Biodesulphurization of fossil fuels: energy, emissions and cost analysis

Abstract

In order to achieve stringent environmental and safety requirements, refineries are in search of “green” and cost-effective methods for crude oil desulphurization. Combined desulphurization technologies are being studied, including bioprocessing to upgrade fossil fuels. Using biodesulphurization (BDS), which is a biochemical process mediated by specific microorganisms, it is possible to desulphurize most of the hydrodesulphurization (HDS) recalcitrant sulphur compounds under mild operating conditions, making it a simple and eco-friendly process. In this study, two BDS process designs are compared, in terms of energy consumption, greenhouse gas emissions and operational costs by following a life cycle assessment (LCA) and life cycle cost (LCC) based methodology. The industrial HDS process is used as the reference technology for sulphur removal from fossil fuels. Different theoretical scenarios were considered and the best BDS results are scaled-up to evaluate a case study of providing ultra low sulphur diesel to an urban taxi fleet. This study exploits the potential of BDS as a cost-effective and eco-friendly alternative or complementary technology to the commonly HDS towards ultra low sulphur fuels.

---

I. Introduction

The combustion of fossil fuel generates emissions of sulphur as sulphur dioxide (SO₂), that is corrosive and toxic, and as fine particulate matter of metal sulphates. These emissions are responsible for damage in many different areas. Gaseous chemical compounds of sulphur constitute a major health hazard when present in the air: the large-ring thiophenes, such as dibenzothiophene, abundant in crude oil, are toxic to mammals;¹ SO₂ gas at high levels can cause bronchial irritation and trigger asthma attacks in susceptible individuals and long-term exposure to combustion-related fine particulate air pollution is an important risk factor for cardio-pulmonary and lung cancer mortality.^2,3 In addition, incomplete burning of liquid fossil fuels causes emissions of aromatic sulphur compounds to the air,⁴ and the oxidation of sulphur compounds in the atmosphere eventually leads to aerosol of sulphuric acid. This aerosol causes acid rains, which are responsible for the corrosion of many infrastructures and monuments, and even affect several living organisms including agricultural crops, thus causing direct damage to the economy.⁵ The aerosol is also harmful to the stratospheric ozone contributing to the hole on the Earth's protective ozone layer.⁶ Lastly, sulphur compounds even prevent functioning of all major pollution control technologies such as automobile catalytic converters⁷ making it more difficult to fight against pollution.

Since gasoline, diesel and non-transportation fuels account for 75 to 80% of the total refinery products,⁸ it is only natural that countries find the reductions of sulphur concentration in fuels as the most effective way to decrease the amount of SO₂ emitted in to the air and limit its prejudicial effects.⁹ Therefore, as a response to the environmental and health problems, there has been an increasing concern regarding the sulphur levels in fossil fuels and its derivatives. Throughout the world countries have been drastically reducing the legal limits on sulphur concentration of these energy sources. For the transportation fuels, namely gasoline and diesel, a 500 ppm limit was admissible in the 90's, but this limit decayed to 10 ppm for EU, since 2009, and 15 ppm for US, since 2006.¹⁰ Furthermore, it has to be expected that the sulphur level in on-road and off-road gasoline and diesel requirements will become stricter in the foreseeable future towards ultra low sulphur fuels, approaching zero sulphur emissions from burned fuels. Therefore, the efficiency of the desulfurization technologies becomes a key point.^8,11

Sulphur is the third most common element in crude oil composition after carbon and hydrogen, but the total sulphur content of crude oil varies from reservoir to reservoir, with an average content of 0.03 to 7.89% by weight.¹² Depending on the amount of sulphur the crude oil can be sweet or sour. When the total sulphur level in the oil is less than 0.5% the oil is called sweet and if it is more than that the oil is called sour.¹³ The sulphur compounds can be found in two forms: inorganic and organic, but the organosulphur compounds such as thiols, sulphides, and thiophenic compounds represent the main source of sulphur found in crude oil.

Hydrodesulphurization (HDS) or hydrotreating is the most commonly used method in the petroleum industry to reduce the sulphur content of crude oil and refined petroleum products. In most cases HDS is performed by co-feeding oil and H₂ to a fixed-bed reactor packed with an efficient inorganic catalyst (e.g. CoMo/Al₂O₃; NiMo/Al₂O₃). During HDS, the sulphur in the organosulphur compounds is converted to H₂S, at high-pressure (1–18 MPa) and high-temperature (200–450 °C), the specific conditions depending on the degree of desulphurization required and the nature of the sulphur compounds in the feed.^13–17 The reactivity of organosulphur compounds varies widely depending on their structure and the local sulphur atom environment.

HDS removes the bulk of inorganic sulphur and light organosulphur compounds but is not effective for removing heterocyclic sulphur compounds such as dibenzothiophene (DBT) and its alkyl derivatives, specially 4- or 6-alkyldibenzothiophene (e.g. mDBT) and 4,6-dialkyldibenzothiophene (e.g. 4,6-dmDBT). Deep desulphurization of gasoline/diesel towards ultra low sulphur (ULS fuels: 10 mg kg⁻¹) is largely restricted by the low-reactivity of these sulphur compounds.^18,19 Moreover, HDS is a very costly option for ultra-deep desulphurization since it implies harsher conditions. For example, the cost of desulphurization of 20 000 barrel of oil per day is as much as $40 million and additional hydrogen and sulphur plant capacities would increase the investment into refinery plant.^13,20

Since the production of ULS automotive fuels has gained enormous interest in the scientific community worldwide, other desulphurization technologies much more efficient and less expensive in removing HDS recalcitrant organosulphur compounds, such as oxidative desulfurization (ODS), extractive desulfurization, adsorptive desulfurization and biodesulphurization (BDS), are being used in test scale and commercial scale projects.^8,16,21–23 Among these, ODS combined with extraction (OEDS) or adsorption is considered to be one of the most promising processes to remove refractory sulphur compounds from diesel fuel. In fact, some oxidative desulphurization processes have been proposed as a tool to upgrade downstream into a conventional refinery for the production of ULS diesel.^24,25

