• Keine Ergebnisse gefunden

4.6 Economic Evaluation

4.6.6 Discussion

The market value for fossil-based ethylene in Brazil is speculated to be high.

The predominant production route is steam cracking of naphtha and the

coun-Table 4.18: Bio-ethylene production cost in USD kg−1C

2H4 resulting from the Monte Carlo simulation with dierent condence levels

Condence Cost 55.2% ≤0.70 87.0% ≤1.50 95.0% ≤1.73 99.0% ≤1.95 99.9% ≤2.10

try often relies on light oil imports for petrochemical production since local oil is mostly heavy. Limited infrastructure for transportation and processing also add to the nal cost. Brazil also produces commercial bio-ethylene via catalytic dehydrogenation of rst generation sugar-cane bio-ethanol and a previous study estimated a production cost in the range of1.13 USD kgC2H4 to1.17 USD kgC2H4

(Haro, Ollero, and Trippe, 2013). The manufacturer, i.e., Braskem applies ra-diocarbon analysis (ASTM D6866) to quantify the content of bio-based carbon and then certify the products, e.g., polyethylene, with a seal labeled "I am green". Although not clearly stated, it is very likely that customers pay a higher price for the certied bio-ethylene than for its fossil counterpart. A sim-ilar strategy could be applied to market bio-ethylene produced by the BG-OCM process.

It is also safe to assume that prices for fossil commodities such as ethylene and natural gas will increase in a short to medium term future due to increasing environmental regulations, implementation of carbon taxes, and depletion of oil resources. The production scales achievable with biogas are insucient to ever fully replace naphtha and ethane-based ethylene, but BG-OCM can act as link technology until novel processes for bio-based platform chemicals are deployed at industrial scale. This enables the use of renewable resources to produce a conventional product with populace acceptance and installed transportation and processing infrastructure.

This study also discovered that the cost assigned to the lights stream as a side product is of major importance to the feasibility of the processing plant.

Improving ethylene yields, e.g., by catalyst modication or by carrying out the reaction in a chemical looping sequence Fleischer et al., 2016, is always desirable but likely only feasible for medium-to-long term industrial implementation. On the other hand, developing multi-product systems in which OCM can be e-ciently integrated oer greater opportunity for its deployment in the short term.

Therefore, it is essential that future studies also concentrate on the utilization

4.6 Economic Evaluation and valorization of the lights stream containing the unreacted methane, such as methanation and dry reforming. Perhaps even a combination of these with CHP may, at the cost of higher capital investments, provide a exible solution, which allows the plant to increase/decrease production amounts for ethylene, electricity, or syngas, depending on market conditions.

4.6.7 Conclusion and Outlook

The production cost for bio-based ethylene has been estimated based on the pre-vious simulation results and cost estimations from Chapters 4.2 to 4.4. Electric-ity for gas compression and steam for amine regeneration represent the highest shares of the total utility cost, while process compressors represent the highest share of the installed equipment cost. The main component of the educts and side products cash ow is the lights stream, which is sold to the adjacent CHP unit at prices between 29 % to 72 % of the price of natural gas in Brazil. On one hand, high energy prices benet the BG-OCM plant because it exports light gases or (indirectly) electricity as a by-product. On the other hand, this also benets alternative projects such as a standalone CHP unit or a biogas up-grade unit, which oer reduced capital investment requirements. Future studies should, thus consider alternative utilization paths for the light stream, e.g., via methanation and reforming reactions, that can be combined with OCM.

A Monte Carlo simulation is employed to estimate the nal ethylene pro-duction cost, which has been found to be 0.53±0.73USD kg−1C

2H4 and lower than the market value of fossil ethylene. Increasing concerns with sustainability from multiple stakeholders could enable bio-ethylene to be marketed at a higher price than its fossil competitor.

5 Conclusion

This thesis has proposed and addressed the application of an heterogeneous catalytic reaction, i.e., Oxidative Coupling of Methane (OCM), to activate methane in biogas for the production of a value-added chemical, i.e., ethylene.

