propylene separation process

Accepted Manuscript Title: Techno-economic feasibility study of membrane based propane/propylene separation process Authors: Ung Lee, Jeongnam Kim, Il Seok Chae, Chonghun Han PII: DOI: Reference:

S0255-2701(16)30665-1 http://dx.doi.org/doi:10.1016/j.cep.2017.05.013 CEP 6997

To appear in:

Chemical Engineering and Processing

Received date: Revised date: Accepted date:

12-12-2016 21-2-2017 20-5-2017

Please cite this article as: Ung Lee, Jeongnam Kim, Il Seok Chae, Chonghun Han, Techno-economic feasibility study of membrane based propane/propylene separation process, Chemical Engineering and Processinghttp://dx.doi.org/10.1016/j.cep.2017.05.013 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Manuscript Prepared for Chemical Engineering and Processing: Process Intensification

Techno-Economic Feasibility Study of Membrane Based Propane/Propylene Separation Process Ung Leea†, Jeongnam Kimb† Il Seok Chaec* Chonghun Hanb* a

AVT Process Systems Engineering (SVT), RWTH Aachen University, Trumstrass 46, Aachen, 52064, Germany b

School of Chemical and Biological Engineering, Seoul National University, Gwanak-ro 1, Gwanak-gu, Seoul 08826, South Korea c

DWI Leibniz-Institut für Interaktive Materialien e.V., Forckenbeckstr. 50, Aachen, 52074, Germany

Graphical Abstract

Highlights      

Manuscript title: Techno-Economic Feasibility Study of Membrane Based Propane/Propylene Separation Process Ung Lee, Jeongnam Kim, Il Seok Chae, and Chonghun Han* Propose minimum specifications of economically attractive C3 separation membrane Techno economic optimization of the propylene separation process. Investigation of membrane specifications in terms of stage cut and pressure drop. Membrane specification changes are identified as the cost and feed condition varies Corresponding Author: Chonghun Han, Professor School of Chemical and Biological Engineering and Institute of Chemical Processes, Seoul National University, Gwanak-ro 1, Gwanak-gu, Seoul 08826, South Korea Tel: +82-2-880-1887 Fax: +82-2-873-2767 E-mail: [email protected]

†

Both authors contributed equally

*Author to whom correspondence should be addressed. Tel.: 82 (0)2/880-1887; E-mail: [email protected]

Manuscript Prepared for Chemical Engineering and Processing: Process Intensification

Abstract One of the most important issues for developing propylene/propane gas separation membranes is to identify the minimum required membrane performances. Herein, we reported the technical and economic feasibility of facilitated propylene transport membranes containing silver nanoparticles, which produces 99.6% purity of propylene with 97% recovery for 46,000 kg/hr capacity. In order to suggest minimum specification, the membrane separation process is compared with advanced distillation process employing direct vapor recompression process. The distillation process is optimized using a stochastic solver interfaced with a commercial process simulator. With the distillation process, the energy and total costs are approximately $4.40 and 18.6$ to produce per ton of propylene, respectively. The single-stage membrane processes with various stage-cut conditions were evaluated to meet the same total propylene production cost. Result indicates that required membrane permeance and selectivity ranges between 11.3 to 251.5 GPU (permeance unit, 1 GPU = 1 × 10-6 cm3 (STP)/(cm s cmHg)) and 61.9 to 1950 depending on stage-cut. Sensitivity analysis was also performed according to membrane costs, feed flow rates and composition.

Keywords: Propylene propane separation, Facilitated transport membranes, Vapor recompression distillation, Economic evaluation, Optimization