In the last decade, BDS has drawn wide attention because of its green processing of fossil fuel. Bioprocesses can potentially provide a solution to the need for improved and expanded fuel upgrading worldwide, because bioprocesses for fuel upgrading do not require hydrogen and produce far less greenhouse gas (GHG) emission than thermochemical processes.^26–28

BDS takes place at low temperatures (<35 °C) and pressure (∼1 atm.) in the presence of microorganisms that can specifically break C–S bonds, especially for DBT and substituted DBTs, thereby releasing the sulphur in a water-soluble, inorganic form. The preferred microorganisms for BDS are those that can selectively remove the sulphur from the recalcitrant compounds through the via 4S-pathway regulated by a dsz operon (e.g. Rhodococcus and Gordonia strains), in which the calorific value of the fuel is preserved.^17,28,29 Exhaustive work has been done towards fossil fuels biodesulphurization in last decades, especially BDS studies using the DBT as a model compound of one of the most recalcitrant compounds in crude oil.^28–38 In fact, this bioprocess has the potential of being developed as a viable technology downstream of classical HDS towards ULS fuels (ULS diesel).^{27,29,39–42} Lower capital and operating costs, ability to be less polluting and produce high valuable by-products, such as biosurfactants,^38,40,43 are some benefits of BDS. Therefore, recent advances have been make to boost BDS efficiency, both by classic microbiology methods and genetic engineering techniques (recombinant strains for overexpression of dsz genes; removal of end-product repression) targeting cost-effectiveness and the ability of the bioprocess to integrate as seamlessly as possible into the existing petrochemical operations to generate ULS diesel (ULSD).^{23,29,36,44,45}

There are very few reports on BDS process designs and cost analysis. In the current study, two BDS process designs are analysed in terms of energy consumption, greenhouse gas emissions (GHG) and costs. The BDS experimental results evaluated in this work are those obtained previously by Alves and Paixão⁴⁶ for DBT desulphurization by Gordonia alkanivorans strain 1B using sugar beet molasses (SBM) as carbon source, either after acid hydrolysis or in a simultaneous saccharification and fermentation (SSF) process with Zygosaccharomyces bailii strain Talf1 enzymes (invertase activity). DBT was used as sulphur source instead of diesel fuel.

The methodology for the analysis of whole processes is based on ISO 14040/44 life cycle standards and is specifically focused on greenhouse gas environmental impact category as in GHG Protocol.⁴⁷ Final energy consumption is taken to evaluate upstream primary energy needs and respective GHG emissions. The best BDS process is scaled-up to evaluate a real case study of providing ULSD to an urban taxi fleet.

II. Methodology

A. Life cycle based methodology

The energy and emissions balance of an engineered process is crucial to identify potential enhancements, inefficient steps and establish a process environmental and economic viable. Life cycle assessment, LCA, follows the ISO 14040 principles, and looks at resource depletion, energy consumption, and substance emissions of all the processes required to achieve the production, use and end-of-life of a product, and second, the assessment of several environmental damages, e.g. global warming potential. The methodology for the analysis of whole processes is based on life cycle standards and is specifically focused on greenhouse gas environmental impact category as in GHG Protocol.⁴⁷ According to the principles of ISO 14040 (ISO 14040–14043), this methodology is composed by four main steps: (1) definition of goal, scope and boundaries; (2) inventory (Life Cycle Inventory, LCI); (3) impact assessment (Life Cycle Impact Assessment, LCIA) and; (4) analysis.

A.1. Goal, scope and boundaries. The main goal of this research is comparing two biodesulphurization processes, whose boundaries are represented in Fig. 1. The desulphurization unit has upstream emissions referring to the gate-to-gate requirements. The main input materials are DBT, bacteria and sugar (C-source). Main output materials are desulphurized product and fermentative medium remains containing microbial biomass. The energy requirements from BDS were compared with those for an industrial case⁴⁸ using the commonly HDS for sulphur removal from fossil fuels.

Fig. 1 Pathways for DBT desulphurization process by G. alkanivorans strain 1B: experimental steps including the respective inputs/outputs.

The gate-to-gate analysis is modular having the advantage of be later on linked to the appropriate overall refining process chain to form a complete cradle-to-gate evaluation. The considered functional unit (FU) is 1 g of sulphur removed.

A.2. Inventory for biodesulphurization. In the inventory analysis for BDS, the raw material requirements, energy, and atmospheric emissions are quantified. As already mentioned, the BDS experimental results evaluated herein are those obtained previously by Alves and Paixão.⁴⁶ Thus, two pathways for DBT desulphurization by Gordonia alkanivorans strain 1B using sugar beet molasses (SBM) treated with 0.25% (w/v) BaCl₂ as carbon source were evaluated. In Path #1: DBT desulphurization process by strain 1B was carried out using SBM treated with BaCl₂ and hydrolysed through acid hydrolysis. In Path #2: the BDS by strain 1B was carried out using SBM treated with BaCl₂ and hydrolysed in a SSF process with Z. bailii Talfl invertase crude extract (1%). In SSF process, both steps (hydrolysis and sugar consumption) occur simultaneously in the course of BDS at the optimal growth and desulphurization conditions for the desulphurizing bacterium (pH = 7.5, 30 °C).

Tables 1 and 2 show the inventories for the two pathways. In these BDS assays, the desulphurized product is 2-hydroxybiphenyl (2-HBP) instead of diesel as in the conventional refining process. The BDS boundaries in analysis, schematized in Fig. 1, include SBM treatment (i.e. sulphate precipitation with BaCl₂), the acid hydrolysis (only in Path #1), the enzymatic hydrolysis (including enzyme production by growing Z. bailii Talf1 in culture medium – only in Path #2), sterilization and desulphurization (including culture medium composition).

