Journal of Cleaner Production 200 (2018) 269e281 Contents lists available at ScienceDirect Journal of Cleaner Production journal homepage: www.elsevier.com/locate/jclepro Lifetime oriented design of natural gas offshore processing for cleaner production and sustainability: High carbon dioxide content Luiz de Medeiros a, Giovani Cavalcanti Nunes b, Alessandra de Carvalho Reis a, Jose lia de Queiroz Fernandes Araújo a, * Ofe a b Federal University of Rio de Janeiro, Brazil State University of Rio de Janeiro, Brazil a r t i c l e i n f o a b s t r a c t Article history: Received 5 February 2018 Received in revised form 19 July 2018 Accepted 27 July 2018 Available online 30 July 2018 Production of natural gas in deepwaters with high gas-to-oil ratio and high carbon dioxide (CO2) content challenges the design of offshore processing due to area and weight limitations. Furthermore, cleaner production and process sustainability impose sending the separated CO2 to early enhanced oil recovery, which has economic beneﬁt but gradually increases %CO2 in raw gas, paralleled by decaying oil and gas ﬂowrates. These conditions favor CO2 capture by membrane permeation (MP) for bulk removal and chemical absorption (CA) for polishing removal. Hybrid MP-CA has greater ﬂexibility to face varying production and %CO2, demanding lifetime-oriented process design. CO2 production proﬁle is estimated adopting %CO2 retained in source rock (0%, 60%) and gas ﬂowrate predicted by empirical production decline curves. Under transient gas production MP area and operational conditions are optimized via non-linear programming at ﬁve points of process lifetime constrained by %CO2 in injected ﬂuid above 75%mol. Treated gas reaches sale speciﬁcation (%CO2 < 3%mol) in the CA unit placed downstream MP. The obtained best design matched targets, but was more impacted by decreasing ﬂowrate of raw gas than by increasing %CO2. © 2018 Elsevier Ltd. All rights reserved. Keywords: Lifetime-oriented design Reservoir decline curve CO2 capture EOR Process optimization Membrane permeation 1. Introduction The carbon budget is the cumulative amount of CO2 in the Abbreviations: Bbl, Barrels of Petroleum Liquids (1 bbl ¼ 0.159 m3); BBL, Billion Barrels of Petroleum Liquids; CO2, Carbon Dioxide; CA, Chemical Absorption; CAPEX, Capital Expenditures; CEPCI, Chemical Engineering Plant Cost Index; COMP, Compressor; E&P, Exploration and Production; ENG, Exported Natural Gas; EOR, Enhanced Oil Recovery; EROI, Energy Return over Invested Energy; E-IG, Energy fraction associated to injection gas; E-RNG, Energy fraction associated to raw natural gas; FPSO, Floating Production Storage and Ofﬂoading; GAMS, General Algebraic Modeling System; GOR, Gas to Oil Ratio; HC, Hydrocarbon; HCDPA, Hydrocarbon Dew-Point Adjustment; HRWH, Heat Recovery Water Heater; IG, Injection Gas; JT, Joule-Thompson; LCC, Life-Cycle Cost; LHV, Lower Heating Value; MDEA, MethylDiethanolamine; MP, Membrane Permeation; MUSD, Million United States Dollars; NG, Natural Gas; NLP, Non-Linear Programming; O&G, Oil & Gas; OPEX, Operational Expenditures; PHW, Pressurized Hot Water; PZ, Piperazine; RNG, Raw Natural Gas; RS, Response Surface; SG, Storage Gas; sm3, standard m3; USD, United States Dollar; WDPA, Water Dew-Point Adjustment. * Corresponding author. E-mail addresses: [email protected] (A. de Carvalho Reis), [email protected] (J.L. de Medeiros), [email protected] (G.C. Nunes), [email protected] (O.Q.F. Araújo). https://doi.org/10.1016/j.jclepro.2018.07.271 0959-6526/© 2018 Elsevier Ltd. All rights reserved. atmosphere corresponding to 450 ppm. For a 50% chance of keeping average global warming below 2 C by 2050, the budget is estimated to be approximately 275 Gt of carbon (1008 Gt CO2) (IPCC, 2014a). To limit emissions rate, carbon taxes and cap-andtrade mechanisms are expanding worldwide (Energy Institute at Haas, 2016), intensifying the risk of “stranded assets” (Carbon Tracker, 2017). Consequently, three quarters of proven reserves of coal, oil, and natural gas may be unburnable (IPCC, 2014b), contributing to reduce life expectancy of Oil & Gas (O&G) business. The Organization of Petroleum Exporting Countries acknowledges that the O&G industry could be overinvesting, building excess capacity (Musarra, 2017), while to achieve the expected return out of capital expenditure (CAPEX) production life needs to be extended. Although proven oil reserves are expanding as offshore exploration and production (E&P) is increasingly moving to remote areas and deeper waters thanks to unstoppable development of offshore E&P (Exploration and Production) technology, the easy oil era has come to an end (ironically, not because reserves are drying). Gerasimchuk et al. (2017) named “zombie energy” the production from these ﬁelds that, although receiving strong government subsidies (“negative carbon taxes”), will remain unburned. However, 270 A. de Carvalho Reis et al. / Journal of Cleaner Production 200 (2018) 269e281 Nomenclature A, A/FEED Permeation area (m2) and permeation area per feed unit (m2/Nm3/h) b Arps' exponent d FPSO working days per y (d) DI Reservoir nominal decline rate (y1) Ei Annual average molar ﬂow of ENG from each FPSO (106 sm3/d) f Objective function in NLP optimization F1 RS input factor #1 CH4 or CO2 feed partial pressure F2 RS input factor #2 CO2 or CH4 feed partial pressure F3 RS input factor #3 area per feed unit (m2/Nm3/h) F3i RS input factor #3 area per feed unit for each stage (m2/Nm3/h) F4 RS input factor #4 permeate pressure (bar) F5 RS input factor #5 feed temperature (K) FFEEDi Feed molar ﬂow rate for each stage (106 sm3/d) Fi Annual average molar ﬂow of RNG fed to each FPSO (106 sm3/d) the installed capacity “locks in” fuel dependency, as they are intensive in capital and long lived, hence production will last to return investments, slowing the transition to lower-carbon energy (Gerasimchuk et al., 2017). Although the extension of subsidies can be argued, the fact is that remote oil and gas reserves pose a general decline in energy return of energy invested (EROI) (Hall et al., 2014), as more energy is demanded for E&P operation, enforcing the need for sustainability-oriented design of E&P. It is worth noting that the Brazilian Pre-Salt oil reserves are distant from shore (>340 km), in ultra-deepwaters (>2000 m) (Gaffney et al., 2010), with high Gasto-Oil Ratio (GOR) e greater than 400 standard m3 of gas/m3 of oil (sm3/m3) e and with association to CO2-rich gas ~44 %mol (Arinelli et al., 2017). Floating Production Storage and Ofﬂoading (FPSO) platforms have been utilized in remote offshore areas without infrastructure for many years but grew in importance with the push by offshore industry into ever deeper waters (Shimamura, 2002). FPSO are preferred for being mobile, self-sufﬁcient, with high storage capacity and without the need of local piping infrastructure for oil transport. Compared to ﬁxed platforms, FPSOs offer the advantages of being more rapidly developed, requiring lower initial investment, and keeping their aggregate value for longer time, since they can be reallocated to other ﬁelds, and having lower abandonment costs (Yu et al., 2015). With huge deepwater reservoirs, Brazilian