Coproduction of Acrylic Acid with a Biodiesel Plant Using CO 2 as Reaction Medium: Process Modeling and Production Cost Estimation

: Producing value-added chemicals from glycerol is imperative for the sustainable future of biodiesel. Despite worldwide e ﬀ orts, the commercial production of acrylic acid from glycerol faces challenges, both technologically and economically. Based on our new technology using CO 2 as a reaction medium in a two-step process to catalytically convert glycerol to acrylic acid, we established computer simulation models to analyze the energy e ﬃ ciency and estimate production costs at di ﬀ erent scenarios. The analysis was conducted in conjunction with published data of a typical, intermediate-sized biodiesel facility, aiming at the feasibility of producing acrylic acid on-site in the context of a circular economy. Variable analysis in response to the market value of glycerol, the source and cost of carbon dioxide recycling, and the changes in process scale and conditions are also presented. Results indicates that a cost-e ﬀ ective supply of CO 2 to the acrylic acid plant is key to the further research and development.


Introduction
Current biodiesel production, although utilizing renewable resources, fits into a linear economy, which is a take-make-dispose model. For a sustainable future of biofuels, it is imperative to move away from a linear economy toward a circular economy with key elements of prioritizing regenerative resources, using waste as a resource, designing for the future, preserving and extending what is already made, and collaborating to create joint value [1]. The common biodiesel production through transesterification of oils produces~10% by weight of crude glycerol as coproducts. Because of the massive volume of biodiesel production, a worldwide effort to valorize glycerol was initiated [2,3]. Among the value-added chemicals that can be derived from glycerol, acrylic acid produced via the intermediate of glycerol dehydration to acrolein received much attention because this appears to be one of the most promising ways to valorize glycerol [4,5]. There have been extensive efforts in recent years to improve glycerol dehydration to acrolein, including glycerol conversion in sub-and supercritical water [6][7][8], as well as selective gas-phase conversions [9][10][11][12][13]. Although these different processes gave adequate glycerol conversion rates and selectivity to acrolein, a continuous process catalyzed by solid acids is desirable for industrial application.
However, despite their positive initial performance, few solid acid catalysts are known to maintain their catalytic activity long enough for practical application without the need for regeneration, as the high functionality of glycerol leads to severe coking on the catalysts. This fast catalyst deactivation remains a major technical problem for the commercial production of acrolein and acrylic acid from glycerol [5,14].

Main Steps of Converting Glycerol to Acrylic Acid
Technical data input to our modeling is based on our experimental laboratory results [15,16]. First, glycerol diluted with water goes through dehydration in the medium of supercritical carbon dioxide catalyzed by a solid acid (ZSM-5 is used in the modeling) to generate acrolein, according to the following equation.
Generation of side products of acetol, acetaldehyde, and propionaldehyde is also considered in this first step.
Carbon dioxide at supercritical conditions, having liquid-like density and gas-like diffusivity, possesses special solvent and transport properties [18]. The major reaction product of acrolein, which is also considered as a coke precursor [5], is soluble in the highly diffusive SC-CO 2 [19] to be quickly moved out of the catalyst bed, preventing it from further secondary reactions that lead to decomposition and coking. Other coke precursors can be dissolved through the combination of water polarity and SC-CO 2 non-polarity, washing out both the hydrophilic and hydrophobic coke precursors and thus enabling long catalyst life [15].
Down the stream, the process is similar to that in the industrial production of acrylic acid via oxidation of propylene to acrolein. The use of CO 2 here also provides a safe reaction medium for performing oxidation reactions within. After pressure release and a condensation step where all unconverted glycerol and acetol (both have a relatively high boiling point) are separated for further refining, the remaining gaseous stream containing acrolein, acetaldehyde, propionaldehyde, and CO 2 is passed through the second reaction of partial oxidation with air catalyzed by a Mo-V based catalyst. Here, acrolein is converted to acrylic acid according to Equation (2).
Acetic acid and a small amount of propionic acid are generated as side products from the oxidation of acetaldehyde and propionaldehyde, respectively.
Acrylic acid is the main revenue product, and acetol and acetic acid are also considered as revenue products. The yield of acrylic acid, acetol, and acetic acid from glycerol is 65 mol%, 12 mol%, and 16 mol%, respectively.