1

Introduction

Light olefins, such as ethylene and propylene, are two large volume organic chemical feedstocks and the key components in the chemical and petroleum industry. For instance, propylene is the monomer for polypropylene and a basic ingredient for many monomers containing acrylic acid, acrylonitrile, propylene oxide, propanol, allyl chloride[1-4]. So far, the recovery of propylene has been typically carried out using distillation, in the so-called C3 splitter process in Naphtha Cracking Center (NCC). This process is generally considered to be one of the most expensive and most energy intensive separation processes[5]. To overcome such disadvantages, membrane-based olefin separation has been proposed and investigated in recent years. More recently, Baker and Low argued that high permeance with the 15-20 propylene/propane selectivity need to be for olefin crackers[6]. However, for the replacement of distillation columns in large-scale process, higher permselective membrane would be required. Much effort has been undertaken to develop many type of membranes including polymers[7, 8], zeolites [9, 10], metal−organic frameworks [11-13], mixed matrix [14-16], and carbon molecular sieves [17-19] for olefin separation. However, most of these membranes suffer from certain limitations. In particular, the similar physicochemical properties of propylene and propane, such as their molecular size, shape and solubility, generally have limited separation performances governed by solutiondiffusion mechanisms in dense-type polymeric membrane. Zeolite membranes like faujasite (FAUtype) have shown a promising single gas permeation of propylene. Nonetheless, FAU-type zeolite has 12-membered oxygen rings and relatively large pores (ca 0.74nm in diameter); consequently, the

Page 2 of 27

Manuscript Prepared for Chemical Engineering and Processing: Process Intensification penetrant-induced plasticization leading to reduced mixed-gas selectivity. On the other hand, olefin separation based on a size-exclusion mechanism has been intensively studied by rational design of the pore size and pore distribution using zeolitic-imidazolate frameworks (ZIFs). Recently, Pan et al.[13] reported that ZIF-8 membrane prepared by a facile hydrothermal seeded growth method showed the highest values of a propylene/propane separation factor up to 50 and a propylene permeability 200 barrer. However, ZIF-8 membranes have often shown no separation performance, likely due to their poor grain boundary structure. Mixed matrix membranes including inorganic fillers are difficult to prepare the ultrathin selective layer, while carbon molecular sieve membranes are brittle and difficult to scale-up the production. Lately, facilitated transport membranes (FTM) have received much attention because they can improve both the permeability and the selectivity of the membrane[20]. For example, Sunderrajan et al. reported silver tetrafluoroborate (AgBF4) dissolved in semicrystalline poly(ethylene oxide) (PEO) showed approximately 80 times higher propylene solubility for propylene than for propane[21]. Using AgBF4 as olefin carrier, Jose et al. reported propylene/propane selectivity up to 160 and propylene permeance higher than 16 GPU for AgBF4 incorporated into poly(N-vinylpyrrolidone) (PVP) with phthalates as a plasticizer[22]. Despite these outstanding achievements, AgBF4 as an olefin carrier can be easily poisoned by the reduction from silver cation to silver metal while its long-term separation. Fortunately, a new system has been devised for robust olefin–paraffin separation based on a polymer matrix embedded with surface-modulated nanoparticles by Chae et al.; these membranes obtained a mixed gas selectivity of approximately 100 for a 50/50 (vol%) propylene–propane mixture[23, 24]. Also, to prevent the reduction AgBF4, PEO/AgBF4 membranes have been recently suggested to include a strong oxidant, Al(NO3)3[25]. Therefore, it is understood that FTM is one of most promising propylene separation membranes, and we adopt it as a basis to evaluate the cost of membrane in the present study. The main energy consumption of membrane-based gas separation processes is by the compressor necessary to compensate pressure drop in both permeate and retentate. Recompression is essential for product liquefaction and retentate stream recycling. Very recently, Huang et al. reported an exhaustive study on pressure ratio between feed stream and permeate stream with an economic viewpoint[26]. There is the other noticeable trade-off relationship between the membrane performance and its cost, i.e., the higher performance brings out higher cost at higher pressure ratio, and vice versa. Recent simulation studies highlight the importance of olefin recovery. Sridhar and Khan[27] simulated the ethylcellulose membrane and evaluated its performance in a single and multistage process. Recently, Motelica et al.[28] reported that no energy is required at such a temperature for column operating. However, the propylene/propane separation has gained less attention in simulation studies as compared with ethylene from ethane and butadiene from a C4mixture separation. In parallel to the development of high performance propylene separation membranes, the distillation process has been also making good progress. Several studies have investigated more energy-efficient technology to mitigate high energy demand. Direct vapor recompression process (VRC) and internally heat-integrated distillation column (HIDiC) have been suggested as promising options. Alcántara-Avila et al.[29] studied optimization of propane/propylene distillation process using mathematical programming, and showed that vapor recompression process can reduce the energy consumption between 58 and 75% when compared with conventional distillation. Chen et