Pre-salt impacted the FPSO scenario. Currently, 178 FPSOs (90 contractor owned and 88 operator owned) are operating worldwide (44 in Brazil) being 126 converted vessels. Additionally, 19 are not working but are available for redeployment (2 in Brazil) and 12 are on order (8 in Brazil), totaling 209 FPSOs (Barton et al., 2017). These FPSO and deep offshore projects demand very sizeable investments that must be optimized and protected (Thiabaud et al., 2011). In fact, for extending use of fossil energy beyond 2050, increased energy efﬁciency is sought in E&P along with minimization of CO2 emissions. It is relevant to the present study that oil production is dependent on the capacity of processing the associated high CO2-rich natural gas (NG). In offshore processing of CO2-rich NG, the main destination for the separated CO2 is Enhanced Oil Recovery (CO2EOR). The CO2 storage potential in EOR is high: 60% of injected CO2 can be retained in the reservoir (Gazalpour et al., 2005). CO2 Equipment foot-print (m2) Annual gas hold-up in the reservoir (106 sm3) Annual average molar ﬂow of IG from each FPSO (106 sm3/d) L MP permeate phase n Number of FPSOs connected to the reservoir PPCH4, PPCO2 Feed CH4 and CO2 partial pressures (bar) PPERM Permeate pressure (bar) qI Initial production ﬂow rate (106 sm3/d) q(t) Production ﬂow rate (106 sm3/d) RECLCH4 RS response #1 CH4 %recovery in permeate RECLCO2 RS response #2 CO2 %recovery in permeate Si Annual average molar ﬂow of SG (106 sm3/d) t Time TFEED Feed temperature (K) x0CO2, xi Reservoir CO2 molar fractions: initial and for y i xENGCO2, xIGCO2 Annual average CO2 molar fractions: for ENG and for IG y year FP Hi Ii reinjection reduces oil density and viscosity, improving its ﬂuidity and increasing reservoir production, monetizing CO2. CO2-EOR recovers 1e3 bbl (barrels) of oil per injected ton of CO2, increasing, thus, the reservoir economic lifetime (Luu et al., 2016). High gas-to-oil ratio (GOR), associated with high CO2 content (Gaffney et al., 2010), challenges the design of FPSOs, due to the impact in area and weight required by the NG processing plant (Andrade et al., 2015). In this case, uncommon steps are needed, such as Water Dew Point Adjustment (WDPA) and Hydrocarbon Dew Point Adjustment (HCDPA), efﬁcient H2S and CO2 removal and high-pressure CO2 reinjection (Formigli Filho et al., 2009). Fig. 1 shows the gas processing steps on the topside of Brazilian Pre-Salt deepwaters FPSO, highlighting CO2 separation while illustrating the ﬂow proﬁle and increasing CO2 content in the reservoir along operating lifetime. The mixed oil, gas and water stream arrives in the FPSO through risers and proceeds to threephase separation, from where each fraction is directed to its treatment. The gas stream is compressed and dehydrated for WDPA avoiding hydrate formation in the transportation pipeline. Next, NG is sent to HCDPA to remove heavier fractions. The gas then proceeds to CO2 separation via membrane permeation (MP), where it is split into permeate e a CO2-rich stream compressed to be dispatched as Injection Gas (IG) e and retentate, a CO2-poor stream compressed and exported to onshore facilities as NG for sale. Compression for dispatching treated NG and CO2-rich ﬂuid for EOR challenge energy and area availability on FPSO topside. Composition speciﬁcation of the CO2-rich stream to be reinjected is also an important design premise. From an environmental point of view CO2-EOR is beneﬁcial because it allows for the storage of part of the CO2 injected while increasing oil recovery (Kwak et al., 2014) - 50% according to the authors. Gazalpour et al. (2005) suggests a “gross” CO2eretention efﬁciency of approximately 60% at CO2 breakthrough if separation and reinjection is not considered after the breakthrough. The unretained CO2 is responsible for a steady rise in CO2 content of produced NG along reservoir operation lifetime, concomitantly to the decrease in NG production due to depletion. This extreme scenario challenges the design of offshore NG processing plants, mainly concerning CO2 separation, demanding advancements in FPSO design for enhanced production (Islam et al., 2012). In fact, sustainability of NG processing e and its survival as A. de Carvalho Reis et al. / Journal of Cleaner Production 200 (2018) 269e281 271 Fig. 1. Typical NG processing on FPSO topside in Brazilian Pre-Salt. energy source in a foreseen low carbon future e depends heavily on CO2 management technologies (Araujo et al., 2017), especially the ones related to its separation and destination. In this sense, optimized design of NG processing plants plays a central role in utilization of NG under stringent speciﬁcations. Most studies reported in the literature propose plants designed at speciﬁc operating point (sizing case) of the production curve, i.e., at one single period of the process lifetime. For instance, Araújo et al. (2017) compared CO2 separation alternatives in ultra-deep waters in technical, economic and environmental terms, facing constant gas ﬂow rate feeds (6,000,000 sm3/d) with three CO2 content scenarios (10%, 30% and 50% mol), considering EOR as CO2 destination. Lifetime was used solely in the economic analysis (20 y). For feeds with higher CO2 content (50% CO2), Araújo et al. (2017) indicated technical advantages of hybrid MP þ CA process consisting of MP for bulk CO2 removal and Chemical Absorption (CA) for polishing. The authors highlighted that, even in drastic conditions, MP þ CA overcome pure CA and pure MP designs. Reis et al. (2017) targeted optimization of MP þ CA under two types of constraints. The ﬁrst forced treated gas to comply with sale speciﬁcation (%CO2 < 3% mol) resulting in single MP process with signiﬁcant hydrocarbon loss (HC Losses) in the injected gas. The second imposed injected gas to a minimum of 75% mol CO2, requiring a CA polishing unit. The authors showed that the hybrid MP þ CA offers more ﬂexibility than MP alternatives. Reis et al. (2017) also used constant gas ﬂowrate with three cases of CO2 content. Kim et al. (2017) approached topside equipment for optimum use of space, but process engineering was an a priori task outside their scope. Thiabaud et al. (2011) focused on key engineering issues in FPSO design and claimed the relevance of lifecycle simulator e a virtual plant used throughout engineering and operational phases, for transients associated to slug ﬂow. The slow lifecycle transient due to production decline (Arps, 1945) and its impact in the topside plant was not covered by the authors. Gallo et al. (2017) studied the energy use of a Brazilian Pre-salt FPSO platform and concluded that the need for gas compression represents the main use of fossil energy (between 38% and 50% of the total energy consumption). Due to the high %CO2 in the gas, it requires CO2 removal, compression and reinjection in the oil ﬁeld, increasing energy needs. The authors estimated a total production curve, with production peak at 7.5 y and with a total life time of 25 y. Their aim was to evaluate power usage along lifecycle (at chosen points in the production curve), for ﬁxed FPSO design, concluding that, for most part of the operation, the power generation system was oversized. The