Process Modeling
Unlike the biodiesel model [17] that was based on well-established processes with years of commercial operations, modeling in this work was based on our experimental data at lab scale that was then projected to larger scales of contemporary engineering practices. SuperPro Designer software (Version 8.5, Intelligen Inc., Scotch Plains, NJ, USA) was used to conduct the modeling work, providing modeling of manufacturing and end-of-pipe treatment processes under a single umbrella, as well as project economic evaluation.
Because this process modeling was motivated to help assess the potential economic feasibilities, bottlenecks, operation targets for process improvement and identify further research and development effort requirements at the early stage of new biobased technology, the model is intended to capture key unit operations for the conversion of glycerol to acrylic acid and the product refining, representative of contemporary industry practices. However, this modeling is not meant to represent the actual design offered by any single technology provider. Although the model is in its relatively preliminary stages in terms of the level of its detail, it is not intended to replace the thorough engineering analysis that is required in the final design and construction of such a plant; we designed a plant that can recycle most of its process water and carbon dioxide in light of a future circular economy, possibly adding extra cost to the establishment. The plant is designed to continuously operate three shifts per day, 350 days per year, converting 3810 metric tons of glycerol (80 wt.%) from an intermediate-sized biodiesel facility to acrylic acid (1700 metric tons) with acetol (314 metric tons) and acetic acid (350 metric tons) as coproducts. It is noteworthy that this is rather small scale comparing to 80-160 metric tons annually for a commercial acrylic acid plant using propylene as feedstock. The SuperPro Designer flowchart is presented in Figure 1, with four color-coded process sections modeling glycerol dehydration to acrolein, acrolein oxidation to acrylic acid, product refining, and CO 2 /energy recycling.
Energies 2020, 13, x FOR PEER REVIEW 4 of 14 noteworthy that this is rather small scale comparing to 80-160 metric tons annually for a commercial acrylic acid plant using propylene as feedstock. The SuperPro Designer flowchart is presented in Figure 1, with four color-coded process sections modeling glycerol dehydration to acrolein, acrolein oxidation to acrylic acid, product refining, and CO2/energy recycling. Glycerol feed is first diluted with recycled water to 30 wt.% (V-101), and pressurized to 7.8 MPa (DP-101). This feedstock of 30 wt.% glycerol is merged with recycled CO2 at the same pressure (MX-101) and flows into the dehydration reactor (RX-101) operated at 320 °C and 7.8 MPa with ZSM-5 as a solid acid catalyst, where the glycerol is converted to acrolein with acetol, acetaldehyde, and propionaldehyde as byproducts. After being depressurized (GTV-10) and cooled with a line of recycled CO2 (HX-105), the reaction effluent passes through a condensation step (HX-101), where part of water and all the acetol and unconverted glycerol are condensed. The condensates go through a distillation (C-101) medium to generate 99.8% acetol as a revenue product. The uncondensed stream (S-114) is mixed with air (MX-102) and flows into the partial oxidation reactor (RX-102), which is operated at 300 °C at near-atmospheric pressure, where the acrolein is converted to acrylic acid with acetic acid and propionic acid as side products; a Mo-V based catalyst is used in this reactor [16].
