]. Linquist L, Pacheco M. Enzyme-based desulfurization process offers energy, CO2 advantages. Oil Gas Journal 1999;97:45–8. Kilbane J. Petroleum bioreﬁning for pollution prevention. Department of Energy, National Petroleum Technology Ofﬁce, Downstream Project Results, vol. 1/2, Fall 2001. Guobin S, Huaiyguin Z, Jianmin X, Guo C, Wangliang L, Huizhou L. Biodesulfurization of hydrodesulfurized diesel oil with Pseudomonas delafieldii R-8 from high density culture. Biochemical Engineering Journal 2006;27:305–9. Li W, Zhang Y, Wang M, Shi Y. Biodesulfurization of dibenzothiophene and other oragniz sulfur compounds by a newly isolated Microbacterium strain ZD-M2. FEMS Microbiology Letters 2005; 247:45–50. Castorena G, Suarez C, Valdez I, Amador G, Fernadez L, Le Borgne S. Sulfur-selective desulfurization of dibenzothiophene and diesel oil by newly isolated Rhodococus sp strains. FEMS Microbiology Letters 2002;215:157–61. DOE. US Department of Energy—Ofﬁce of Energy Efﬁciency and Renewable Energy Industrial Technologies Program. Gasoline Biodesulfurization, May 2003. Cui-Qing M, Jin-Hui F, Yi-Yong Z, Xiao-Feng C, Bai-Ping S, ZhaoBin Z, et al. Methods for the preparation of a biodesulfurization biocatalyst using Rhodococus sp. Chemosphere, accepted in 2006, in press. See online available electronic version at: /www.sciencedirect.comS. Van Hamme J, Singh A, Owen P. Recent advances in petroleum microbiology. Microbiology and Molecular Biology Reviews 2003; 67:503–49. Monticello D. Riding the fossil fuel biodesulfurization wave. Chemtech 1998;28:38–45. Kayser K, Bielaga-Jones B, Jackowski K, Odusan O, Kilbane J. Utilization of organosulfur compounds by axenic and mixed cultures of Rhodococcus rhodocrous IGTS8. Journal of General Microbiology 1993;139:3123–9. Denome S, Olson E, Young K. Identiﬁcation and cloning of genes involved in speciﬁc desulfurization of dibenzothiophene by Rhodococcus sp. strain IGTS8. Applied and Environmental Microbiology 1993;59:2837–43. Kilbane J. Microbial biocatalyst developments to upgrade fossil fuels. Current Opinion in Biotechnology 2006;17:305–14. Pennig R. Petroleum reﬁning a look at the future. Hydrocarbon Processing 2001;2:45–6.