work did not explore process design. Barrera et al. (2015) presented another exergy analysis of a Brazilian FPSO. Their ﬁndings indicated the highest energy loss associated to the exhaust gases from gas turbines and the gas injection system as the second highest exergy sink. Araújo et al. (2017) explored the process design gap in the literature and the loss of thermal energy from gas turbines to supply heat for solvent regeneration in amine-based CO2 separation. A set of performance indicators, under the premises adopted, favored hybrid capture e membrane permeation (MP) for bulk CO2 separation followed by chemical absorption (CA) as a CO2 removal polishing step. However, the service split between MP and CA was not explored to improve process performance. Filling this gap, Reis et al. (2017) optimized the hybrid arrangement of Araújo et al. (2017), with a ﬁxed gas production (6,000,000 sm3/d) and different content of CO2 in the raw gas (10%, 30% and 50% mol CO2). Optimization used a response surface (RS) model to represent MP separation under two types of constraint e a threshold on maximum CO2 content in treated gas, and a threshold of minimum CO2 content in the injected CO2-rich gas. The later type provided the optimum split of CO2 separation service between MP and CA units. Reis et al. (2017) did not include in their analysis the production curve, which presents variations in gas production and, due to CO2 injection, CO2 content increases along production lifecycle. The present work targets sustainability of long lived FPSO, aiming at designs with CO2 reinjection, which impacts CO2 content in produced gas along project lifetime. Furthermore, production cycle is accelerated, reducing project expectancy to 20 y, to face carbon budget constraints, differently from the 25 y production lifetime adopted by Gallo et al. (2017). For the relevant technology niche of offshore processing of CO2-rich NG, the main literature gap targeted by this work consists in evaluating process performance in a dynamic scenario of gas production decline, along with increasing CO2 content in the produced gas resulting from CO2-EOR. Specifically, this work approaches the impact of varying feed conditions (ﬂowrate and composition) on FPSO design. Due to the ﬂexibility of the hybrid process composed by MP and CA units, as defended by Reis et al. (2017), this technology route was selected as CO2 separation pathway. It was evaluated in terms of Capital Expenditures (CAPEX), footprint and HC losses on a time-varying scenario of feed ﬂow rate and CO2 content. 272 A. de Carvalho Reis et al. / Journal of Cleaner Production 200 (2018) 269e281 Finally, CO2 management is the road for sustainability of fossilbased energy. Processing CO2-rich NG demands capital and energy intensive upstream processes that lower EROI of NG exported to onshore facilities, reinforcing the need of careful gas process engineering to face the stringent offshore scenario and move towards sustainability with constrained carbon budget. Although the FPSO produces oil besides associated NG, the focus is on the gas processing plant, which, in the Pre-salt applications occupy nearly 60% of the deck area and manages large amount of CO2. For instance, in the period 2010e2016, 4,600,000 t of CO2 was reinjected in Lula and Sapinho a ﬁelds (Petrobras, 2017). 2005). Fig. 3 is a simpliﬁed representation of the four streams, considering the systems boundaries: raw NG (RNG), which is processed in FPSOs, generating NG stream to be exported (ENG), the CO2-rich stream for EOR injection (IG) and storage gas stream (SG, pure CO2 captured into source rocks). The construction of the two CO2 content proﬁle (for 0% and 60% of injected CO2 stored in reservoir rock) adopts reservoir constant pressure and GOR along process lifetime. Additionally, the raw NG stream (RNG) is considered water and sulfur free, and the condensate ﬂow rate generated in the HCDPA unit (Joule-Thomson unit) is added to the oil produced, which is not represented in Fig. 3. 2. Background information 2.2.1. Scenario 1e0% CO2 storage The reservoir is modeled as a perfectly mixed tank. Hence, the reservoir CO2 content in a given y corresponds to the CO2 content in the raw NG produced in this same y. The approach assumes that the variables used in the balances are constant throughout a y, and that the average production molar ﬂow for a given y is the arithmetic average of two consecutive yearly values (present and previous y). Eqs. (4) and (5) describe reservoir balances, while Eqs. (6) and (7) refer to balances of a FPSO. Eqs. (4) and (5) correspond to annual reservoir balances of gas and CO2, respectively. Certain infra-structure concepts characteristic of oil and gas E&P are necessary to develop the analysis in Sec. 3. These concepts are deﬁned and formalized in this section. 2.1. Gas production curve Arps (1945) presents empirical production decline curves that remain widely used in predicting production proﬁle for oil and gas E&P. Arps proposes three decline proﬁles depending on time (t, y), initial production rate (qI, 106 sm3/d), reservoir nominal decline rate (DI, y1) and Arps' exponent (b). According to this last parameter, proﬁles are classiﬁed as exponential (b¼0), hyperbolic (0 < b<1) and harmonic (b¼1). Eqs. (1)e(3) present, respectively, the rate expressions for these three kinds of proﬁles. q(t) ¼ qI ∙ exp (- DI ∙ t) q(t) ¼ qI / ((1 þ b ∙ DI ∙ t) (1) 1/b ) q(t) ¼ qI / (1 þ DI ∙ t) Hiþ1 ¼ Hi e Fi ∙ d ∙ n þ Ii. ∙ d ∙ n (4) xiþ1 ∙ Hiþ1 ¼ xi ∙ Hi e xi ∙ Fi ∙ d ∙ n þ xIGCO2 ∙ Ii ∙ d ∙ n (5) Fi ¼ Ei þ Ii (6) xi ∙ Fi ¼ xENGCO2 ∙ Ei þ xIGCO2 ∙ Ii (7) (2) (3) Decline curves are normally constructed to predict oil production. However, if gas to oil ratio (GOR) is considered constant along the reservoir lifetime, it is expected that the gas production presents a proﬁle proportional to the oil production proﬁle. With this premise, Fig. 2 illustrates Arps decline curves representing the gas production decline of a FPSO. 2.2. Curve of CO2 content in the reservoir To represent the increase of CO2 content in a reservoir in which the reinjection practice is used along its entire lifetime, global mass balance and CO2 balance are herein proposed focusing solely on the produced NG. Two scenarios are considered: (a) CO2 storage in the source rock does not occur (0% CO2 storage); and (b) 60% of injected CO2 is retained in the reservoir (60% CO2 storage) (Gazalpour et al., Fig. 2. Arps decline curves for gas production of a FPSO. Fig. 3. Representation of a cluster of FPSOs with EOR: (a) 0% CO2 storage; (b) 60% CO2 storage [RNG ≡ Raw NG, IG ≡ Injection Gas, ENG ≡ Exported NG, SG ≡ Storage Gas (pure CO2)]. A. de Carvalho Reis et al. / Journal of Cleaner Production 200 (2018) 269e281 where Hi is annual gas hold-up in the reservoir (106 sm3); Fi is annual average molar ﬂow (106 sm3/d) of raw NG (RNG) fed to each FPSO; Ii is annual average molar ﬂow (106 sm3/d) of injection gas (IG) from each FPSO; Ei is annual average molar ﬂow (106 sm3/d) of exported NG (ENG) from each FPSO; xi is annual average CO2 molar fraction in the reservoir; xIGCO2 is annual average CO2 molar fraction in the injection gas (IG); xENGCO2 is annual average CO2 molar fraction of exported NG (ENG); d is number of FPSO working days per y (d); and n is the number of FPSOs connected to the reservoir. 