Page 3 of 27

Manuscript Prepared for Chemical Engineering and Processing: Process Intensification al.[30] compared VRC and HIDiC in terms of cost and energy demand and they concluded that VRC scheme was superior to the HIDiC on the basis of uniform minimum temperature difference for heat transfer. On the other hand, Olujić et al.[31] performed a similar study, but they showed HIDiC structure exhibits 25% energy saving over VRC. Ho et al.[32] presented dynamic simulation of HIDiC structure and demonstrated their control configuration can handle the HIDiC under various disturbances. Adopting HIDiC process, however, in existing plant may need larger capital cost investment and introduce more complex configuration as compared with the VRC process. The superstructure optimization can be used for obtaining the optimum distillation process. The superstructure includes almost all possible process alternatives and the optimum process is systematically selected using mathematical optimization techniques. The superstructure optimization is applied in wide ranges of processes including heat exchanger network [33, 34], reaction network [35], distillation process [36, 37], and flowsheet optimization[38-40]. Comprehensive review is available in Trespalacios and Grossmann[41]. The purpose of this paper is to find the minimum required membrane specifications, i.e. selectivity and permeance, for an economically-attractive propylene separation process. Our strategy is divided two parts. The distillation process is firstly optimized in terms of cost. A MINLP model is constructed in MATLAB interfaced with a commercial simulator and solved stochastically using genetic algorithm. The distillation process configuration and operation condition are simultaneously optimized. Once the optimum propylene production cost of the distillation process is determined, permeance and selectivity of the membrane process having the same production cost are decided using one stage membrane separation model depending on the various operating conditions. Sensitivity analysis was also carried out to identify the influence of the important design parameters such as propylene feed composition, flow rate, and membrane unit cost.

2

Process Modeling and Optimization 2.1

Propylene distillation process

The Naphtha Cracking Center (NCC) is comprised of the steam thermal cracking, quenching, compressing, and purification process and converts naphtha to light olefins such as ethylene and propylene. As shown in Fig. 1, naphtha enters the thermal cracking furnace, where naphtha is broken down into small molecules with the presence of catalysts. The hot gas stream goes through the heat exchange section for preventing degradation and high pressure steam generation. Light olefins in gas phase are than extracted in the primary fractionator and desuperheated in the quench tower. The desuperheated stream is passed thorough the gas compression section and washed down to remove acid gases and water. Lastly, light olefins are fractionated in the purification process which includes demethnaizer, dedethanizaer, C2 splitter, depropanizer, C3 splitter, debutanizer, C4 extraction and others. The yields of furnaces are typically ranged 60 to 75% for ethane and 90 to 93% for propane and that of the propylene from propane is about 18 to 19% by weight depending on the furnace operating conditions[42]. The light gases, methane and hydrogen, are recovered from the overhead of the demethanizer and used as fuel gas. Note that the demethanizer can be operated in a wide pressure range[42]. The heavier components (C2+) from the demethanizer bottom are directed to deethanizer