It conceptualized and investigated the techno-economic feasibility of a Biogas-based Oxidative Coupling of Methane (BG-OCM) process on an industrial pro-duction scale by means of process modeling and simulation followed by cost estimations. First-principle models have been developed and validated for all reaction and separation steps. Mathematical optimization has been used to de-sign a cost-ecient process, which resulted in a reaction section containing two OCM adiabatic Packed-Bed Reactors (PBRs) in series, a CO2 removal section consisting of standalone amine absorption, and a distillation section applying a Recycle Split Vapor (RSV) scheme. The BG-OCM is coupled with a Combined Heat and Power (CHP) unit to exploit the caloric value of the light o-gases and an economic evaluation is performed by estimating the bio-ethylene pro-duction cost and comparing it with the market value of fossil ethylene.

This thesis has also provided methodological contributions to the eld of Pro-cess Systems Engineering by modifying the Probability of Improvement (PI) method (Carpio, Giordano, and Secchi, 2018; Jones, 2001). A k-Nearest Neigh-bor (kNN) classication method has been proposed to map owsheet conver-gence behavior on a probabilistic basis. This improves how the PI method handles failed or non-converged simulations and makes it better suited for Sequential-Modular (SM) owsheet optimization or for any problem wherein the black-box model evaluation may fail. A Python package (Bbop) has been programmed to enable Aspen Plus connectivity and stochastic and Surrogate-Assisted Optimization (SAO) using the modied PI method and other algo-rithms available in public libraries. The package is applied to optimize the design of each section of the BG-OCM process. Although Dierential Evolu-tion (DE) located the best soluEvolu-tions in all test cases, the PI algorithm is able to nd very close solutions at a fraction of the computational cost. Hence, it can provide signicant time savings for optimization with expensive simulation models.

The OCM reaction section has been optimized to enable a C2 product yield of 16.12 %in adiabatic regime with a La2O3(27%)/CaO catalyst described in

literature (Stansch, Mleczko, and Baerns, 1997). This performance is in ac-cordance with recent patents on commercial implementations of the Natural Gas-based Oxidative Coupling of Methane (NG-OCM) processes (Siluria Tech-nologies, 2015). However, the high inlet temperatures required (≈1013 K) and the high sensitivity of the reaction performance to the operating conditions, e.g., reactors' inlet temperatures, pose serious threats to technical feasibility.

In the patents, this performance is achieved at much lower inlet temperatures (≈873 K), but there are currently no published kinetic model for low tempera-ture OCM catalysts that could be used for process simulation. Therefore, there is the need to carry out further experiments and modify or develop kinetic mod-els for OCM catalyst systems operating at low temperatures, adiabatic regime, and under CO2 dilution conditions in order to provide a better assessment of the attainable reaction performance and of the operation feasibility with biogas.

The optimization of the CO2 removal section has shown that the use of a hy-brid separation process involving Gas-Separation Membranes (GSM) and amine absorption is unlikely to bring any economic advantage if the OCM reaction is carried out near atmospheric pressures. The hybrid separation scheme requires a gas compression step in the upstream of the GSM, which drastically increases electricity consumption and capital investment in compressors. Standalone ab-sorption at low pressure (3.13 bar) carried out with a 30 wt%aqueous solution of Monoethanolamine (IUPAC: 2-aminoethan-1-ol) (MEA) provided lower cap-ital and operational expenses as well as higher product (C2H4) recoveries, hence being the most cost-ecient solution.

A distillation conguration using a Recycle Split Vapor (RSV) scheme has been simulated based on several dierent owsheets and wide operating ranges described in patents on OCM and natural gas processing. The RSV aims at using process streams as refrigeration utility in the condensers of the demeth-anizer and C2-splitter distillation columns. The process is simulated and its economic performance is compared to a traditional distillation scheme, wherein only external refrigeration is applied for that purpose. The RSV scheme can provide 24 % reduction in the ethylene separation cost. Further work in this direction should include optimization of both congurations and the compar-ison with alternative separation technologies such as Pressure-Swing Adsorp-tion (PSA). Finally, dierent transient regime analyses are essential to ensure feasible process control and start-up and shut-down procedures.

Finally, an economic evaluation has been carried out based on the simulation results. The bio-ethylene production cost is estimated using utility, equipment, educts, and side product costs and compared to the market value of fossil ethy-lene. It is demonstrated that, due to limited methane conversion in the reaction, the feasibility of BG-OCM is highly dependent on the fate and the sales value of

the lights stream containing the unreacted methane, which is the top product of the demethanizer column. In this study, this stream is sold at a fraction of the natural gas price in Brazil to be energetically recovered in an adjacent CHP unit.