2.2.2. Scenario 2e60% CO2 storage The only differences between Scenario 2 and Scenario 1 are the reservoir balances, which now have ﬂow rates of CO2 captured into the source rocks as shown in Eqs. (8) and (9). Si represents annual average molar ﬂow rate (106 sm3/d) of pure CO2 captured into source rock (SG), where Si is proportional to Ii according to the coefﬁcient of injected CO2 retained in source rocks (60%). Hiþ1 ¼ Hi e Fi ∙ d ∙ n þ Ii ∙ d ∙ n e Si ∙ d 3. Methods This section presents the premises and equations used in this work. 3.1. Gas production curve The reservoir under analysis mimics the Libra ﬁeld in the Brazilian Pre-Salt, whose E&P plan predicts nine FPSOs in operation (Gaffney et al., 2010). The Replicant FPSO (a replicated standard design) is considered, with maximum gas production capacity of 6,000,000 sm3/d (Andrade et al., 2015). For the construction of the gas production curve shared by each FPSO, it is considered that the maximum production is steeply reached by the end of the ﬁrst y, remaining constant for ﬁve y at the “Production Plateau” and declining hyperbolically (0 < b < 1) from the sixth y onwards. Table 1 presents the premises for constructing the gas production proﬁle of a FPSO. 3.2. Curve of reservoir CO2 content Table 2 shows the parameters and values adopted for both scenarios. 3.3. Process description Raw NG from the dehydration unit is pressurized to 50 bar, sent to the HCDPA and to the membrane permeation (MP) unit. The MP unit produces a low-pressure CO2-rich permeate and a highTable 1 Premises for estimation of gas production curve. Parameter Value b DI (y1) Maximum production (106 sm3/d) Production plateau duration (y) Project lifetime (y) 0.5a 0.17b 6 5 20 a Arps' exponent, Eq. (2). Reservoir nominal decline rate (DI), Eq. (2). Table 2 Premises for estimation of reservoir proﬁle of %CO2 content. Parameter Value Hold-up (BBL) GOR (sm3/m3) Number of FPSOs x0CO2 xIGCO2 xENGCO2 d Lifetime (y) 3.65a 500b 9a 0.44c 0.9d 0.03e 350 20 a Gaffney et al. (2010). Pinto et al. (2014). GOR ¼ Gas to oil ratio. Arinelli et al. (2015). x0CO2: CO2 molar fraction in RNG at operation startup. d xIGCO2: annual CO2 molar fraction in the injection gas (IG). e xENGCO2: annual CO2 molar fraction of the exported natural gas (ENG). b c (8) xiþ1 ∙ Hiþ1 ¼ xi ∙ Hi e xi ∙ Fi ∙ d ∙ n þ xIGCO2 ∙ Ii ∙ d ∙ n - Si ∙ d(9) b 273 pressure NG retentate, which is compressed to 50 bar and cooled to 40 C to enter at the bottom of the chemical absorption (CA) column. Lean aqueous methyldiethanolamine (MDEA) - piperazine (PZ) is fed at the top, yielding a NG top stream (<3% mol CO2). The bottom stream with CO2-rich solvent is expanded and heatintegrated with the hot lean solvent, before entering the stripper for regeneration. The stripper top product is mixed with the permeate from the MP to form a CO2-rich stream, sent to compression and then injected into the reservoir for enhanced oil recovery (EOR). The cooled lean solvent receives water makeup and is pumped and cooled to 40 C returning to the CA absorber. The gas product from the absorber (<3% mol CO2) is compressed to 150 bar, cooled and sent as Exported NG (ENG). The reboiler heat duty in the stripper is supplied by pressurized hot water (PHW) at 190 C. PHW is produced by the Heat Recovery Water Heater (HRWH), which recovers heat from the hot ﬂue gas exhausted by the gas turbine. Heat from the HRWH supplies the heat demand of CA, which is considered cost-free up to the maximum heat recovery capacity of the HRWH at about 75 MW per 100 MW of power. It is worth noting that the FPSOs presently operating in the Brazilian Pre-Salt reservoirs have in average four gas turbines, three in operation and one in standby mode, totaling 100 MW of installed capacity. The power generated in the gas turbines is used to drive the compressors of sale NG (ENG to pipeline) and of CO2 to EOR (Araújo et al., 2017). Fig. 4a displays the conceptual diagram of the MP þ CA hybrid process. Equipment sizing of the MP þ CA hybrid process uses mass and energy balances obtained by simulations in Aspen HYSYS environment, except for MP module areas, which are obtained through non-linear optimization (employing GAMS - General Algebraic Modeling System), depicted in Fig. 4b. 3.4. Optimization of membrane permeation (MP) modules The optimization procedure uses MP response surface (RS) model and object function proposed by Reis et al. (2017). In this work, CH4 %Recovery in permeate (RECLCH4, %) and CO2 %Recovery in permeate (RECLCO2, %) were chosen as MP output responses against ﬁve relevant independent input factors: (a) CH4 feed partial pressure (PPCH4, bar); (b) CO2 feed partial pressure (PPCO2, bar); (c) Permeation area per unit of gas feed (A/Feed, m2/Nm3/h); (d) Permeate pressure (PPERM, bar), and (e) Gas feed temperature (TFEED, K). It is worth mentioning that the factors were ordered according to the modeled response. The RS formula for Y¼RECLCH4 (%), with Y as dependent variable, is written as a function of the independent 274 A. de Carvalho Reis et al. / Journal of Cleaner Production 200 (2018) 269e281 Fig. 4. NG processing: (a) conceptual diagram; (b) solving steps; HC¼Hydrocarbon, CA¼Chemical Absorption, MP ¼ Membrane Permeation, DP ¼ dewpoint, MDEA ¼ MethylDiethanolamine, PZ¼Piperazine. factors F1 ≡ PPCO2 (bar), F2 ≡ PPCH4 (bar), F3 ≡ A/Feed (m2/(Nm3/h)), F4 ≡ PPERM (bar) and F5 ≡ TFEED (K) according to Eq. (10), using the following parameter values: B0 ¼ 2.79258; B1 ¼ 0.958001; B2 ¼ 0.06217; B3 ¼ 8.51175; B4 ¼ 1.60924; B5 ¼ 0.00596025; B6 ¼ 0.00195653; B7 ¼ 0.00593021; B8 ¼ 0.849712; B9 ¼ 0.0119562; B10 ¼ 0.00293719; B11 ¼ 0.274505; B12 ¼ 0.00304543; B13 ¼ 0.588134; B14 ¼ 0.0355225; B15 ¼ 0.00455962. On the other hand, the RS formula for Y¼RECLCO2 (%), with Y as dependent variable, is written as a function of F1 ≡ PPCH4 (bar), F2 ≡ PPCO2 (bar), F3 ≡ A/Feed (m2/(Nm3/h)), F4 ≡ PPERM (bar) and F5 ≡ TFEED (K) in Eq. (11), with the following parameters: B0 ¼ 36.9348; B1 ¼ 3.45146; B2 ¼ 39.4261; B3 ¼ 9.76926; B4 ¼ 0.00259308; B5 ¼ 0.0368232; B6 ¼ 7.14655; B7 ¼ 0.0252322; B8 ¼ 0.0111323; B9 ¼ 0.0362618; B10 ¼ 0.000736667; B11 ¼ 0.482942; B12 ¼ 0.108566; B13 ¼ 1.18341; B14 ¼ 0.0242281; and B0 ¼ 0.00728997. Y¼B0 þB1∙F1 þB2∙F2 þB3∙F3 þB4∙F4 þB5∙F21 þB6∙F24 þ (B7∙F2 þB8∙F3 þB9∙F4 þB10∙F5)∙F1þ(B11∙F3 þ B12∙F4)∙F2 þ (B13∙F4 þ B14∙F5)∙F3 þ B15∙F4∙F5 f ¼ n X ðF3 Þi $ðFFEED Þi (12) i¼1 3.5. Performance metrics Design performances are evaluated with four metrics: lifecycle cost (LCC), process footprint, carbon footprint and hydrocarbon losses in the injection gas (IG). 