Page 4 of 27

Manuscript Prepared for Chemical Engineering and Processing: Process Intensification where ethane, acetylene and ethylene are separated as the overhead stream. The bottom stream of the deethanizer contains C3+ components and sent to depropanizer[42]. The overhead stream of the depropanizer contains chemical grade propylene (92 to 95 mol%), and the C3 splitter further purifies propylene to the polymer grade propylene (over 99.5 mol%) [43]. Alkane products such as ethane and propane are recycled as feedstock, and light olefins are supplied for raw materials of petrochemicals. According to the U.S. Department of Energy, propane and propylene separation is generally the most energy intensive process because of the low relative volatility for the C3 mixtures. It also requires two distillation towers with a large number of stages over 100 and is accompanied by majority capital costs [4]. Herein, we modeled thermally integrated distillation process through VRC in order to mitigate high-energy consumption. External heat integration through VRC is often superior to the internal heat integration in terms of both energy consumption and economics[30]. Fig. 2 indicates the process flow diagram of the VRC C3 splitter process. The conventional propylene production process typically divides distillation column into two owing to height limitations and these columns are connected with a pump. Due to the high-purity requirement, the C3 splitter typically requires more rectifying stages than stripping stages[44], thus it was assumed that feed stage is located in the second column. The propane/propylene mixture entering the distillation column contains 95mol% propylene. The vapor stream exiting top of the column (stream 5) is compressed to high enough pressure to form a positive temperature difference in reboiler. Consequently, the hot vapor stream can be used as hot utility to drive reboiler as it condenses and returns to the column. Most of the energy requirement in VRC distillation process is the compressor work. Thus, overhead vapor stream is split and only a fraction is sent to compressor to minimize compressor duty. The propylene distillation process design can be formulated as a MINLP: min 𝑓(𝑥, 𝑦) subject to 𝑔𝑖 (𝑥, 𝑦) ≤ 0 ℎ𝑗 (𝑥, 𝑦) = 0 𝑥 ∈ 𝑋 ⊂ 𝑅𝑛 , 𝑦 ∈ 𝑁 𝑚 (P) where x and y are vectors consisting of n continuous and m integer variables. Continuous variables represent the feed stream vapor fraction, compressor inlet flow rate, and compressor pressure ratio and the integer variable is used for the feed stage of the column. 𝑓(𝑥, 𝑦), 𝑔(𝑥, 𝑦), and ℎ(𝑥, 𝑦) represent the objective function, inequality, and equality constraints, respectively. The distillation process involves a large number of equality constraints. These constraints describe thermodynamics, energy and mass balance resulting challenging MINLPs. In order to handle the MINLP efficiently with tractable computation time, we decompose the original problem into optimization and simulation part and equality constraints related to process simulation are explicitly calculated in a commercial process simulator. This decomposition simplifies the optimization problem, thus optimal solution can be found with relatively smaller population and generation of stochastic solver. Aspen Plus V8.6, was used for the simulation of the distillation process. Peng Robinson equation of state (EOS) was employed to predict thermodynamic properties of propane/propylene mixture. Modeling parameters such as feed conditions and column specifications are decided based upon industrial data(for propylene recovery, feed flow rate, total number of stage and feed pressure) or directly taken from the literature (e.g. top product composition[44, 45]). Table 1 summarizes the input parameters that were used for distillation process. Equality constraints are implemented in

Page 5 of 27

Manuscript Prepared for Chemical Engineering and Processing: Process Intensification Aspen Plus directly. The reflux flow rate is decided to satisfy the product purity, and column top pressure is calculated according to the condenser temperature through Aspen Plus design specification option. In this way, the MINLP model can avoid equality constraints, thus default stochastic solver imbedded in MATLAB can be used without modification. The