Table 1 Inventory for DBT desulphurization process by G. alkanivorans strain 1B through Path #1 (note: O.I.: orbital incubator; In.: inoculum. *Quantity and cost are referred to the actual values calculated for assays using 150 mL of BDS culture medium³¹)

Process	Item	Quantity*	Cost*
INPUTS
Source material	SBM	1.800 g	0.11 € per kg
	DBT	250 μM	—
SBM treatment	Incubation (BaCl2)	0.085 MJ	0.096 € per kW per h
	Centrifugation	0.201 MJ	0.096 € per kW per h
SBM acid hydrolysis	HCl 37%	7.386 g	0.08 € per kg
	Incubation	0.114 MJ	0.096 € per kW per h
	NaOH	0.047 g	4.65 € per kg
Biodesulphurization
Microbial growth medium	NH4Cl	0.183 g	19.13 € per kg
	KH2PO4	0.375 g	15.20 € per kg
	Na2HPO4·2H2O	0.375 g	15.20 € per kg
	MgCl2·6H2O	0.026 g	7.84 € per kg
	Metals solution	0.075 g	1.00 € per kg
	NaOH	0.025 g	4.58 € per kg
	Sterilization	0.026 MJ	0.096 € per kW per h
	Incubation (O.I.)	0.190 MJ	0.096 € per kW per h
G. alkanivorans 1B (2% In.)	2% of sterilization and O.I.	0.004 MJ	0.0003 € per kW per h
OUTPUTS
2-HBP production (maximum production rate)		175 μM (5.76 μM h^−1)
Sulphur removed		0.0056 g
Microbial biomass		4.8 g L^−1

---

Table 2 Inventory for DBT desulphurization process by G. alkanivorans strain 1B through Path #2 (note: O.I.: orbital incubator; In.: inoculum. *Quantity and cost are referred to the actual values for assays using 150 mL of BDS culture medium³¹)

Process	Item	Quantity*	Cost*
INPUTS
Source material	SBM	1.800 g	0.11 € per kg
	DBT	250 μM	—
SBM treatment	Incubation (BaCl2)	0.085 MJ	0.096 € per kW per h
	Centrifugation	0.201 MJ	0.096 € per kW per h
Biodesulphurization
Microbial growth medium	NH4Cl	0.183 g	19.13 € per kg
	KH2PO4	0.375 g	15.20 € per kg
	Na2HPO4·2H2O	0.375 g	15.20 € per kg
	MgCl2·6H2O	0.026 g	7.84 € per kg
	Metals solution	0.075 g	1.00 € per kg
	NaOH	0.025 g	4.58 € per kg
	Sterilization	0.026 MJ	0.096 € per kW per h
	Incubation (O.I.)	0.190 MJ	0.096 € per kW per h
G. alkanivorans 1B (2% In.)	2% of sterilization and O.I.	0.004 MJ	0.0003 € per kW per h
Z. bailii Talf1 invertases (1%)	1% of sterilization and O.I.	0.002 MJ	0.0001 € per kW per h
	Sterilization	0.026 MJ	0.096 € per kW per h
	Incubation (O.I.)	0.733 MJ	0.096 € per kW per h
OUTPUTS
2-HBP production (Maximum production rate)		249 μM (7.78 μM h^−1)
Sulphur removed		0.0080 g
Microbial biomass		5.9 g L^−1

---

Sulphur (S) mass balance is used to quantify the quantity of its removal from DBT. Each mol of DBT has one mole of S, therefore each mole of 2-HBP produced is equivalent to one mole of S removed by strain 1B. In Path #1 was observed a S removal of 0.0056 g (i.e. 70% SRE – Sulphur Removal Efficiency) after 75 h at 30 °C, while in Path #2 was observed a S removal of 0.008 g (99.6% SRE) after 48 h at 30 °C.

For nutrients used in culture medium, the energy consumption and GHG emission were estimated by the Ecoinvent 2.0 database⁴⁹ adapted to the Portuguese electricity fuels generation mix. The remaining energy inputs, for the used equipment, were derived from the device specifications and operational hours or by direct measurements. Only operational processes were accounted for, i.e. equipment production and the final fuel use/distribution were not included.

A.2.1. Reference crude oil hydrotreating. A petroleum refinery processes crude oil into thousands of products using a thermochemical processing technology. The refinery receives the crude oil, a complex mixture containing many hydrocarbon compounds, and uses distillation processes to separate pure product streams. Because the crude oil is contaminated (to varying degrees) with compounds of sulphur, nitrogen, oxygen, and metals, cleaning operations through hydrotreatment with hydrogen input at high-pressure and temperature in presence of metal catalysts are common.

Fig. 2 shows the most energy-intensive processes of a petroleum refinery (desalting, atmospheric distillation, vacuum distillation and hydrotreating), representing over 95% of the total energy requirements of U.S. petroleum refineries.⁵⁰ In particular, for HDS the oil feed to the hydrotreater unit is mixed with hydrogen-rich gas before entering a fixed-bed reactor. In the presence of a metal-oxide catalyst, hydrogen reacts with the oil feed to produce hydrogen sulfide, ammonia, saturated hydrocarbons, and other free metals. A large refinery may have five or more hydrotreaters, but most HDS processes have essentially the same process flow (Fig. 3).

Fig. 2 Simplified flow diagram of a petroleum refinery for the production of fuels/feedstocks, considering only the most energetic processes (adapted from F. Associates⁵⁰). The percentages of refinery products represent percent by mass of total refinery output. Other* includes still gas, petroleum coke, asphalt and petrochemical feedstocks.

Fig. 3 Typical catalytic hydrotreating process flow diagram (adapted from Energetics Incorported⁴⁸).