3.5.1. Lifecycle cost (LCC) Araújo et al. (2017) employed LLC as an economic indicator, according to Eq. (13), where t is the project lifetime in y and E-ENG is the equivalent amount of low heating value of ENG, in GJ, exported per y by the FPSO. Since LCC is a function of CAPEX and OPEX, an analysis of these parameters is necessary. (10) Y¼B0 þB1∙F2 þB2∙F3 þB3∙F4 þB4∙F21 þB5∙F22 þB6∙F23 þ(B7∙F2 þB8∙F3 þB9∙F4 þB10∙F5)∙F1þ(B11∙F3 þ B12∙F4)∙F2 þ (B13∙F4 þ B14∙F5)∙F3 þ B15∙F4∙F5 (11) The permeation of ethane, heavier hydrocarbons and nitrogen were considered negligible. Another constraint was applied in the injected gas, imposing the content of CO2 to be greater than 75% mol. The decision variables are the values of factor F3 in the RS models of all involved membrane stages. In order to force the NLP optimization to remain within the range of adherence of RS models, lower and upper bounds were imposed on the feasible range of factor F3 in Eqs. (10) and (11) i.e. 0.4 m2/(Nm3/h) F3 2.1 m2/(Nm3/ h). The objective function is deﬁned in Eq. (12) as the total permeation area of the MP train. This objective respond to decision variables (F3)I, the permeation area per unit feed of all stages i, where the feed ﬂow rate of stage 1 (FFEED)1 (Nm3/h) is a ﬁxed parameter. More details about the optimization model are present in Reis et al. (2017). LCC ¼ (CAPEXOffshore þ t ∙ OPEX) / (t ∙ E-ENG) (13) 22.214.171.124. CAPEX. Mass and energy balances obtained by process simulation (Aspen HYSYS) are used for equipment sizing needed to estimate CAPEX of six FPSOs processing conditions (ﬁve related to the RNG evaluated feeds and the “Sizing FPSO” condition) for each scenario. CAPEX is ﬁrst estimated based on onshore plants, according to the procedure presented by Turton et al. (2012) with CPCI corresponding to 2016 reference y and using a cost intensiﬁcation factor of two applied due to the installation of facilities on FPSOs e i.e. CAPEXOffshore ¼ 2∙ CAPEXOnshore e as described in Araújo et al. (2017). Table 3 lists additional premises for economic analysis. 126.96.36.199. OPEX. OPEX was computed according to Araújo et al. (2017), who state that, in the FPSO context, the main OPEX component corresponds basically to the speciﬁed NG with 3% CO2 which is burnt for power generation to support the power demand of CO2 separation technologies. In this work, OPEX evaluation also A. de Carvalho Reis et al. / Journal of Cleaner Production 200 (2018) 269e281 275 4. Results and discussion Table 3 Economic premises. Item Premise CEPCI NG Price MDEA-PZ price MP price MP lifetime (y) d Project lifetime (y) 541.7 (2016)a USD 6/Million BTUb USD 1937.27/tb USD 50/m2c 5c 350 20 Reservoir analysis results are used to obtain RNG ﬂowrates and compositions fed to each FPSO along 20 y of operation. Next, performance metrics results of each FPSO design are computed. Detailed values of gas production curve, %CO2 of reservoir gas curve and HYSYS process ﬂow diagrams are available in Supplements A and B (Supplementary Materials). a Value obtained from Jenkins (2017). Prices of NG, Methyl Diethanolamine (MDEA) and Piperazine (PZ) obtained from Reis et al. (2017) and Araujo et al. (2017) as (0.464*1900USD/t MDEA þ 0.036*2419USD/t PZ)/0.5 for aqueous solution of 46.4% w/w MDEA and 3.6% w/w PZ. c Value obtained from H€ agg et al. (2017). 4.1. Gas production curve b Fig. 5 shows a FPSO production proﬁle along 20 y of operation, with Arps hyperbolic decline from the sixth y onwards, which provides RNG ﬂowrates along operation lifetime. considered MP replacement (5 y) and was calculated for each scenario and their 5 feed conditions, over a y of production. 188.8.131.52. Exported natural gas (ENG). To complete the LCC calculation, it is necessary to obtain the amount of ENG generated during the project lifetime (20 y). ENG was also computed in PJ (1015 J) using LHV (Lower Heating Value) for each scenario and their ﬁve feed conditions over a y of production. 3.5.2. Process footprint Calculation of footprints for the MP, CA and Compressor (COMP) skids were performed using the correlations presented in Araújo et al. (2007), displayed in Table 4. 3.5.3. Carbon footprint To compare the environmental impact caused by each scenario for the ﬁve feed conditions, the Carbon Footprint parameter was deﬁned, which evaluates the amount of CO2 emitted (t) to generate 1 MW-e of ENG over a y of production, as presented in Eq. (14). The calculation is based on data present in Araújo et al. (2017) that establishes the emission of 517 g of CO2 for the production of 1 kWe-h. Carbon Footprint (tCO2 / MWe ENG) ¼ (CO2 emitted / EeENG) 4.2. Curve of reservoir CO2 content Fig. 6 shows the CO2 content proﬁle in the reservoir along its lifetime (20 y) for the two scenarios. The ﬁrst scenario considers no occurrence of CO2 retention in the source rocks from the injected CO2. That is, all injected CO2 is added to the gas hold-up in the reservoir. For the second scenario, it is assumed that 60% of the injected CO2 is trapped in the source rock not reaching the reservoir gas hold-up. Consequently, Scenario 2 generates a proﬁle with lower CO2 content values when compared to values of Scenario 1. 4.3. RNG ﬂow rate to FPSO RNG ﬂow rate is computed at ﬁve points of the production curve: 1, 5, 10, 15 and 20 y. Figs. 5 and 6 provide FPSO feed ﬂowrates and respective CO2 content for the ﬁve points. It is considered that the increase in CO2 content occurs at the expense of equivalent reduction in the methane fraction of the RNG while the remaining components are maintained at constant fractions. Additional RNG conditions are from Reis et al. (2017) and Araújo et al. (2017). Table 5 shows RNG conditions for the ﬁve evaluated points. (14) 3.5.4. HC losses Hydrocarbon Losses (HC Losses) are indirectly measured by the energy fraction associated to raw NG (E-RNG) stream that is wasted in the injected gas (E-IG), according to Eq. (15). The LHV of the IG is used to compute HC Losses. HC Losses (%) ¼ (E-IG / EeRNG) ∙ 100% (15) Fig. 5. Gas production curve of a FPSO. Table 4 Footprints (FP) correlations of main gas processing equipment. Skid Correlation (FP in m2) Nomenclature a Skid ¼ 0:00296,A FPMP MP AMP ¼ MP area (m2) MP Compressora CAa a Araújo et al. (2017). Skid FPCompressor ¼ 0:8017,ðPowerÞ2=3 pﬃﬃﬃ Skid FPCA ¼ 18 ððDAbs þ DReg Þ= 2Þ2 Power ¼ compressor power (kW) DAbs ¼ diameter (m) of CA absorber DReg ¼ diameter (m) of CA regenerator 276 A. de Carvalho Reis et al. / Journal of Cleaner Production 200 (2018) 269e281 4.5.1. LCC The calculation of LCC requires the values of CAPEX (“Sizing” FPSO), OPEX and ENG, those last two obtained for the lifetime of the project. Fig. 6. Reservoir %CO2 versus time proﬁle. 4.4. FPSO designs Five FPSO designs were performed for each of the two scenarios, accordingly to RNG conditions presented in Table 5, totaling ten designs for gas processing on the FPSO. Each design includes skids for HCDPA, CO2 separation (MP and CA) and gas compression (IG and ENG). In both scenarios, the ﬁnal FPSO design is the assemblage of the largest equipment pieces for every skid along the ﬁve RNG feed conditions (Table 5) in each scenario: MP modules, CA units, compression skids and respective auxiliary equipment (heat exchangers and vessels included in the skids). This resulting FPSO design, for each scenario (0% and 60% CO2 storage) is denominated “Sizing FPSO” and is a ﬂexible platform able to operate at the most stringent feed ﬂow rate (y 1 to 6 in the Production Plateau) and CO2 composition condition, covering the ﬁve evaluated RNG feeds. Supplements C and D (Supplementary Materials) present simulation results for equipment sizing. Tables E12 and E13 of Supplement E (Supplementary Materials) show equipment sizing results for each feed condition and for 0% and 60% CO2 storage cases. It is worth noting that, for both cases, “Sizing FPSOs” consist of equipment sized at Points 1 and 2 at which production ﬂow rate is at its maximum value (i.e., ﬁrst and ﬁfth y). Since production ﬂow rates at these points are equal, as they deﬁne the Production Plateau, it can be concluded that “Sizing FPSO” is dictated by the largest ﬂowrate at the Production Plateau with 6,000,000 sm3/ d and CO2 content between 44.72% and 46.53% mol. 4.5. Performance metrics FPSOs performances are analyzed according to four metrics: Lifecycle Cost (LCC), process footprint, carbon footprint and hydrocarbon losses. Detailed results of performance metrics are available in Supplement F (Supplementary Materials). 