Given the large uncertainties in the cost estimations, a Monte Carlo simulation with 10,000 samples is performed and the resulting bio-ethylene production cost is 0.53±0.73USD kg−1C2H4 with 87 % of the values below the upper bound for fossil ethylene market value, which lies around1.5 USD kgC−1

2H4. Therefore, the process concept is likely to be feasible in locations with high prices for energy, natural gas, and petrochemicals and that have extensive bio-resources available for large-scale biogas production, e.g., Brazil, China, and India. In Brazil, the concept can be used to utilize biogas derived from the Anaerobic Digestion (AD) of vinasse in the bioethanol industry and potentially add incen-tive for the implementation of these treatment units. The BG-OCM concept can also be interesting in Europe, wherein more customers may be willing to pay a premium for a plastic, e.g. polyethylene, derived from bio resources. It would be interesting to compare this project to a standalone CHP that simply explores biogas energetically or a biogas upgrade unit that produces biomethane as a direct substitute for pipeline natural gas or as a vehicular fuel. Further alternatives to be explored are the methanation and reforming reactions, which can be employed as standalone or in combination with OCM.

The use of catalysis and the shift towards renewable feedstock are two prin-ciples of green chemistry (Anastas and Warner, 1998). In this thesis, ethylene production, which is commonly achieved by a thermochemical process and oil-derived feedstock, is achieved by a catalytic process with a renewable feedstock.

However, the feedstock and process substitution alone does not necessarily yield a sustainable process nor supply chain. This issue is best addressed by means of Life-Cycle Inventory (LCI) and Life-Cycle Analysis or Assessment (LCA). For instance, biogas could be used to produce ethylene and polymers, thus reducing naphtha/ethane consumption and improving impact categories such as carbon dioxide emissions and natural resources depletion. However, other categories such as land use may be aected negatively. Also, a full system expansion is necessary for this type of analysis in order to ensure equal outputs for a scenario-based comparison. Biogas is currently used for electricity production and, if it is directed for ethylene production via OCM instead, other sources must replace biogas for electricity production. If, for example, this is achieved by natural gas or coal, there may be a net increase in carbon dioxide emissions and natural resources depletion. Therefore, there is still the need to perform detailed analyses in order to properly assess the merits of a BG-OCM process concept on an environmental and social basis.

Appendix A

Process Models

This chapter contains complementary information and validation plots for the process models utilized within this thesis.

A.1 OCM Reactor

Figures A.1 to A.6 show educt conversion and product yield at700°C(913 K) and 830°C (1103 K) under the conditions summarized in Table 2.1. In these plots, lines are results obtained with the isothermal Plug-Flow Reactor (PFR) model while crosses are experimental data from (Stansch, Mleczko, and Baerns, 1997) for comparison. The error bars are not from experiments (these have not been reported), but rather the average deviations between model and experi-mental data obtained by (Stansch, Mleczko, and Baerns, 1997).

0 10 20 30 40 50 60 mcat/VSTP [kgcat*s*Nm-3]

0.00 0.05 0.10 0.15 0.20 0.25

X CH4

700°C (Stansch 1997) 700°C (Model) 830°C (Stansch 1997) 830°C (Model)

Figure A.1: Methane conversion for dierent contact times in isothermal lab-scale Packed-Bed Reactor at 700°C and 830°C. Comparison of reactor model predictions and experimental data from (Stansch, Mleczko, and Baerns, 1997).

0 10 20 30 40 50 60

mcat/VSTP [kgcat*s*Nm-3] 0.0

0.2 0.4 0.6 0.8 1.0 1.2

X O2

Oxygen Conversion for Different Contact Times

700°C (Stansch 1997) 700°C (Model) 830°C (Stansch 1997) 830°C (Model)

Figure A.2: Oxygen conversion for dierent contact times in isothermal lab-scale Packed-Bed Reactor at 700°C and 830°C. Comparison of reactor model predictions and experimental data from (Stansch, Mleczko, and Baerns, 1997).

A.1 OCM Reactor

Ethylene Yields for Different Contact Times

700°C (Stansch 1997) 830°C (Stansch 1997) 700°C (Model) 830°C (Model)

Figure A.3: Ethylene yield for dierent contact times in isothermal lab-scale Packed-Bed Reactor at 700°Cand 830°C. Comparison of reactor model predictions and experimental data from (Stansch, Mleczko, and Baerns, 1997).