reduced MINLP for the distillation process can be expressed as 𝑎𝑛𝑛𝑢𝑎𝑙𝑖𝑧𝑒𝑑 𝐶𝐴𝑃𝐸𝑋 + 𝑂𝑃𝐸𝑋 2 min 𝑆𝑃𝐶 = 𝑡𝑜𝑛𝑛𝑒 +𝛾𝑝(𝑥, 𝑦) 𝑎𝑛𝑛𝑢𝑎𝑙 𝑝𝑟𝑜𝑝𝑒𝑛𝑒 𝑝𝑟𝑜𝑑𝑢𝑐𝑡𝑖𝑜𝑛 ( 𝑦𝑟 ) Subject to 0 ≤ 𝑥1 ≤ 1 1 ≤ 𝑥2 ≤ 4 1 ≤ 𝑥3 ≤ 1000 (tonne/hr) 101 ≤ 𝑦 ≤200 (1) Where 𝑥1 , 𝑥2 , and 𝑥3 are vapor fraction of the feed stream, compressor ratio, and compressor inlet flow rate. The objective function of the MINLP is specific production cost (SPC). 𝛾 and 𝑝(𝑥, 𝑦) are penalty parameter (1e7) and penalty function checking the convergence of process simulator. When the process simulator converges without error 𝑝(𝑥, 𝑦) returns zero. Otherwise, a large positive value is assign to objective function as the 𝑝(𝑥, 𝑦) becomes one. Equipment sizing and cost estimation are carried out using methods by Biegler[46] and Hasan et al.[47] The fixed capital investment was calculated from direct and indirect cost which can be estimated using factors specified by Douglas[48] and Peters et al.[49]. Total capital investment (CAPEX) and production cost (OPEX) were calculated once fixed capital cost value was obtained. Several assumptions were made to evaluate CAPEX and OPEX. Economic depreciation was assumed to be 5% over 15 years. The plant was assumed to be operated 8640 hours per year. Cooling water and electricity cost are assumed as $0.013/m3 and $0.065/kWh according to the Seider et al.[50]. The calculation of equipment cost estimation consists of equipment sizing and simple factoring. With regard to the purchase cost(PC) of the packed columns, we use the former method presented in Hasan et al. [47] PC = 𝐹𝑀 𝐶𝑉 + 𝐶𝑃𝐿 + 𝑉𝑃 𝐶𝑃𝐾 + 𝐶𝐷𝑅 , where 𝐹𝑀 , 𝐶𝑣 , 𝐶𝑃𝐿 , 𝑉𝑃 , 𝐶𝑃𝐾 , 𝐶𝐷𝑅 are the factor for materials of construction, the free on board purchase of the empty vessel, the added cost for column platforms and ladders, the packing volume, the installed cost of the packing for unit volume, and the installed cost of high performance liquid distributors and redistributors, respectively. 𝐹𝑀 =2.1, and 𝐶𝑃𝐾 = 40 are specified with constant value, and others are determined by following equation 𝐶𝑉 = exp[7.2756 + 0.18255 ln(𝑊) + 0.02297 ln(𝑊)2 ] 𝑡 𝑊 = 𝜋(39.37 ∙ 𝐷 + 𝑡𝑠 )(39.37(0.8𝐷 + 𝐿)) ∙ 𝜌𝑠 𝐶𝑃𝐿 = 300.9(3.281𝐷)²(3.281𝐿)0.80161 𝜋 𝑉𝑃 = 4 (3.281 ∙ 𝐷)²(3.281𝐿) 𝜋

𝐶𝐷𝑅 = 3 ∙ 125 ∙ 4 ∙ (3.281 ∙ 𝐷)².

Where D is the column packing diameter, L is the column packing length, ρ = 0.284lb/in3 is the density of carbon still, and 𝑡𝑠 = 2in is the shell thickness. The simple factoring method is considered in other equipment cases according to Biegler. [46] The base equipment costs (BC) is given by a

Page 6 of 27

Manuscript Prepared for Chemical Engineering and Processing: Process Intensification power law expression (𝐵𝐶 = 𝐶0 ∙ (𝑆⁄𝑆 )𝛼 ) 0 where 𝐶0 and 𝑆0 are the base costs and capacities, and the exponent 𝛼 is less than one. The kind of equipment determines those exact values with reference to Biegler. [46] An update factor(UF) is considered for accounting of difference between existing data and current case. The update bare module cost (BMC) is lastly calculated using BMC = UF ∙ BC ∙ (MF + MPF − 1). The material and pressure factor (MPF) and the module factor (MF) by equipment types are decided based on Biegler. [46] The optimization of the VCR distillation process is carried out using genetic algorithm (GA) in MATLAB by interfacing it with Aspen Plus using ActiveX Control. Recently, this method is widely implemented in the process flowsheet optimization [51, 52], reactor design [53], process operation condition optimization [54]. In this method, complex mathematical information and derivative of models are not required due to the fact that GA is based on function evaluations. 50 generations with 50 individuals are used. We choose 70% of crossover fraction, one elite population, and Gaussian mutation is employed in the GA solver. The maximum population size and five stall generations are used for termination criteria.