Data concerning energy consumption of a typical U.S. refinery may be found in Energetics Incorporated.⁴⁸ This is for crude oil with 1.26% wt sulphur (as for U.S. crude oil mix of 1998 according to Energy Information Administration) and diesel fuel with 500 ppm. This implies a sulphur mass balance of 0.012 g of S removal per g of crude oil (96% removal capability). The electricity consumption of the process, excluding the transmission and generation losses, was found to be 19 000 BTU per bbl_feed (ref. 48) (68 kW h kg_diesel⁻¹) which corresponding a final energy consumption of 20.3 MJ per g of S removed. This will be the reference value to compare with the BDS results. HDS and the production of hydrogen for HDS account for 44% of the diesel production pathway GHG emissions, which means roughly 135 g kg_diesel⁻¹ (ref. 51) (100 g kg_diesel⁻¹ is due to the production of H₂ alone). These GHG emissions are highly dependent on the electricity fuels mix and therefore will not be used for comparison.

A.3. Impact categories and cost analysis. Cumulative energy demand (CED) or primary energy demand, final energy and global warming potential (GWP) are the chosen impact categories. GHG or CO₂ equivalent emissions are based on a 100 year time frame. The national electricity mix of Portugal in 2012 will be used for primary energy demand calculations. The mix of production is composed by 18% renewable, 5% nuclear and 77% fossil thermal power plants, with the specific CO₂ emissions of 380 g kW⁻¹ h⁻¹.⁵² The fossil energy demand (FED) analysis (defined as the percentage of fossil fuels used in the overall process) coupled with the operational costs is fundamental to assess the viability of the processes used in the system.

For the cost analysis only the operational costs were regarded as a consequence of the different inputs considered in BDS processes (Paths #1 or #2) and the respective unity costs are presented in Tables 1 and 2. The operation costs throughout the bioreactor lifetime are only a part of LCC cost analysis but useful to compare the two BDS process designs in monetary units.

A.4. Uncertainty and sensitiveness analysis. The basic uncertainty⁵³ is accounted for and reflected in the minimum (Min) and maximum (Max) values of the primary energy (estimated using the physical content method followed by the International Energy Agency).

Scale-up scenarios were envisaged for the best BDS pathway including some changes to the process towards a more cost-effective industrial process. The different scenarios considered were: use of 100% renewables for electricity generation (scenario 1), heating recovered from other refinery process and avoidance of some electrical energy to generated it (scenario 2), and replacement of industrial residue SBM by sucrose (scenario 3). The sensitiveness of results to these scenarios is highlighted and discussed.

B. Implications of BDS in a taxi fleet

As already stated, from the standpoint of preventing damage to the environment by sulfur oxides and from the standpoint of the increasing rigor of exhaust gas regulations for cars, demand has grown in society for a reduction in the sulfur content of car fuels. Present regulations relating to sulphur content have become increasingly rigorous demanding ULSD oils with no more than 10 ppm of sulphur by mass.

The reduction in the sulphur content of refined diesel oil may easily be correlated, by direct combustion mass balance, with lower SO_x emissions at diesel exhaust. For a typical diesel car exhaust, SO₂ emissions are estimated accordingly to the equation, considering a fuel consumption (FC) of 7 L/100 km and a density of 840 g L⁻¹ (ρ_diesel):

This means that a reduction of the 500 ppm limit to the 10 ppm limit implies a reduction of 0.05 g km⁻¹ to 0.001 g km⁻¹. In this study it is considered a case study application where a hypothetical diesel fuel exiting a crude oil refinery with 500 ppm sulphur will be subjected to the best BDS selected by the previously described life cycle based methodology.

The fleet to study this impact is a typical urban taxi fleet. For example, taking a diesel car fleet, composed by 3100 conventional diesel vehicles, with an average daily distance travelled of about 207 km,⁵⁴ translates into ∼45 000 L of consumed diesel per day. This diesel amount could have been subjected to a BDS process becoming an ULSD and consequently decrease the SO₂ emissions by the fleet.

III. Results and discussion

A. BDS: energy and emissions

The results of energy consumption and GHG emissions calculated for the two BDS process designs using SBM as C-source and DBT as S-source (Paths #1 and #2) are presented in Table 3, detailing the contribution of each step within the different process phases (SBM treatment with BaCl₂, SBM acid hydrolysis and BDS). FED is nearly equal to the FED of the electricity mix that is 82%. In addition, Fig. 4 and 5 permit visualize the overall final energy consumption and GHG emissions per g of S removed in both BDS process designs (Path #1 and Path #2), distinguishing the three process phases input.

Table 3 Energy consumption and GHG emissions for each step of the different phases within the two BDS process designs (Paths #1 versus #2) for an input of 250 μM DBT. The steps are accordingly to the respective pathway inventory