A condensation step (HX-102) separates the second reaction effluent into two streams. The gaseous stream (S-124) contains CO2, N2, and O2, and the CO2 is recycled. The condensate stream (S-104), consisting of acrylic acid, acetic acid, propionic acid, and water, passes through a liquid-liquid extraction (DX-101) with ethyl acetate as the extractant, which can be recycled (via stream S-122 and unit V-102). The bottom partition of the extractor mainly consists of water (S-123), which is recycled. The top partition, comprised of extracted acrylic acid, acetic acid, and propionic acid in ethyl acetate, is first added with hydroquinone as the inhibitor (MX-104) to prevent polymerization of acrylic acid, before passing through two stages of distillation. The first stage of distillation (C-104) recovers ethyl acetate (via stream S-122); the second distillation generates acetic acid (99.5%) at the top and acrylic acid (99.1% plus 0.23% hydroquinone as anti-polymerization additive) at the bottom.
To efficiently utilize process heat and CO2, the gaseous stream after the condensation of the second reaction effluent (S-124) is first cooled with a recycled line of CO2 (HX-106), followed by compression to 2.2 MPa (G-101) and chilling to −20 °C (HX-107), so the CO2 in the liquid phase is separated from gaseous N2 and O2 (CSP-101) and recycled. The recycled CO2 line (S-118), going through two stages of compression (DP-102 and G-102) to reach 7.8 MPa, is heated with the effluent  Glycerol feed is first diluted with recycled water to 30 wt.% (V-101), and pressurized to 7.8 MPa (DP-101). This feedstock of 30 wt.% glycerol is merged with recycled CO 2 at the same pressure (MX-101) and flows into the dehydration reactor (RX-101) operated at 320 • C and 7.8 MPa with ZSM-5 as a solid acid catalyst, where the glycerol is converted to acrolein with acetol, acetaldehyde, and propionaldehyde as byproducts. After being depressurized (GTV-10) and cooled with a line of recycled CO 2 (HX-105), the reaction effluent passes through a condensation step (HX-101), where part of water and all the acetol and unconverted glycerol are condensed. The condensates go through a distillation (C-101) medium to generate 99.8% acetol as a revenue product. The uncondensed stream (S-114) is mixed with air (MX-102) and flows into the partial oxidation reactor (RX-102), which is operated at 300 • C at near-atmospheric pressure, where the acrolein is converted to acrylic acid with acetic acid and propionic acid as side products; a Mo-V based catalyst is used in this reactor [16].
A condensation step (HX-102) separates the second reaction effluent into two streams. The gaseous stream (S-124) contains CO 2 , N 2 , and O 2 , and the CO 2 is recycled. The condensate stream (S-104), consisting of acrylic acid, acetic acid, propionic acid, and water, passes through a liquid-liquid extraction (DX-101) with ethyl acetate as the extractant, which can be recycled (via stream S-122 and unit V-102). The bottom partition of the extractor mainly consists of water (S-123), which is recycled. The top partition, comprised of extracted acrylic acid, acetic acid, and propionic acid in ethyl acetate, is first added with hydroquinone as the inhibitor (MX-104) to prevent polymerization of acrylic acid, before passing through two stages of distillation. The first stage of distillation (C-104) recovers ethyl acetate (via stream S-122); the second distillation generates acetic acid (99.5%) at the top and acrylic acid (99.1% plus 0.23% hydroquinone as anti-polymerization additive) at the bottom.
To efficiently utilize process heat and CO 2 , the gaseous stream after the condensation of the second reaction effluent (S-124) is first cooled with a recycled line of CO 2 (HX-106), followed by compression Energies 2020, 13, 6089 5 of 14 to 2.2 MPa (G-101) and chilling to −20 • C (HX-107), so the CO 2 in the liquid phase is separated from gaseous N 2 and O 2 (CSP-101) and recycled. The recycled CO 2 line (S-118), going through two stages of compression (DP-102 and G-102) to reach 7.8 MPa, is heated with the effluent line of the dehydration reaction (S-121), before flowing back to the dehydration reactor (RX-101) via the mixer (MX-101).