2.2

Membrane separation process

Herein, we build a single stage membrane separation process in order to estimate the permeance, and selectivity of the membrane. We assume perfect mixing in the feed stream and both permeate and retentate sides. Fig. 3 describes the process flow diagram of the considered membrane separation process. A fraction of the retentate is recycled, and pressure drop resulting from membrane separator is compensated using a compressor. The recycle of the retentate lowers the feed concentration of the membrane separator, thus membrane specifications such as selectivity are varying as the propylene pressure drop and stage cut changes. The compression of the recycle stream increases the steam temperature and eventually increases compression energy. Additional cooler is installed to lower the temperature, and it cools down the compressor outlet to make stream 2 saturated vapor. We neglect pressure drop in the cooler because it is considerable small compare to the membrane separator. Stage cut and pressure drop along the membrane separator are degrees of freedom. Thus, we vary these variables and observe how required permeance and selectivity are changed to meet the same SPC as in the distillation process. The calculation is carried out sequentially starting with mass balance. 𝑚̇1,𝑖 = 𝑚̇9,𝑖 + 𝑚̇4,𝑖 ∑ 𝑚̇9,𝑖 = 0.97 ∑ 𝑚̇1,𝑖 𝑚̇9,𝑝𝑟𝑜𝑝𝑦𝑙𝑒𝑛𝑒 = 0.996 ∑ 𝑚̇9,𝑖 (2) where 𝑚̇ indicates molar flow rate. The input and out streams of the membrane process can be easily calculated using equation (2). To calculate mass balance along the membrane separator, we define stage cut, 𝜃, and decided relevant recycle flow rate, stream 5. Note that low stage cut results in high recycle flow rate. ∑ 𝑚̇

𝜃 = ∑ 𝑚̇8,𝑖 2,𝑖

𝑚̇2,𝑖 = 𝑚̇7,𝑖 + 𝑚̇1,𝑖

Page 7 of 27

Manuscript Prepared for Chemical Engineering and Processing: Process Intensification 𝑚̇2,𝑖 = 𝑚̇8,𝑖 + 𝑚̇3,𝑖 𝑚̇3,𝑖 = 𝑚̇5,𝑖 + 𝑚̇4,𝑖

(3)

Once the mass balance is completed, energy balance is carried out. Thermodynamic properties are also calculated using Peng Robinson EOS. Required work for compressors is calculated using equation (4). 𝜂𝑖𝑠𝑒𝑛 𝑠 𝑊̇𝑐𝑜𝑚𝑝 = 𝑚̇𝑖𝑛 𝜂 (ℎ𝑜𝑢𝑡 − ℎ𝑖𝑛 ) (4) 𝑚𝑒𝑐ℎ

The isentropic (𝜂𝑖𝑠𝑒𝑛 ) and mechanical efficiency (𝜂𝑚𝑒𝑐ℎ ) of the compressor use default value of Aspen Plus isentropic compressor model. Enthalpies of inlet and outlet stream are calculated from flash calculation. The economic evaluation of membrane process is following once the mass and energy balance are completed. Unit operations except membrane separator are sized and economically evaluated according to Biegler [46] and Hasan et al.[47]. The cost information used for the distillation process is also applied to the membrane separation process. In order to suggest minimum specifications of the membrane separation process, the membrane separator cost along with its area are decided based upon the distillation process so as to equate the total costs of both processes. Note that total cost of the membrane separation process is higher even without membrane separator cost at the low stage-cut because of high recycle flow rate and consequent high capital and operational cost of compressors. The cost of the membrane separator can be calculated using equations (5). 𝑆𝑃𝐶𝑚𝑒𝑚𝑏𝑟𝑎𝑛𝑒 (𝐶𝑚𝑒𝑚 ) = 𝑆𝑃𝐶𝑑𝑖𝑠𝑡𝑖𝑙𝑙𝑎𝑡𝑖𝑜𝑛 𝐶𝑚𝑒𝑚 = 𝐴𝑚𝑒𝑚 𝑈𝐶𝑚𝑒𝑚 (5) where 𝐶𝑚𝑒𝑚 , 𝐴𝑚𝑒𝑚 , and 𝑈𝐶𝑚𝑒𝑚 indicate membrane separator equipment cost, required membrane area, and unit cost of membrane. Herein, we assumed the unit cost of FTM to be $100/m 2, which is typical production cost of FTM containing silver nanoparticles for lab scale experiments. The permeance is obtained using Molar flux equation (equation (6)). In this relationship, 𝑃 𝑙

permeance of a component i, 𝑖, can be calculated when molar flux and pressure drop are given. With the molar flux that we calculated from the stage-cut and membrane area, the required permeance of the membrane separator can be presented with respect to pressure drop 𝑗𝑖̇ =