Process design (Path #)	Process phase	Step	Final Energy		Primary energy (MJ)	GHG
			MJ	MJ gS^−1		gCO2 eq	kgCO2 eq gS^−1
a Minimum and maximum values.
Path #1	SBM treatment (BaCl2)	Incubation	0.086	15.283	0.189 (0.163–0.200)^a	9.036 (8.630–9.460)	1.614 (1.541–1.689)
		Centrifugation	0.201	35.924	0.444 (0.383–0.470)	21.240 (20.260–22.266)	3.793 (3.618–3.976)
		Total	0.287	51.207	0.633 (0.546–0.670)	30.24 (28.93–31.76)	5.406 (5.159–5.665)
	SBM acid hydrolysis (only in Path #1)	Reagents	0.002	0.316	0.004 (0.003–0.004)	0.187 (0.178–0.196)	0.033 (0.032–0.035)
		Incubation	0.114	20.377	0.252 (0.217–0.266)	12.048 (11.507–12.613)	2.151 (2.055–2.252)
		Total	0.116	20.694	0.256 (0.256–0.270)	12.24 (11.69–12.81)	2.185 (2.087–2.287)
	BDS	Reagents	0.001	0.185	0.002 (0.001–0.002)	18.872 (17.982–19.776)	3.370 (3.211–3.531)
		Metal solution	0.000	0.013	0.000	0.009 (0.009–0.009)	0.002 (0.002–0.002)
		Sterilization	0.026	4.613	0.057 (0.049–0.060)	2.727 (2.605–2.855)	0.487 (0.465–0.510)
		Incubation	0.190	33.986	0.420 (0.363–0.444)	20.094 (19.192–21.036)	3.588 (3.427–3.756)
		1B inoculum	0.004	0.776	0.010 (0.008–0.010)	0.834 (0.796–0.873)	0.149 (0.142–0.156)
		Total	0.222	39.573	0.489 (0.421–0.516)	42.53 (40.58–44.55)	7.596 (7.247–7.955)
Path #2	SBM treatment (BaCl2)	Incubation	0.086	10.741	0.189 (0.163–0.200)	9.036 (8.630–9.460)	1.134 (1.083–1.187)
		Centrifugation	0.201	25.248	0.444 (0.383–0.470)	21.240 (20.260–22.266)	2.666 (2.543–2.794)
		Total	0.287	35.989	0.633 (0.546–0.670)	30.24 (28.93–31.76)	3.800 (3.626–3.982)
	BDS	Reagents	0.001	0.130	0.002 (0.001–0.002)	18.872 (17.982–19.776)	2.368 (2.257–2.482)
		Metal solution	0.000	0.009	0.000	0.009 (0.009–0.009)	0.001 (0.001–0.001)
		Sterilization	0.026	3.242	0.057 (0.049–0.060)	2.727 (2.605–2.855)	0.342 (0.327–0.358)
		Incubation	0.190	23.886	0.420 (0.363–0.444)	20.094 (19.192–21.036)	2.522 (2.409–2.640)
		1B inoculum	0.004	0.273	0.010 (0.008–0.010)	0.834 (0.796–0.873)	0.052 (0.050–0.055)
		Invertases	0.002	0.545	0.005 (0.004–0.005)	0.417 (0.398–0.437)	0.105 (0.100–0.110)
		Total	0.224	28.085	0.493 (0.425–0.521)	42.94 (40.98–44.99)	5.391 (5.143–5.646)
Complete BDS process – Path #1			0.624	111.474	1.377 (1.188–1.456)	85.04 (81.16–89.08)	15.187 (14.493–15.908)
Complete BDS process – Path #2			0.511	64.074	1.126 (0.971–1.191)	73.22 (69.87–76.71)	9.190 (8.769–9.627)

---

Fig. 4 Final energy consumption per g of S removed (MJ g_S⁻¹) by G. alkanivorans strain 1B in main phases within DBT desulphurization process designs: Path #1 vs. Path #2 overview (note: the reference value for HDS energy is 20.3 MJ g_S⁻¹).

Fig. 5 GHG emissions per g of S removed (kg_{CO2 eq} g_S⁻¹) by G. alkanivorans strain 1B in main phases within DBT desulphurization process designs: Path #1 vs. Path #2 overview.

The results detailed in Table 3 clearly indicate that for DBT desulphurization by strain 1B using SBM, the most critical steps in terms of final energy consumption (MJ g_S⁻¹) were the centrifugation and the incubation. In Path #1 the energy consumption within these steps was 105.57 MJ g_S⁻¹, while in Path #2 it was 59.88 MJ g_S⁻¹, but for both BDS pathways the level of energy consumed accounted >93% of the total energy consumption. In the same way, the incubation and centrifugation steps in both BDS processes were the main responsible for the overall GHG emissions (kg of CO₂ equivalents per g of S removed) together with the major nutrients of the culture medium (Path #1: 14.52/15.19 kg_{CO2 eq} g_S⁻¹; Path #2: 8.69/9.14 kg_{CO2 eq} g_S⁻¹).

Moreover, the total monetary cost, including consumables and processes' energy consumption, for BDS process through Path #1 is of 8.2 € per g S, whereas for BDS process through Path #2 is 5.3 € per g S (roughly, 40% due to material consumables and 60% due to electricity consumption).

Thus, these results indicate that Path #2 design is the best BDS concept process in terms of desulphurization ability, cost, energy and GHG emissions aiming for a scale-up.

Despite the total energy consumption in this BDS process be 3-fold higher than our reference value for HDS (20.3 MJ g_S⁻¹), this value accounted a total desulphurization of the initial product (DBT instead crude oil) while HDS desulphurized the crude oil to a fuel with 500 ppm of sulphur. Moreover, in BDS process the microbial biomass produced (5.9 g L⁻¹) can also be further valorized, for example to single cell protein, biosurfactants,^39,40,43 etc., which also contributes to decrease the overall costs of the process towards its scale-up.

B. Uncertainty and sensitive analysis

Aiming for a future industrial application of BDS process through Path #2 into a refinery, three scenarios were established and analyzed towards a more cost-effective process. The different scenarios are described in Fig. 6, which presents the results for the energy consumption, GHG emissions and costs associated to each hypothetical scenario.

Fig. 6 Comparison of different scenarios for BDS process through Pathway #2 in terms of final energy consumption (MJ g_S⁻¹), GHG emissions (kg_{CO2 eq} g_S⁻¹) and costs (€ per g S). Scenario 1: 100% renewable energy for electricity generation, FED ∼ 0; scenario 2: heating recovered from other refinery process; scenario 3: use of sucrose instead of SBM; optimum: combines scenarios 1, 2 and 3 in the same process design. All these scenarios suppose a 99.6% SRE.