Estimation of Production Costs
A depreciable life of 10 years was assumed for the project. Estimation of production costs was based on the flowchart above ( Figure 1) that forms the "Base Model" of this study. The following categories that could be specified in SuperPro Designer were considered in the estimation of production costs.

1.
Material-related information. This includes the different unit costs (e.g., purchase price, selling price, waste treatment/disposal cost) of pure components, and stock mixtures.

2.
Stream-related information. This includes the classification of input and output streams into different categories (e.g., raw materials, revenues, wastes, etc.).

3.
Operation-related information. This includes the costs of labor and utilities (heat transfer agents and power) for each operation.

4.
Equipment-related information. This includes the capital and operating costs for each equipment (e.g., purchase cost, installation, maintenance, consumables, etc.).
We used the built-in cost model and database in SuperPro Designer that estimated each equipment purchase cost based on size, the material of construction, and other characteristics, and the cost was brought up to the year of 2017 with an inflation rate of 2% per year. All the process vessels were specified to meet The American Society of Mechanical Engineers (ASME) standards of vessel construction that would increase the cost. Concerning construction materials, all the unit operation vessels or reactors are specified to be constructed of stainless steel with the exceptions of the two large CO 2 compressors (G-101 and G-102), having carbon steel as the specified construction material.
We followed the methodology of parametric input of Haas et al. [17] to the SuperPro Designer for capital investment adjustment. Direct fixed capital was estimated based on the total equipment purchase cost plus the installation cost. Unlisted equipment purchase cost was estimated as 20% of the listed equipment purchase cost, and unlisted equipment installation cost is two times the unlisted equipment purchase cost. Working capital was estimated to cover 30 days of utilities, labor, and raw materials, and the startup and validation cost account for 5% of the direct fixed capital. In the same way, operating cost was adjusted considering the facility-dependent cost that included depreciation, maintenance, and miscellaneous costs of insurance, local taxes, and factory expenses. A lumped estimate of 16,800 labor-hours per year was assigned to labor cost. Because environmental pollution regulations vary from location to location, no precise calculation of waste stream treatment costs was attempted. However, in the annual operating budget, $50,000 was allocated for waste stream disposal charges.
Because this study focused on process modeling and production cost estimation at the early stage of a developing technology, economic factors of economic life, internal rate of return, corporate tax rate, salvage value, debt fraction, construction interest rate, long term interest rate, and marketing and distribution expenses, were not accounted for. The existence of governmental credits or subsidies was not considered.
Based on the input of material throughputs and operation conditions, SuperPro Designer would first solve the mass and energy balances, and then perform economic calculations. Importantly, the establishment of the Base Model allowed adjustment of input parameters for variable analysis in response to the market value of glycerol, the source and cost of carbon dioxide recycling, and the changes in process scale and conditions. All the following tables and figures are based on our modeling and economic calculations using SuperPro Designer.

Overview of Base Model
A summary of the production cost estimation is presented in Table 2. Detailed information on the data and terminology is provided in the annotated economic evaluation report generated by SuperPro Designer (Supplemental Information). Approximately $7.5 million total capital investment is needed to establish the designed facility converting glycerol to acrylic acid. Also listed in Table 2 are major operating costs and revenue values of our input into the modeling. The purchasing price of crude glycerol is taken from the selling price reported in Haas et al. [17]. This price and those for revenue products represent a medium value of their fluctuating prices in the past ten years. The total labor cost is estimated as $462,000 annually, slightly less than that ($490,600) of an intermediate-sized biodiesel facility reported in Haas et al. [17], since this project has fewer number of unit operations. Generally, this "Base Model" acrylic acid plant is profitable, although the profit margin is smaller, comparing to the intermediate-sized biodiesel facility [17]. Because the model is relatively preliminary in the level of its detail, and it is not meant to replace the thorough engineering analysis that is required in the final design and construction of such a plant, we will focus on the difference between the unit production cost and unit production revenue (profit margin) relative to the Base Model in the ensuing analysis and discussion. This will help to assess the potential economic feasibilities, bottlenecks, operation targets for process improvement and identify further research and development effort requirements at the early stage of a new biobased technology.
Breaking down annual operating costs reveals more information (Figure 2). Unlike the intermediate-sized biodiesel facility where the soybean oil feedstock accounts for 88% of total operating cost, the largest portion of annual operating cost is the utilities, followed in order by raw materials, facility-dependent costs, and labor (Figure 2A). Dividing the operating cost by the four designed process sections shows that the dehydration section occupies the highest percentage, followed by the CO 2 /Energy recycle section ( Figure 2B). After analysis of the utilities category ( Figure 2C), we found that the top contributors to utility costs are standard power and CaCl 2 brine used for deep cooling. It is noteworthy to point out that the most expensive piece of equipment (~$1 million) is the CO 2 compressor (G-101) that initially pressurizes a large volume of CO 2 from near atmospheric pressure to 2.2 MPa before deep cooling to separate and recycle the CO 2 . The standard power consumption of this CO 2 compressor ranks third-highest among all unit operations, after the dehydration reactor (RX-101) and the partial oxidation reactor (RX-102) that handle reactions at high temperatures of 300 • C. This analysis indicates that the use of SC-CO 2 as a reaction medium and its recycling entails a high operating cost, especially in terms of utilities.
Model in the ensuing analysis and discussion. This will help to assess the potential economic feasibilities, bottlenecks, operation targets for process improvement and identify further research and development effort requirements at the early stage of a new biobased technology. Breaking down annual operating costs reveals more information (Figure 2). Unlike the intermediate-sized biodiesel facility where the soybean oil feedstock accounts for 88% of total operating cost, the largest portion of annual operating cost is the utilities, followed in order by raw materials, facility-dependent costs, and labor (Figure 2A). Dividing the operating cost by the four designed process sections shows that the dehydration section occupies the highest percentage, followed by the CO2/Energy recycle section ( Figure 2B). After analysis of the utilities category ( Figure  2C), we found that the top contributors to utility costs are standard power and CaCl2 brine used for deep cooling. It is noteworthy to point out that the most expensive piece of equipment (~$1 million) is the CO2 compressor (G-101) that initially pressurizes a large volume of CO2 from near atmospheric pressure to 2.2 MPa before deep cooling to separate and recycle the CO2. The standard power consumption of this CO2 compressor ranks third-highest among all unit operations, after the dehydration reactor (RX-101) and the partial oxidation reactor (RX-102) that handle reactions at high temperatures of ~300 °C. This analysis indicates that the use of SC-CO2 as a reaction medium and its recycling entails a high operating cost, especially in terms of utilities.