184.108.40.206. CAPEX. From equipment sizing at six FPSO conditions (ﬁve related to the RNG evaluated feeds listed in Table 5 and the “Sizing FPSO”) for each scenario (0% and 60% CO2 storage), their respective CAPEX were calculated with the procedure proposed by Turton et al. (2012) including the inventories of MDEA/PZ for CA solvent. Fig. 7a shows the decreasing proﬁles of CAPEX resulting from changing conditions of ﬂow rate and CO2 content in RNG. CAPEX decline is justiﬁed by the reduction in equipment size due to the decrease of RNG ﬂow rate. To cope with this, a speciﬁc CAPEX-1 index is proposed as the CAPEX divided by the energy content of RNG, USD/MW RNG. However, since a single design is to be chosen for the FPSO operating along the entire lifetime (the “Sizing FPSO”), the implemented CAPEX is constant, independently of the CO2 content in RNG. Therefore, a speciﬁc CAPEX-2 index is proposed dividing the CAPEX of the “Sizing FPSO” (MUSD 235 for 0% CO2 storage, and MUSD 233 for 60% CO2 storage) by RNG energy content of each point in gas production curve. Fig. 7a evidences that CO2 content does not impact FPSO design, which is dominated by RNG ﬂowrate: the farther in the production curve is the sizing point used for FPSO design, the smaller the production ﬂow rate and the smaller the calculated CAPEX. Fig. 7b indicates that speciﬁc CAPEX-1 index suffers mild impact by CO2 content in RNG, conﬁrming that equipment size is mainly affected by RNG ﬂowrate. A minimum speciﬁc CAPEX-1 value is presented at the third sizing point (10 y of operation). Fig. 7c shows that the speciﬁc CAPEX-2 index is practically the same for the ﬁrst production points, since the “Sizing FPSO” is mainly composed by the FPSOs design associated to these feed conditions as the largest pieces of equipment resulted from the conditions of largest feed ﬂow rates. However, speciﬁc CAPEX-2 increases signiﬁcantly with RNG ﬂow reduction after the Production Plateau, indicating that the plant is oversized for ﬂow rates past the Production Plateau period. Differently from CAPEX and speciﬁc CAPEX-1, at the end of lifetime speciﬁc CAPEX-2 presents z14% reduction for Scenario 2 (52.68% CO2 in RNG, 1,260,000 sm3/d) in comparison to Scenario 1 (64.16% CO2 in RNG, 1,260,000 sm3/d). Speciﬁc CAPEX-2 is nearly invariant for the two scenarios at the Production Plateau (6,000,000 sm3/d, % CO2). This behavior conﬁrms that CAPEX is more impacted by feed ﬂow rate than by the CO2 content in the RNG feed. Fig. 8 illustrates the investment costs related to MP, CA and COMP skids as components of the Total CAPEX. Skid costs comprehend the equipment present: (a) MP skid includes total MP area; (b) CA skid assembles absorption and stripping columns, heat exchangers, vessels and pumps; and (c) COMP skid comprises Table 5 RNG Conditions at evaluated gas production points. Parameter Value Temperature (ºC) Pressure (bar) Point Number y RNG Flow Rate (106 sm3/d) RNG composition (%mol) 50 15 1 1 6.00 2 5 6.00 N2 0.24, C2 4.90, C3 3.44, i-C4 1.25, n-C4 2.65, i-C5 0.94, nC5 1.59, n-C6 1.56 Scenario 1 CO2 44.72 51.01 CH4 38.71 32.42 Scenario 2 CO2 44.25 46.53 CH4 39.18 36.90 3 10 3.35 4 15 1.93 5 20 1.26 58.00 25.43 49.43 34.00 61.83 21.60 51.34 32.09 64.16 19.27 52.68 30.75 A. de Carvalho Reis et al. / Journal of Cleaner Production 200 (2018) 269e281 Fig. 7. CAPEX proﬁles for ﬁve FPSO designs: (a) total CAPEX; (b) speciﬁc CAPEX-1; and (c) speciﬁc CAPEX-2 [Two scenarios: (1) 0% CO2 storage; (2) 60% CO2 storage; Speciﬁc CAPEX-1: Total CAPEX at corresponding RNG feed conditions/RNG energy content; Speciﬁc CAPEX-2: “Sizing FPSO” CAPEX (MUSD 235 for 0% CO2 storage, and MUSD 233 for 60% CO2 storage)/RNG energy content; 1 ¼1st y, 2 ¼ 5th y, 3 ¼ 10th y, 4 ¼ 15th y and 5 ¼ 20th y]. vessels, heat exchangers, drivers and compressors. Fig. 8a and c shows that COMP is the dominant CAPEX, which explain invariance of Total CAPEX upon CO2 reduction observed in Fig. 7a, while compressor power (and hence CAPEX) decreases steeply with decreasing gas ﬂowrate. In the ﬁrst y, MP and CA CAPEX's are in the same order of magnitude and correspond only to z10% of COMP CAPEX in both scenarios. MP CAPEX, similarly to COMP, is nearly invariant with CO2 content in RNG, being dominated by RNG ﬂowrate. However, as shown in Fig. 8c, CAPEX of CA is affected by both ﬂow rate and CO2 composition. CA cost decreases with increasing CO2 content in RNG feed for a given production point. This behavior results from MP, whose retentate is poorer in CO2 with increasing CO2 content in RNG as seen in the GAMS optimization results in Supplement D (Supplementary Materials). Hence, CA has lower CO2 content in its feed, demanding smaller columns and less costs. For points 4 and 5 in Fig. 8c (respectively, 61.83% and 64.16% CO2 in 277 Fig. 8. CAPEX of skids for ﬁve FPSO designs: (a) COMP CAPEX; (b) MP CAPEX; and (c) CA CAPEX [Two scenarios: (1) 0% CO2 storage; (2) 60% CO2 storage, where 1 ¼1st y, 2 ¼ 5th y, 3 ¼ 10th y, 4 ¼ 15th y and 5 ¼ 20th y, Design stands for the “Sizing FPSO”]. RNG), in Scenario 1, MP already speciﬁes NG, dismissing the need of the CA polishing step. The MP and CA costs in the “Sizing FPSO” are equal to those of Point 1 in both scenarios, since MP area and CA size are the same in the two FPSOs sized on the Production Plateau. Since this work adopts hybrid MP þ CA CO2 removal, it must be highlighted that Fig. 8b and c do not aim to compare costs of MP and CA because they perform different services, processing feeds with different %CO2. These ﬁgures also present the %CO2 of each skid feed. MP and CA are hence complementary and do not compete in this study. It is worth mentioning that, in all cases, the added costs of MP and CA represent less than 20% of Total CAPEX, with COMP imposing the main capital costs. 220.127.116.11. OPEX. OPEX as a function of time (Fig. 9), for each feed condition, shows proﬁle similar to that observed for gas production. To ﬁnd the value of OPEX over the 20 y of production, a linear regression was performed yielding OPEX as a function of gas production in Eq. (16). The parameters of the linear regression are given in Table 6. 