From the comparison of the data presented in Fig. 6, it is highlighted that the use of SBM, a cheaper agroindustrial residue, as the C-source for the biocatalysts production during BDS process reveals to be in overall more expensive than if commercial sucrose was used as the C-source (scenario 3). In fact, despite of sucrose be 3-fold more expensive than SBM its use for BDS can reduce considerably the overall energy consumption, GHG emissions and costs associated to the process, since the prior C-source pretreatment phase is avoided. The use of sucrose instead SBM (scenario 3) led to a decrease in the final energy consumption from 64.1 to 28.1 MJ g_S⁻¹ and consequently to a reduction of total cost from 3.8 to 2.9 € per g S. The BDS steps requiring heat could potentially use recovered heat from other industrial processes in the refinery, which can contribute to lower significantly the overall process energy consumption, GHG emissions and costs (scenario 2). Moreover, the GHG balance can be mitigated by using electricity from renewables (scenario 1). The use of 100% of renewable energy sources is beneficial regarding the GHG emissions, since the majority of energy inputs are electricity based. Thus, for a BDS scale-up in an oil refinery it must be considered the use of sucrose as C-source, the use of heating recovered from other refinery process in some steps (e.g. incubation for bacterial growth; culture medium sterilization) and the use of 100% of renewable energy whenever possible to mitigate the GHG emissions, simultaneously. This “optimum scenario” was added in Fig. 6, from which it can be observed that a final energy of 4.20 MJ with an associated cost of 1.06 € is needed per g of S removed.

Prior studies have already highlighted the ability of G. alkanivorans strain 1B to desulphurize several thiophenic organosulphur compounds, such as benzothiophene, DBT, DBT-sulfone, 4-mDBT and 4,6-dmDBT, either pure or in a model oil mixture.^32,55 However, the bacterial growth in field conditions may be significantly different leading to different desulphurization yields. For example, Bhatia and Sharma³⁰ reported a change from 92% sulphur removal ability to 26.38–71.42%, for the bacteria Pantoea agglomerans D23W3, depending on the crude oil for the same ambient temperature and desulphurization hours. In this context, it was also assayed an hypothetical worst case scenario in which it was considered an ability of only 26% of G. alkanivorans strain 1B desulphurize crude oil, i.e. a 26% SRE from crude oil instead the 99.6% SRE demonstrated for DBT desulphurization through Path #2. Therefore, in this scenario with a lower SRE but maintaining the operation conditions above described for the “optimum scenario”, consequently designated as “optimum – 26% SRE”, zero GHG emissions are expected, but the overall final energy consumption and associated cost increase to 16.15 MJ g_S⁻¹, and 4.06 € per g S, respectively (i.e. 3.8-fold of increase relatively to optimum – 99.6% SRE scenario in Fig. 6).

Based in results predicted both for “optimum – 26% SRE” and “optimum – 99.6%” scenarios, an evaluation of the benefits of include a BDS – Path #2 – optimum unit upstream or downstream of HDS units (Fig. 3) into a conventional petroleum refinery, such as that described in Energetics Incorporated⁴⁸ refining a crude oil containing 1.26% of S, or of use the BDS – Path #2 – optimum as an eco-friendly alternative process to HDS was further performed. In Table 4 are presented the values estimated for the overall energy consumption and cost analysis towards the desulphurization per L of crude oil or HDS' diesel (hydrodesulphurized diesel) associated to these three options. From this table, it is evident that the most cost-effective option to produce ULSD is the coupling of BDS downstream HDS. This combination, needing only an extra of 1.73–6.65 MJ per L HDS' diesel for the BDS unit with a corresponding operation cost of 0.43–1.63 € per L HDS' diesel, may ensure cutting down on huge investment required for an ultra-deep HDS to generate ULSD (≤10 ppm). Thus, the downstream option for an industrial BDS application seems to be the best choice, despite in the upstream option the prior S removal from recalcitrant compounds using the BDS may also lead to some adjustments in the HDS required conditions, turning them less severe (e.g. eventual decrease in temperatures and/or pressures) and therefore less costly and less polluting to produce zero sulphur diesel. But, independently of the upstream or downstream integration option, the GHG emissions regarding this BDS process design are null in a renewables context considered for the optimum scenarios (26% or 99.6% SRE), otherwise the carbon footprint would be higher.

Table 4 Energy consumption and cost analysis for the coupling of a BDS – Path #2 – optimum unit, either upstream or downstream, into an HDS petroleum refinery and for the use of BDS – Path #2 – optimum as an alternative process to the industry's conventional technology towards the production of ULSD

Energetic needs and operational costs	Optimum scenario	BDS – Path #2 upstream HDS^c	BDS – Path #2 downstream HDS^d	BDS – Path #2 as alternative to HDS^e
a HDS desulphurized diesel (500 ppm of S).b Optimum – 26% SRE: 16.15 MJ gS^−1, 4.06 € per g S; optimum – 99.6% SRE: 4.20 MJ gS^−1, 1.06 € per g S.c Desulphurization of crude oil (1.26% of S) in a BDS unit towards 10% of S removal.d Desulphurization of HDS' diesel in a BDS unit, from 500 to 10 ppm of S.e Desulphurization of crude oil, from 12 600 to 10 ppm of S (our reference value for the energetic needs of HDS process to desulphurize a crude from 12 600 ppm to 500 ppm of S is 20.3 MJ gS^−1, which corresponds to 210.9 MJ per L crude oil).
MJ L^−1 crude oil or HDS' diesel^a	26% SRE^b	17.09	6.65	174.99
	99.6% SRE^b	4.44	1.73	45.49
€ per L crude oil or HDS' diesel	26% SRE	4.30	1.63	44.03
	99.6% SRE	1.12	0.43	11.45

---

This BDS process design as an alternative technology to the industry's solution (HDS) is still an unviable option, because despite the final energy required (45.49–174.99 MJ per L crude) be lower than for HDS (211 MJ per L crude) it implies too much investment to achieve ULSD (11.45–44.03 € per L crude), in comparison with the options of integrate BDS as a complementary technique to HDS (Table 4). Moreover, the traditional thermochemical technique is also able of remove several other compounds from crude oil besides sulphur.

In order to get more realistic cost analysis data for a coupled BDS–HDS refinery, further studies must be carried out, firstly testing real crude with G. alkanivorans to determine its SRE from crude oil and then making field studies in a pilot scale refinery.

C. Diesel taxi fleet application

Considering the option of coupling BDS – Path #2 – optimum downstream HDS into the conventional refinery described by Energetics Incorporated⁴⁸ to desulphurize a crude oil and generate an ULSD, to accomplish the imposed strict European limit for diesel S-content of 10 ppm, further calculations were carried out accounting the use of this ULSD by a taxi fleet.