Factor Effects on Base Model
Factors and their effects on the Base Model are presented in Figure 3. Here we define the unit production revenue minus unit production cost for acrylic acid (AA) as "Unit(R-C)"; when this value becomes negative, the plant will not be profitable. The glycerol purchasing price slightly affects the total investment for the facility but has a more pronounced impact on the annual operating cost. This Base Model plant can be marginally profitable if the feedstock price rises to $0.40/kg ( Figure 3A).
Historically, the acrylic acid price greatly fluctuated, heavily dependent on the petroleum market and the supply of propylene. Nonetheless, the global acrylic acid market is expected to reach USD $22.6 billion by 2022, with demand in superabsorbent polymer production as a key driving factor for market growth, according to a study by Grand View Research, Inc., San Francisco, CA, USA [20]. It appears that acrylic acid priced at $2/kg is the break-even point for the Base Model ( Figure  3B).
With further research and development in reaction catalysis, it is reasonable to achieve higher acrylic acid yield and lower reaction temperatures. The profit margin can be increased by 74% if the acrylic acid yield can be increased by 15% to reach 80 mol% ( Figure 3C). If the reaction temperature for both dehydration and partial oxidation can be lowered to 280 °C, the total capital investment and annual operating cost will be reduced by 0.7% and 1.8%, respectively; utility costs will be reduced by 4.3%, while the Unit(R-C) will be increased by 14.8% ( Figure 3D).

Factor Effects on Base Model
Factors and their effects on the Base Model are presented in Figure 3. Here we define the unit production revenue minus unit production cost for acrylic acid (AA) as "Unit(R-C)"; when this value becomes negative, the plant will not be profitable. The glycerol purchasing price slightly affects the total investment for the facility but has a more pronounced impact on the annual operating cost. This Base Model plant can be marginally profitable if the feedstock price rises to $0.40/kg ( Figure 3A).
Historically, the acrylic acid price greatly fluctuated, heavily dependent on the petroleum market and the supply of propylene. Nonetheless, the global acrylic acid market is expected to reach USD $22.6 billion by 2022, with demand in superabsorbent polymer production as a key driving factor for market growth, according to a study by Grand View Research, Inc., San Francisco, CA, USA [20]. It appears that acrylic acid priced at $2/kg is the break-even point for the Base Model ( Figure 3B).
With further research and development in reaction catalysis, it is reasonable to achieve higher acrylic acid yield and lower reaction temperatures. The profit margin can be increased by 74% if the acrylic acid yield can be increased by 15% to reach 80 mol% ( Figure 3C). If the reaction temperature for both dehydration and partial oxidation can be lowered to 280 • C, the total capital investment and annual operating cost will be reduced by 0.7% and 1.8%, respectively; utility costs will be reduced by 4.3%, while the Unit(R-C) will be increased by 14.8% ( Figure 3D).

Effect of CO2 Input on Base Model
In our research and development of the SC-CO2 technology to extend the service life of solid acid catalysts, the amount of CO2 input has not been optimized. The Base Model requires cycling 9270 kg per hour of CO2, 50 times the molar input of glycerol. Our simulation results shown in Figure 4 demonstrate the critical need to reduce the CO2 input for the improvement of this new technology.

Effect of CO 2 Input on Base Model
In our research and development of the SC-CO 2 technology to extend the service life of solid acid catalysts, the amount of CO 2 input has not been optimized. The Base Model requires cycling If the CO2 input can be cut down by 50%, the total capital investment will be reduced by 20% from $7.47 million to $5.97 million; the annual operating cost will be decreased by 17% from $3.99 million to $3.33 million, and the Unit(R-C) will be increased by 140% from 0.27 to $0.65/kg AA. Further cutting down CO2 input to 38% of that of the Base Model will further reduce capital investment and operating cost, elevating the profit margin ( Figure 4A).
Breaking down the operating cost by category reveals that the utility costs exhibit a significant reduction with the decreasing CO2 input, and raw materials rise to the top cost category ( Figure 4B). Among the utilities used, the reduction of standard power consumption is pronounced ( Figure 4C). Future research should prioritize the reduction of CO2 input without sacrificing its function of extending catalyst service life.