278 A. de Carvalho Reis et al. / Journal of Cleaner Production 200 (2018) 269e281 Fig. 9. OPEX time proﬁle [Two scenarios: (1) 0% CO2 storage; (2) 60% CO2 storage; 1 ¼ 1st y, 2 ¼ 5th y, 3 ¼ 10th y, 4 ¼ 15th y and 5 ¼ 20th y]. Table 6 Regression parameters for OPEX proﬁle in Eq. (16). Table 7 Regression parameters for ENG and E-ENG proﬁles in Eq. (17). Parameter Scenario 1 Scenario 2 Parameter Scenario 1 Scenario 2 a b 0.260 8.521 0.407 8.406 a b 5.413 7.266 2.396 7.205 OPEX ¼ a þ b ∙ gas production (16) 18.104.22.168. Exported natural gas - ENG. The calculated proﬁles of ENG and E-ENG (Fig. 10) followed the same procedure for estimating OPEX, obtaining the linear regression of Eq. (17) for ENG and E-ENG with parameters shown in Table 7. ENG ¼ a þ b ∙ gas production (17) Table 8 summarizes economic analysis values, making it possible to calculate LCC, according to Eq. (13). It is observed that Scenario 2 presents a lower cost. LCC values found in this work are compatible with those of Reis et al. (2017) for MP þ CA hybrid CO2 removal, which ranged from 1.5 to 1.8 USD/GJ-ENG. The authors, however, considered in their analysis the maximum production of 6,000,000 sm3/d along the entire project lifetime. 4.5.2. Process footprint Using MP total area in the MP þ CA process, the diameter of the Table 8 Economic analysis values. Parameter Scenario 1 Scenario 2 “Sizing FPSO” CAPEX (MUSD) OPEX (MUSD) ENG (PJ) LCC (USD/GJ-ENG) 235 801 373 2.15 233 794 428 1.86 CA columns and compressor power for the six FPSO designs in each scenario, the footprints associated to these skids were calculated by correlations in Table 4 as proposed in Araújo et al. (2017) and are shown in Fig. 11a. Total Footprint decays mainly from the decrease in equipment size resulting from reduction of RNG ﬂowrate. Since Footprint is directly related to equipment size, the Footprint proﬁles behave like the CAPEX proﬁle: footprints decrease with decreasing RNG ﬂowrates. However, Scenario 2 (lower CO2 content in RNG) shows slightly higher footprint after the Production Plateau, being more pronounced after the 10th y of operation. Similarly to the speciﬁc CAPEX-1 and speciﬁc CAPEX-2 indexes, Fig. 11b and c presents speciﬁc Footprint-1 (Total Footprint divided Fig. 10. ENG time proﬁle [Two scenarios: (1) 0% CO2 storage; (2) 60% CO2 storage; 1 ¼1st y, 2 ¼ 5th y, 3 ¼ 10th y, 4 ¼ 15th y and 5 ¼ 20th y; PJ ¼ 1015 J]. A. de Carvalho Reis et al. / Journal of Cleaner Production 200 (2018) 269e281 Fig. 11. Footprint proﬁles for ﬁve FPSO designs: (a) total Footprint; (b) speciﬁc Footprint-1; and (c) speciﬁc Footprint-2 [Two scenarios: (1) 0% CO2 storage; (2) 60% CO2 storage; speciﬁc Footprint-1: Total Footprint/RNG energy content; speciﬁc Footprint-2: Sizing FPSO Footprint (1582 m2 for 0% CO2 storage and 1558 m2 for 60% CO2 storage)/RNG energy content; 1 ¼1st y, 2 ¼ 5th y, 3 ¼ 10th y, 4 ¼ 15th y and 5 ¼ 20th y; Design stands for the “Sizing FPSO”]. by RNG energy content) and speciﬁc Footprint-2 (“Sizing FPSO” Total Footprint - 1582 m2 for 0% CO2 storage and 1558 m2 for 60% CO2 storage, divided by RNG energy content). After the end of the Production Plateau, Scenario 2 implies lower speciﬁc footprints than Scenario 1 (~10%). Footprint is more inﬂuenced by RNG feed ﬂowrate than by its % CO2, and speciﬁc Footprint-2 is most impacted by the production curve, corroborating the conclusion that the gas processing plant will be oversized for ﬂow rates below 6,000,000 sm3/d (i.e., after the Production Plateau). Fig. 12a, b and 12c decompose the footprint by equipment assemblage. MP and COMP skids are slightly sensitive to RNG CO2 content, being mainly impacted by its ﬂow rate, while CA skid is inﬂuenced by both variables. Although CAPEX of CA is larger than CAPEX of MP (Fig. 11b and c), the proﬁles are reversed for Footprints (Fig. 12b and c). The “Sizing FPSO” has the same sizes of MP and CA skids of Point 1 in both scenarios, so their respective footprints are also the same. MP 279 Fig. 12. Footprint values for ﬁve FPSO designs: (a) total Footprint; (b) speciﬁc Footprint-1; and (c) speciﬁc Footprint-2 [Two scenarios: (1) 0% CO2 storage; (2) 60% CO2 storage; 1 ¼1st y, 2 ¼ 5th y, 3 ¼ 10th y, 4 ¼ 15th y and 5 ¼ 20th y; Design stands for the “Sizing FPSO”]. and CA skids together represented only 30% of Total Footprint, with COMP skids being responsible for the main footprint. Similarly to CAPEX, Fig. 12b and c do not aim to compare the footprints of MP and CA skids, as they perform different and complementary services, processing feeds with different CO2 contents. Fig. 12b and c also present the %CO2 of each skid feed. 4.5.3. Carbon footprint Dividing FPSO CO2 emissions by ENG energy is a way of smoothing the impact of feed ﬂow rate variation, evidencing the huge inﬂuence of CO2 content in the feed. This, in fact, occurs, since the point-to-point analysis reveals that higher %CO2 feeds (Scenario 1) generate larger CO2 emissions to achieve ENG speciﬁcations (3% CO2), as can be seen in Fig. 13. 4.5.4. HC losses Fig. 14 presents HC losses. For Scenario 1 (higher %CO2 in RNG), 280 A. de Carvalho Reis et al. / Journal of Cleaner Production 200 (2018) 269e281 and intermediate CO2 content. Both speciﬁc CAPEX and footprint have similar behaviors: constant at the Production Plateau of gas producing curve and increasing with reduction in RNG ﬂowrate beyond the ﬁfth y. MP and COMP are mildly sensitive to %CO2 of RNG but are highly impacted by RNG ﬂowrate. CA is affected by both variables. On entire production curve, HC losses were low e a beneﬁt of hybrid MP þ CA e from 4% to 5% for all FPSO designs. Acknowledgements OQF Araújo and JL de Medeiros acknowledge ﬁnancial support from PETROBRAS S.A. (0050.0096933.15.9) and CNPq-Brazil (309640/2016-4 and 311076/2017-3). Fig. 13. Carbon Footprint [1 ¼1st y, 2 ¼ 5th y, 3 ¼ 10th y, 4 ¼ 15th y and 5 ¼ 20th y]. Appendix A. Supplementary data Supplementary data related to this article can be found at https://doi.org/10.1016/j.jclepro.2018.07.271 References Fig. 14. HC losses in injection gas (IG) [1 ¼1st y, 2 ¼ 5th y, 3 ¼ 10th y, 4 ¼ 15th y and 5 ¼ 20th y]. HC losses are higher in Points 2 and 3. However, for Points 4 ad 5 (ﬁnal period of the production curve, i.e., smaller RNG ﬂowrates), Scenario 2 shows higher HC losses than Scenario 1 (~5% higher). RNG ﬂow rate does not dominate the inﬂuence on the behavior of HC losses. Regardless of the scenario and production curve period, HC losses are small, being inferior than 4.6% for all FPSO designs, a beneﬁt granted by the hybrid MP þ CA technology, as presented by Reis et al. (2017). 