In this context, for a taxi fleet of 3100 vehicles performing 207 km daily, it would be necessary the biodesulphurization of about 45 000 L of HDS' diesel (500 ppm of S) per day. So, considering the BDS – Path #2 process design within the optimum scenario, it would be necessary 4 ton of sucrose, 1% (v/v) of invertases from Z. bailii Talf1 and 2% (v/v) of G. alkanivorans 1B inoculum, to desulphurize the required amount of HDS' diesel. This post-treatment of the HDS' diesel by BDS to generate ULSD (10 ppm of S) would have a final energy consumption ranging from 26 to 299 GJ with associated costs ranging from 7000 to 20 000 €, for the “optimum – 99.6% SRE” and “optimum – 26% SRE”, respectively. Moreover, as already stated, the GHG emissions regarding this BDS – Path #2 – optimum process will be null in a renewable context. Using the ULSD, the amount of SO₂ emissions by the fleet would decrease from 32.09 kg to levels up to 0.64 kg per day.

IV. Conclusions

The LCA and LCC based methodology applied to two BDS processes using G. alkanivorans strain 1B highlighted the BDS process in a SSF approach with invertases (BDS – Path #2) as the most efficient in terms of sulphur removal ability, cost, energy and GHG emissions. For SBM as C-source and DBT as S-source, G. alkanivorans was able to reduce 99.6% of S-content consuming 64.1 MJ g_S⁻¹ of final energy with GHG emissions of 9.14 kg_{CO2 eq} g_S⁻¹ and costs associated of 3.8 € per g S. Despite SBM be 3-fold cheaper than the commercial sucrose, the pretreatment required for the use of this agroindustrial residue as C-source for BDS overcomes this benefit. So, taking into account a scale-up scenario with sucrose as C-source, renewables for electricity and heating waste recovering (optimum scenario), the overall energy, GHG emissions and costs associated decrease significantly to 4.2 MJ g_S⁻¹ S, ∼0 kg_{CO2 eq} g_S⁻¹ and 1.06 € per g S, respectively. This final energy value is roughly a quarter of the reference value for HDS (20.3 MJ g_S⁻¹), nevertheless the BDS – Path #2 – optimum process doesn't imply an increase of the carbon footprint, since it mitigate the GHG emissions.

BDS process cannot be considered yet an industrial alternative to the HDS, nevertheless it can be an advantageous eco-friendly complementary technique to HDS towards ultra low sulphur fuels. The current study points out for the application of BDS – Path #2 – optimum downstream HDS as the best cost-effective conceptual design to apply into an oil refinery. Once it is able of desulphurize HDS recalcitrant compounds selectively, BDS integration may led to the accomplishment of the stringent European limit of <10 ppm for S-content on fuels, which otherwise may imply the necessity of more severe conditions within HDS units. Deep desulphurization is a very costly option and is not environmental friendly because implies higher GHG emissions and substantially increase the carbon footprint.

However, further scale-up studies in a pilot plant are necessary prior to the integration of BDS technology on a real refinery towards an industrial cost-effective and less polluting combined technology.

Abbreviations

2-HBP	2-Hydroxybiphenyl
BDS	Biodesulphurization
CED	Cumulative energy demand
DBT	Dibenzothiophene
FC	Fuel consumption
FED	Fossil energy demand
FU	Functional unit
GHG	Greenhouse gas
GWP	Global warming potential
HDS	Hydrodesulphurization
In.	Inoculum
LCA	Life cycle assessment
LCC	Life cycle cost
M	Molar mass
O.I.	Orbital incubator
SBM	Sugar beet molasses
SRE	Sulphur removal efficiency
SSF	Simultaneous saccharification and fermentation
ULS	Ultra low sulphur
ULSD	Ultra low sulphur diesel

Acknowledgements

This work was funded by FEDER funds through POFC-COMPETE and by national funds through Fundação para a Ciência e Tecnologia (FCT) in the ambit of the project Carbon4Desulf (FCOMP-01-0124-FEDER-013932) of LNEG. This work was also supported by Fundação para a Ciência e a Tecnologia (FCT), through IDMEC, under LAETA-UID/EMS/50022/2013. Ana F. Ferreira is pleased to acknowledge the FCT for the provision of the scholarship SFRH/BPD/95098/2013 and Carla Silva the FCT 2012 researcher competition (IF/00181/2012).