Effect of No CO2/Energy Recycling
Changes are made to the Base Model by removing the process section of CO2/Energy recycling shown in Figure 1, while keeping other process conditions the same. In this NR-Model, there is a  If the CO 2 input can be cut down by 50%, the total capital investment will be reduced by 20% from $7.47 million to $5.97 million; the annual operating cost will be decreased by 17% from $3.99 million to $3.33 million, and the Unit(R-C) will be increased by 140% from 0.27 to $0.65/kg AA. Further cutting down CO 2 input to 38% of that of the Base Model will further reduce capital investment and operating cost, elevating the profit margin ( Figure 4A).
Breaking down the operating cost by category reveals that the utility costs exhibit a significant reduction with the decreasing CO 2 input, and raw materials rise to the top cost category ( Figure 4B). Among the utilities used, the reduction of standard power consumption is pronounced ( Figure 4C). Future research should prioritize the reduction of CO 2 input without sacrificing its function of extending catalyst service life.

Effect of No CO 2 /Energy Recycling
Changes are made to the Base Model by removing the process section of CO 2 /Energy recycling shown in Figure 1, while keeping other process conditions the same. In this NR-Model, there is a source of CO 2 as input to the plant, requiring two gas compressors and an extra heater to bring the input CO 2 to the required pressure and temperature before entering the dehydration reactor. Five scenarios are considered for the NR-Model in contrast to the Base Model ( Figure 5): (1) the input CO 2 has a pressure of 1 Bar with no cost; (2) the input CO 2 has a pressure of 10 Bar with no cost; (3) the input CO 2 has a pressure of 20 Bar and costs $15 per metric ton; (4) the input CO 2 at 1 Bar has no cost and the volume can be cut down by 50%; and (5) the input CO 2 at 1 Bar can be cut down by 50% in volume and costs $25 per metric ton.
Energies 2020, 13, x FOR PEER REVIEW 10 of 14 source of CO2 as input to the plant, requiring two gas compressors and an extra heater to bring the input CO2 to the required pressure and temperature before entering the dehydration reactor. Five scenarios are considered for the NR-Model in contrast to the Base Model ( Figure 5): (1) the input CO2 has a pressure of 1 Bar with no cost; (2) the input CO2 has a pressure of 10 Bar with no cost; (3) the input CO2 has a pressure of 20 Bar and costs $15 per metric ton; (4) the input CO2 at 1 Bar has no cost and the volume can be cut down by 50%; and (5) the input CO2 at 1 Bar can be cut down by 50% in volume and costs $25 per metric ton.   Although the total capital investment can be lowered by~5% if we do not recycle CO 2 /Energy in the first scenario, the annual operating cost would be increased by 10%, resulting in a 78% lower profit margin ( Figure 5A) compared to the Base Model. This proves the advantage of including CO 2 /Energy recycling in the design of our Base Model. If the input CO 2 is at 10 Bar (scenario 2), significant reductions of capital investment (by 31%) and operating cost (by 12%) can be achieved with an increase of profit margin by 100%, comparing to the Base Model. This analysis indicates that the initial compression of a large volume of CO 2 input at near atmospheric pressure is an energy-intensive process, also requiring costly large equipment. Further rise of the input CO 2 pressure to 20 Bar (scenario 3) would render a plant close to the break-even point even if the CO 2 costs $15 per metric ton. As we have shown in the analysis for the Base Model in Section 3.3, the reduction of CO 2 input volume has a pronounced impact on lowering investment and operating costs. The same is true for the NR-Model if the CO 2 input is cut down by 50% (scenario 4), and the plant would be close to the break-even point even if the CO 2 costs $25 per metric ton (scenario 5). Figure 5B shows the relative percentage changes of each operating cost category for all the five scenarios in comparison with the Base Model, reinforcing that (1) initial compression of a large volume of CO 2 is costly in terms of both utilities and facility; (2) the top two utilities cost items are the standard power and CaCl 2 brine which are mainly used for CO 2 compression and cooling ( Figure 5C); and (3) the reduction of CO 2 input would significantly increase profit margin. It is noteworthy to point out that the future development of this bio-based technology should consider cascading processing in conjunction with emerging Carbon Capture and Utilization or Recycling (CCU or CCR) technologies that can efficiently provide CO 2 .