5. Conclusions This work evaluated the design of FPSOs oriented by lifetime parameters, namely, decreasing production ﬂow rate simultaneously with increasing CO2 content in the reservoir resulting from CO2-EOR, considering hybrid MP þ CA for CO2 separation. Flow rate variation was predicted by Arps hyperbolic decline, while CO2 content proﬁle in the reservoir was calculated for two scenarios: (1) 0% CO2 storage; (2) 60% CO2 storage. Designs were evaluated though performance metrics - CAPEX, OPEX, footprint, emissions and HC losses in the injected gas (IG), for ten feed conditions along the production curve. Calculations showed, in both scenarios, the “Sizing FPSO” being composed only by equipment related to operation conditions at the ﬁrst and ﬁfth y, both located in the Production Plateau of the gas producing curve, where maximum ﬂow rate of raw NG occurs. The “Sizing FPSO” corresponds to design at maximum feed ﬂow (6,000,000 sm3/d) and %CO2 between 44.7% and 46.5%. The resulting FPSO is oversized for other feed conditions in the remaining lifetime. CAPEX and Footprint as speciﬁc values (CAPEX and Footprint of “Sizing FPSO”/RNG energy content), conﬁrmed that FPSO design must be performed considering maximum production ﬂow rate Andrade, A.M.T., Vaz, C.E.M., Ribeiro, J., Lopreato, L.G.R., Nascimento, R.F.S., 2015. Offshore production units for pre-salt projects. In: OTC-25691-MS, Offshore Technology Conference, Houston, USA. Araújo, O.Q.F., Reis, A.C., de Medeiros, J.L., Nascimento, J.F., Grava, W.M., Musse, A.P., 2017. Comparative analysis of separation technologies for processing carbon dioxide rich natural gas in ultra-deepwater oil ﬁelds. J. Clean. Prod. 155, 12e22. https://doi.org/10.1016/j.jclepro.2016.06.073. Arps, J.J., 1945. Analysis of decline curves. Analysis of decline curves. Trans. Am. Inst. Min. Metall. Pet. Eng. 160, 228e247. https://doi.org/10.2118/945228-G. Arinelli, L.O., Trotta, T.A.F., Teixeira, A.M., de Medeiros, J.L., Araújo, O.Q.F., 2017. Offshore processing of CO2 rich natural gas with supersonic separator versus conventional routes. J. Nat. Gas Sci. Eng. 46, 199e221. https://doi.org/10.1016/ j.jngse.2017.07.010. Barrera, J.E., Bazzo, E., Kami, E., 2015. Exergy analysis and energy improvement of a Brazilian ﬂoating oil platform using Organic Rankine Cycles. Energy 88, 67e79. https://doi.org/10.1016/j.energy.2015.03.091. Barton, C., Hambling, H., Albaugh, E.K., Mahlstedt, B., Davis, D., 2017. Worldwide survey of ﬂoating production, storage and ofﬂoading (FPSO) Units. Available in: www.offshore-mag.com/content/dam/offshore/print-articles/volume-77/08/ 2017FPSO-072017-Ads.pdf. (Accessed 23 April 2018). Carbon Tracker, 2017. Stranded assets. Available in: www.carbontracker.org/terms/ stranded-assets/. (Accessed 8 August 2018). Energy Institute at Haas, 2016. Time to unleash the carbon market? Available in: https://energyathaas.wordpress.com/2016/06/20/time-to-unleash-the-carbonmarket/. (Accessed 23 April 2018). Formigli Filho, J.M., Pinto, A.C.C., Almeida, A.S.A., 2009. Santos Basin's pre-salt reservoirs development e the way ahead. In: OTC 19953. 2009 Offshore Technology Conference, Houston, 4-7 May, USA. https://doi.org/10.4043/19953MS. Gaffney, Cline, Associates, September 15th. 2010. Review and Evaluation of Ten Selected Discoveries and Prospects in the Pre-Salt Play of the Deepwater Santos Basin, Brazil. CG/JW/RLG/C1820.00/GCABA.1914, ANP. Available in: www.anp. gov.br/?dw¼39132. (Accessed 20 January 2018). Gallo, W.L.R., Gallego, A.G., Acevedo, V.L., Dias, R., Ortiz, H.Y., Valente, B.A., 2017. Exergy analysis of the compression systems and its prime movers for a FPSO unit. J. Nat. Gas Sci. Eng. 44, 287e298. https://doi.org/10.1016/ j.jngse.2017.04.023. Gazalpour, F., Ren, S.R., Tohidi, B., 2005. CO2 EOR and storage in oil reservoirs. Oil Gas Sci. Technol. 60 (3), 537e546. Gerasimchuk, I., Bassi, A.M., Ordonez, C.D., Doukas, A., Merrill, L., Whitley, S., 2017. Zombie Energy: Climate Beneﬁts of Ending Subsidies to Fossil Fuel Production. Working paper Published by the International Institute for Sustainable Development. Available in: https://www.iisd.org/sites/default/ﬁles/publications/ zombie-energy-climate-beneﬁts-ending-subsidies-fossil-fuel-production.pdf. (Accessed 28 May 2018). €gg, M.B., Lindbråthen, A., He, X., Nodeland, S.G., Cantero, T., 2017. Pilot Ha demonstration-reporting on CO2 capture from a cement plant using hollow ﬁber process. Energy Procedia 114, 6150e6165. https://doi.org/10.1016/ j.egypro.2017.03.1752. IPCC, 2014a. Intergovernmental Panel on Climate Change. Climate Change 2014, Synthesis Report, Summary for Policymakers. Available in: www.ipcc.ch/report/ ar5/. (Accessed 24 April 2018). Intergovernmental Panel on Climate Change, 2014b. Climate Change 2014, Synthesis Report, ipcc.ch/pdf/assessment-report/ar5/syr/AR5_SYR_FINAL_All_Topics.Pdf. (Accessed 24 April 2018). Islam, A.B.M.S., Jameel, M., Jumaat, M.Z., Shirazi, S.M., Firas, A., 2012. Review of offshore energy in Malaysia and ﬂoating spar platform for sustainable A. de Carvalho Reis et al. / Journal of Cleaner Production 200 (2018) 269e281 exploration. Renew. Sustain. Energy Rev. 16, 6268e6284. https://doi.org/ 10.1016/j.rser.2012.07.012. Jenkins, S., 2017. Current economic trends: CEPCI January (prelim.) and December (ﬁnal). Available in: www.chemengonline.com/current-economic-trends-cepcijanuary-prelim-and-december-ﬁnal/. (Accessed 20 January 2018). Hall, C.A.S., Lambert, J.G., Balogh, S.B., 2014. EROI of different fuels and the implications for society. Energy Policy 64, 141e152. https://doi.org/10.1016/ j.enpol.2013.05.049. Kim, S.-K., Roh, M.-I., Kim, K.-S., 2017. Evaluation of feasibility index in the arrangement design of an offshore topside based on the automatic transformation of experts' knowledge and the fuzzy logic. Ocean Eng. 130, 284e299. https://doi.org/10.1016/j.oceaneng.2016.11.057. Kwak, D.H., Yun, D., Binns, M., Yeo, Y.K., Kim, J.K., 2014. Conceptual process design of CO2 recovery plants for enhanced oil recovery applications. Ind. Eng. Chem. Res. 53, 14385e14396. https://doi.org/10.1021/ie502110q. Luu, M.T., Milani, D., Abbas, A., 2016. Analysis of CO2 utilization for methanol synthesis integrated with enhanced gas recovery. J. Clean. Prod. 112, 3540e3554. https://doi.org/10.1016/j.jclepro.2015.10.119. Musarra, S.P., 2017. Offshore Leadership Forum Addresses CAPEX Challenges. Offshore Magazine, 02/15/2017. Available in: www.offshore-mag.com/articles/ print/volume-77/issue-2/offshore-leadership-forum/offshore-leadershipforum-addresses-capex-challenges.html. (Accessed 24 May 2018). 281 Petrobras, 2017. Petrobras focus on its strengths. Available in: www. agenciapetrobras.com.br/upload/documentos/apresentacao_Rh7L2Aabtr.pdf. (Accessed 27 May 2018). Pinto, A.C.C., Vaz, C.E.M., Branco, C.C.M., Ribeiro, J., 2014. An evaluation of large capacity processing units, for ultra deep water and high GOR oil ﬁelds. In: OTC25274-MS, Offshore Technology Conference, Houston, USA. https://doi.org/ 10.4043/25274-MS. Reis, A.C., de Medeiros, J.L., Nunes, G.C., Araújo, O.Q.F., 2017. Upgrading of natural gas ultra-rich in carbon dioxide: optimal arrangement of membrane skids and polishing with chemical absorption. J. Clean. Prod. 165, 1013e1024. https:// doi.org/10.1016/j.jclepro.2017.07.198. Shimamura, Y., 2002. FPSO/FSO: state of the art. J. Mar. Sci. Technol. 7, 59e70. https://doi.org/10.1007/s007730200013. Thiabaud, P., Gros, F., Tourdjman, J., Guaraldo, M., 2011. Beneﬁts of life cycle dynamic simulation for FPSO projects and offshore developments. In: SPE-143532-MS. , Brazil. https://doi.org/10.2118/143532-MS. Brasil Offshore, 14e17 June, Macae Turton, R., Bailie, R.C., Whiting, W.B., Shaeiwitz, J.A., Bhattacharyya, D., 2012. Analysis, Synthesis, and Design of Chemical Processes, fourth ed. Prentice Hall, Michigan, USA. ISBN-13: 978-0-13-261812-0. Yu, L.C., King, L.S., Hoon, A.T.C., Yean, P.C.C., 2015. A review study of oil and gas facilities for ﬁxed and ﬂoating offshore platforms. Research Journal of Applied Sciences. Eng. Technol. 10 (6), 672e679. https://doi.org/10.19026/rjaset.10.2476.