Notes and references

1. S. E. Murphy, S. Amin, K. Coletta and D. Hoffman, Chem. Res. Toxicol., 1992, 4, 491–495.
2. C. A. Pope III, JAMA, J. Am. Med. Assoc., 2002, 287, 1132–1141.
3. G. Mohebali and A. S. Ball, Microbiology, 2008, 154, 2169–2183.
4. K. F. Ho and S. C. Lee, Sci. Total Environ., 2002, 289, 145–158.
5. J. Bender and H.-J. Weigel, Agron. Sustainable Dev., 2011, 31, 81–89.
6. P. A. Denis, Theor. Chem. Acc., 2010, 127, 621–626.
7. M. M. Maricq, R. E. Chase, N. Xu and P. M. Laing, Environ. Sci. Technol., 2002, 36, 283–289.
8. I. V. Babich and J. A. Moulijn, Fuel, 2003, 82, 607–631.
9. G. Mohebali, A. S. Ball, A. Kaytash and B. Rasekh, Microbiology, 2008, 154, 878–885.
10. OPEC, World Oil Outlook, 2011.
11. P. Derikvand, Z. Etemadifar and D. Biria, Int. Biodeterior. Biodegrad., 2014, 86, 343–348.
12. S. Mehran, B. Amarjeet and M. Argyrios, Biotechnol. Adv., 2007, 25, 570–596.
13. D. Duissenov, Master in Science Dissertation, Faculty of Engineering Science and Technology, Norwegian University of Science and Technology, Trondheim, Norway, 2013.
14. D. D. Whitehurst, T. Isoda and I. Mochida, Adv. Catal., 1998, 42, 345–471.
15. R. Shafi and G. J. Hutchings, Catal. Today, 2000, 59, 423–442.
16. R. Javadli and A. de Klerk, Appl. Petrochem. Res., 2012, 1, 3–19.
17. V. Chandra Srivastava, RSC Adv., 2012, 2, 759–783.
18. I. Mochida and K. H. Choi, J. Jpn. Pet. Inst., 2004, 47, 145–163.
19. B. H. Alsolami, Ph. D. Thesis, Delft University of Technology, The Netherlands, 2012.
20. M. J. Grossman, M. K. Lee, R. C. Prince, V. Minak-Bernero, G. N. George and I. J. Pickering, Appl. Environ. Microbiol., 2001, 67, 1949–1952.
21. M. Q. Wang, D. Kodjak and H. Cai, J. Automot. Saf. Energy, 2011, 2, 271–293.
22. G. A. Adegunlola, J. K. Oloke, O. N. Majolagbe, E. A. Adebayo, C. O. Adegunlola, A. G. Adewoyin and F. O. Adegunlola, Eur. J. Exp. Biol., 2012, 2, 400–403.
23. M. Mužic and K. Sertić-Bionda, Chem. Biochem. Eng. Q., 2013, 11, 101–108.
24. Z. X. Jiang, H. Y. Lü, Y. Zhang and C. Li, Chin. J. Catal., 2011, 32, 707–715.
25. J. T. Sampanthar, H. Xiao, J. Dou, T. Y. Nah, X. Pong and W. P. Kwan, Appl. Catal., A, 2006, 63, 85–93.
26. D. J. Monticello, Curr. Opin. Biotechnol., 2000, 11, 540–546.
27. S. Labana, G. Pandey and R. K. Jain, Lett. Appl. Microbiol., 2005, 40, 159–163.
28. J. J. Kilbane, Curr. Opin. Biotechnol., 2006, 17, 305–314.
29. N. Gupta, P. K. Roychoudhury and J. K. Deb, Appl. Microbiol. Biotechnol., 2005, 66, 356–366.
30. S. Bhatia and D. K. Sharma, Biochem. Eng. J., 2010, 50, 104–109.
31. L. Alves, M. Melo, D. Mendonça, F. Simões, J. Matos, R. Tenreiro and F. M. Gírio, Enzyme Microb. Technol., 2007, 40, 1598–1603.
32. L. Alves, S. Marques, J. Matos, R. Tenreiro and F. M. Gírio, Chemosphere, 2008, 70, 967–973.
33. S. Aggarwal, I. A. Karimi and D. Y. Lee, FEMS Microbiol. Lett., 2011, 315, 115–121.
34. S. Aggarwal, I. A. Karimi and G. R. Ivan, Mol. Biosyst., 2013, 9, 2530–2540.
35. S. M. Paixão, P. D. Teixeira, T. P. Silva, A. V. Teixeira and L. Alves, New Biotechnol., 2013, 30, 598–606.
36. A. A. Nuhu, Rev. Environ. Sci. Biotechnol., 2013, 12, 9–23.
37. S. Bhatia and D. K. Sharma, Environ. Sci. Pollut. Res., 2012, 19, 3491–3497.
38. P. Derikvand, Z. Etemadifar and D. Biria, Microbiology, 2015, 84, 65–72.
39. S. Bandyopadhyay, R. Chowdhury and C. Bhattacharjee, Green Chem. Lett. Rev., 2014, 7, 288–295.
40. S. Bandyopadhyay, R. Chowdhury, C. Bhattacharjee and S. Pan, Fuel, 2013, 113, 107–112.
41. J. Yang and I. W. Marison, Biochem. Eng. J., 2005, 27, 77–82.
42. M. Mukhopadhyay, R. Chowdhury and P. Bhattacharya, AIChE J., 2007, 53, 2188–2197.
43. J. Raheb and M. J. Hajipour, Energy Sources, Part A, 2011, 33, 2113–2121.
44. M. Shavandi, M. Sadeghizadeh, A. Zomorodipour and K. Khajeh, Bioresour. Technol., 2009, 100, 475–479.
45. M. Soleimani, A. Bassi and A. Margaritis, Biotechnol. Adv., 2007, 25, 570–596.
46. L. Alves and S. M. Paixão, Appl. Biochem. Biotechnol., 2014, 172, 3297–3305.
47. World Resources Institute and W. B. C. for S. Development, Product Life Cycle Accounting and Reporting Standard, 2013.
48. Energetics Incorporated, Energy Bandwidth for Petroleum RefiningProcesses, U. S., 2006.
49. M. Frischknecht, R. Althaus, H. J. Doka, G. Dones, R. Heck, T. Hellweg, S. Hischier, R. Jungbluth, N. Nemecek, T. Rebitzer and G. Spielmann, Overview and methodology, final report ecoinvent v2.0 no. 1, 2007.
50. F. Associates, Cradle-to-Gate Life Cycle Inventory of Nine Plastic Resins, 2011.
51. National Energy Technology Laboratory, An Evaluation of the Extraction, Transport and Refining of Imported Crude Oils and the Impact on Life Cycle Greenhouse Gas Emissions, 2009.
52. EDP, A World Full of Energy, 2013.
53. R. Frischknecht, N. Jungbluth, H.-J. Althaus, G. Doka, R. Dones, T. Heck, S. Hellweg, H. Roland, T. Nemecek, G. Rebitzer and M. Spielmann, Int. J. Life Cycle Assess., 2004, 1–7.
54. A. F. Ferreira, J. Ortigueira, L. Alves, L. Gouveia, P. Moura and C. Silva, Bioresour. Technol., 2013, 144, 156–164.
55. L. Alves, R. Salgueiro, C. Rodrigues, E. Mesquita, J. Matos and F. M Gírio, Appl. Biochem. Biotechnol., 2005, 120, 199–208.

---