Ethane Yields for Different Contact Times

700°C (Stansch 1997) 830°C (Stansch 1997) 700°C (Model) 830°C (Model)

Figure A.4: Ethane yield for dierent contact times in isothermal lab-scale Packed-Bed Reactor at 700°Cand 830°C. Comparison of reactor model predictions and experimental data from (Stansch, Mleczko, and Baerns, 1997).

0 10 20 30 40 50 60 mcat/VSTP [kgcat*s*Nm-3]

0.00 0.01 0.02 0.03 0.04 0.05 0.06

Y CO2

700°C (Stansch 1997) 830°C (Stansch 1997) 700°C (Model) 830°C (Model)

Figure A.5: Carbon Dioxide yield for dierent contact times in isothermal lab-scale Packed-Bed Reactor at 700°C and 830°C. Comparison of reactor model predictions and experimental data from (Stansch, Mleczko, and Baerns, 1997).

0 10 20 30 40 50 60

mcat/VSTP [kgcat*s*Nm-3] 0.01

0.01 0.02 0.02 0.03 0.03 0.03 0.04 0.05

Y CO

Carbon Monoxide Yields for Different Contact Times

700°C (Stansch 1997) 830°C (Stansch 1997) 700°C (Model) 830°C (Model)

Figure A.6: Carbon Monoxide yield for dierent contact times in isothermal lab-scale Packed-Bed Reactor at 700°C and 830°C. Comparison of reactor model predictions and experimental data from (Stansch, Mleczko, and Baerns, 1997).

A.2 Carbon Dioxide Removal

A.2 Carbon Dioxide Removal

Figures A.7 to A.18 show the solubility of dierent gases in water and in aqueous or pure Monoethanolamine (IUPAC: 2-aminoethan-1-ol) (MEA) at dierent temperatures. Lines are predictions using the model described in Section 2.2.2, whereas crosses are experimental data from dierent sources. Agreement with experimental data is very good in the range of operating conditions, which is between 1 bar to 32 bar and 313 K to 403 K. If available, error bars for the experimental data are also shown.

0.0 0.5 1.0 1.5 2.0 2.5 3.0

Liquid Mole Fraction of Hydrogen [mol/mol] 10-3 0

0.5 1 1.5 2 2.5

Total Pressure [Pa]

107 Solubility of Hydrogen in Water as a P-x Diagram

311K Gillespie et al. 1980 366K Gillespie et al. 1980 422K Gillespie et al. 1980 478K Gillespie et al. 1980 311K Model

366K Model 422K Model 478K Model

Figure A.7: Solubility of Hydrogen in Water at310.891 K,366.459 K,422.004 K and 477.554 K: Model predictions and experimental data from (Gillespie and Wilson, 1980).

0.0 1.0 2.0 3.0 4.0 5.0 6.0 Liquid Mole Fraction of Nitrogen [mol/mol] 10-4 0

1 2 3 4 5 6 7 8

Total Pressure [Pa]

106 Solubility of Nitrogen in Water as a P-x Diagram

313K Baranenko et al. (1990) 353K Baranenko et al. (1990) 433K Baranenko et al. (1990) 313K Model

353K Model 433K Model

Figure A.8: Solubility of Nitrogen in Water at 273 K,353 K, and433 K: Model predictions and experimental data from (Baranenko et al., 1990).

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6

Liquid Mole Fraction of Methane [mol/mol] 10-3 0

5 10 15

Total Pressure [Pa]

106 Solubility of Methane in Water as a P-x Diagram

313K Kiepe et al. (2003) 353K Kiepe et al. (2003) 373K Kiepe et al. (2003) 313K Model

353K Model 373K Model

Figure A.9: Solubility of Methane in Water at 313.15 K and 373.29 K: Model predictions and experimental data from (Kiepe et al., 2003).

A.2 Carbon Dioxide Removal

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0

Liquid Mole Fraction of Ethylene [mol/mol] 10-3 0

106 Solubility of Ethylene in Water as a P-x Diagram

311K Davis et al. (1960) 360K Davis et al. (1960) 394K Davis et al. (1960) 310K Model

360K Model 394K Model

Figure A.10: Solubility of Ethylene in Water at311 K,360 K, and394 K: Model predictions and experimental data from (Davis and McKetta, 1960).