Effect of Threefold Scale-Up
Although we intended to conduct the process modeling and production cost estimation of a cascading acrylic plant based on a typical intermediate-sized biodiesel facility, larger scales of biodiesel facilities exist. Therefore, we scale up the Base Model to evaluate the scaling impact. Two scenarios are considered here as shown in Figure 6 in contrast to the Base Model: (1) a threefold scale-up in term of glycerol input volume (and proportionally other inputs of materials) keeping all process conditions the same as those of the Base Model, and (2) a threefold scale-up in term of glycerol input volume keeping all process conditions the same as those of the Base Model, except that the CO 2 input is reduced by 50%.
The threefold scale-up (3 × scale) plant requires approximately a doubled capital investment and 2.6 times the annual operating cost, comparing to the Base Model, and the profit margin is doubled as indicated by the Unit(R-C). Reducing CO 2 input by 50% significantly lowers the capital investment and operating cost, leading to a more than tripled profit margin ( Figure 6A).
With the threefold scale-up, the relative percentage of utilities cost jumps to nearly a half of the total operating cost, while that of the facility-dependent cost shrinks ( Figure 6B), although now the most expensive piece of equipment for initial CO 2 compression (G-101) costs nearly $2.9 million, indicating an energy-intensive plant. Reducing CO 2 input by 50% brings the percentage of the cost of utilities down to the same level as that for the Base Model ( Figure 6B). Looking further into the itemized utility costs ( Figure 6C) reveals that the scale-up increases the percentage of standard power cost to 48% of the total utilities cost, again mainly due to the large power consumption for compressing the CO 2 . It is clear that even for a threefold scale-up, the reduction of CO 2 input should be prioritized in further research and development.
Although continuous development of better catalysts remains as one way to extend catalyst life [21], the potential of "green" process engineering should be explored. We would like to point out that Carbon Capture and Utilization or Recycling (CCU or CCR) technologies are advancing quickly. In the North American market, ammonia producers in the United States are used to a price range of around US$3-15 per metric ton for bulk gaseous/supercritical CO 2 , which varies significantly depending on the location; the price for pipelined CO 2 has historically been in the range of US$9-26 per metric ton, which includes the cost of the pipeline infrastructure (capital and operational costs) [22]. Ethanol is already a part of the bioeconomy. A typical ethanol plant producing 50 million gallons of ethanol per year also produces~150,000 metric tons of high-purity CO 2 annually [23]. These are potential sources that could provide CO 2 to an acrylic acid plant in a cost-effective manner.
Energies 2020, 13, x FOR PEER REVIEW 12 of 14 of ethanol per year also produces ~150,000 metric tons of high-purity CO2 annually [23]. These are potential sources that could provide CO2 to an acrylic acid plant in a cost-effective manner.  Investment ($ million) Annual operating cost ($ million) Unit(R-C) ($/kg AA)

Conclusions and Future Perspective
This study establishes simulation models in the context of a circular economy to analyze the technical and economic feasibility of producing acrylic acid from glycerol in conjunction with a typical intermediate-sized biodiesel facility, based on our new technology of using carbon dioxide as reaction medium in a two-step process to catalytically convert glycerol to acrylic acid. The results of our process modeling and production cost estimation indicate that coproduction of acrylic acid with an intermediate-sized biodiesel plant can be economically feasible, contributing to the bioeconomy and circular economy, and the profit margin can be doubled with a threefold scale-up.
Other than market factors such as fluctuating prices of glycerol and acrylic acid, our simulation reveals that from the technical standpoint, the reduction of CO 2 input without sacrificing its function of extending catalyst service life has the highest positive impact on the economic benefit, followed by further increase of acrylic acid yield and reduction to milder reaction conditions. The profit margin can be increased by 74% if the acrylic acid yield can be increased by 15% to reach 80 mol%, whereas the profit margin can be increased by 140% if the CO 2 input can be cut down by 50%. Our results clearly pointed out directions for further research and development of this new technology in the way that should consider cascading processing in conjunction with emerging Carbon Capture and Utilization or Recycling Technologies that can cost-effectively provide CO 2 .
New biobased technologies are emerging as scattered dots, and it is now the time to connect the dots to develop cascading and integrated technologies towards a circular economy.