0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0

Liquid Mole Fraction of Ethane [mol/mol] 10-4 0

106 Solubility of Ethane in Water as a P-x Diagram

311K Culberson and McKetta Jr. (1950) 378K Culberson and McKetta Jr. (1950) 444K Culberson and McKetta Jr. (1950) 311K Model

378K Model 444K Model

Figure A.11: Solubility of Ethane in Water at311 K,378 K, and 444 K: Model predictions and experimental data from (Culberson and McKetta, 1950).

0 1 2 3 4 5 6 7 Liquid Mole Fraction of Carbon Monoxide [mol/mol] 10-4 0

1 2 3 4 5 6

Total Pressure [Pa]

106Solubility of Carbon Monoxide in Water as a P-x Diagram

310K Gillespie et al. (1980) 366K Gillespie et al. (1980) 310K Model

366K Model

Figure A.12: Solubility of Carbon Monoxide in Water at 310 K and 366 K: Model predictions and experimental data from (Gillespie and Wil-son, 1980).

0.000 0.005 0.010 0.015 0.020 0.025

Liquid Mole Fraction of Carbon Dioxide [mol/mol]

0 2 4 6 8 10 12

Total Pressure [Pa]

106 Solubility of Carbon Dioxide in Water as a P-x Diagram 298K Lucile et al. (2012) 348K Lucile et al. (2012) 393K Lucile et al. (2012) 298K Model

348K Model 393K Model

Figure A.13: Solubility of Carbon Monoxide in Water at 298 K, 348 K, and 393 K: Model predictions and experimental data from (Lucile et al., 2012).

A.2 Carbon Dioxide Removal

0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0

Liquid Mole Fraction of Hydrogen [mol/mol] 10-3 0

106 Solubility of Hydrogen in MEA as a P-x Diagram

323K Kling and Maurer (1991) 373K Kling and Maurer (1991) 423K Kling and Maurer (1991) 323K Model

373K Model 423K Model

Figure A.14: Solubility of Hydrogen in MEA at323 K,373 K, and423 K: Model predictions and experimental data from (Kling and Maurer, 1991).

0.0 0.2 0.4 0.6 0.8 1.0 1.2

Liquid Mole Fraction of Methane [mol/mol] 10-3 0

106 Solubility of Methane in a 3 kmol/m3 Aqueous Solution of MEA

298K Carrol et al. (1998) 348K Carrol et al. (1998) 398K Carrol et al. (1998) 298K Model

348K Model 398K Model

Figure A.15: Solubility of Methane in aqueous MEA with3 kmol m−3 at298 K, 348 K, and398 K: Model predictions and experimental data from (Carroll et al., 1998).

0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 Liquid Mass Fraction of Amine [kg/kg]

0.8 1 1.2 1.4 1.6 1.8 2

Liquid Mole Fraction of Ethylene [mol/mol]

10-4 Solubility of Ethylene in aqueous MEA solution

298K Sada and Kito (1972) 288K Sada and Kito (1972) 298K Model

288K Model

Figure A.16: Solubility of Ethylene in aqueous MEA with dierent concentra-tions, atmospheric pressure, and at 288 Kand 298 K: Model pre-dictions and experimental data from (Sada and Kito, 1972).

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0

Liquid Mole Fraction of Ethane [mol/mol] 10-3 0

2 4 6 8 10 12 14

Total Pressure [Pa]

106 Solubility of Ethane in a 3 kmol/m3 Aqueous Solution of MEA

298K Jou and Mather (2006) 348K Jou and Mather (2006) 398K Jou and Mather (2006) 298K Model

348K Model 398K Model

Figure A.17: Solubility of Ethane in aqueous MEA with 3 kmol m−3 at 298 K, 348 K, and 398 K: Model predictions and experimental data from (Jou and Mather, 2006).

A.2 Carbon Dioxide Removal

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 CO2 Loading in the Amine Solution [mol

CO2/mol

MEA] 10-5

100 105

CO 2 Partial Pressure [kPa]

Solubility of Carbon Dioxide in 30wt% aqeuous MEA

313K Jou et al. (1995) 353K Jou et al. (1995) 393K Jou et al. (1995) 313K Model

353K Model 393K Model

Figure A.18: Solubility of Carbon Dioxide in 30 wt% aqueous MEA solution at 313 K, 353 K, and393 K: Model predictions and experimental data from (Jou, Mather, and Otto, 1995).