𝑃𝑖 𝑓 (𝑝𝑖 𝑙

𝑝

− 𝑝𝑖 )

𝑓

𝑝

(6) where 𝑙, 𝑝𝑖 , and 𝑝𝑖 and indicate membrane

thickness, partial pressure of component i on the feed side and permeate side of the membrane. The permeability is often expressed in Barrer but, commercial membranes are typically expressed in terms 𝑃

of permeance ( 𝑙𝑖). Selectivity of the membrane separator is defined using equation (7) once permeance is calculated. 𝛼𝑖𝑗 =

𝑃𝑖 𝑃𝑗

=

𝑥𝑖,𝑝𝑒𝑟𝑚 (𝑥𝑗,𝑓𝑒𝑒𝑑 𝑃𝑓𝑒𝑒𝑑 −𝑥𝑗,𝑝𝑒𝑟𝑚 𝑃𝑝𝑒𝑟𝑚 )

(7)

𝑥𝑗,𝑝𝑒𝑟𝑚 (𝑥𝑖,𝑓𝑒𝑒𝑑 𝑃𝑓𝑒𝑒𝑑 −𝑥𝑖,𝑝𝑒𝑟𝑚 𝑃𝑝𝑒𝑟𝑚 )

Where 𝑥𝑖,𝑝𝑒𝑟𝑚 and 𝑥𝑖,𝑓𝑒𝑒𝑑 represent mole fraction of component i in permeate and feed sides of a membrane. In order to understand and suggest minimum specification of membrane separation process, permeance, selectivity, membrane area, permeate outlet pressure, CAPEX and OPEX are presented with respect to stage-cut and propylene pressure difference (ΔPpy). We define the propylene pressure difference using equation 8. ΔPpy = 𝑥𝑝𝑟𝑜𝑝𝑦𝑙𝑒𝑛𝑒,𝑓𝑒𝑒𝑑 𝑃𝑓𝑒𝑒𝑑 − 𝑥𝑝𝑟𝑜𝑝𝑦𝑙𝑒𝑛𝑒,𝑝𝑒𝑟𝑚 𝑃𝑝𝑒𝑟𝑚 (8)

Page 8 of 27

Manuscript Prepared for Chemical Engineering and Processing: Process Intensification The single stage membrane separation process is modeled using Microsoft Excel. Vapor phase flash is carried out to estimate the network of compressors. Simplex algorithm uses to find the membrane separator equipment cost and subsequent area.

3

Results and Discussion 3.1

Distillation process optimization result

The optimization of propylene distillation process is carried out using Aspen Plus V.8.6 and MATLAB R2014b on a desktop computer with Intel Xeon E5-2630 (2.6 GHz) processor and 128 GB RAM running window 7. In the GA optimization, the computation of MINLP problem takes about 35 hours and it is resulted from the process simulation time. The computation time for single population roughly takes one minute and the GA terminated with 38 generations. Note that we used the five consecutive stall generations as the termination criteria which is satisfied within the specified generation size. Table 2 summarizes the optimization result of the VRC distillation process. The objective function i.e., 𝑆𝑃𝐶𝑑𝑖𝑠𝑡𝑖𝑙𝑙𝑎𝑡𝑖𝑜𝑛 is 18.6$ per ton of propylene production. The energy cost for unit propylene production is about $4.40 per ton. At the optimum point, the column top pressure reaches at 16.7 bar where the overhead vapor stream can be condensed without external cooling media. The compressor work is reduced by 4% as fraction of the overhead vapor stream is directed to the condenser. The equipment cost, CAPEX, and OPEX of the distillation process are listed in the Table 3, 4, and 5. Note that operating labor cost is assumed constant for both the distillation process and membrane separation process. The total amount of propylene production is calculated as 368 ton/yr. The optimization of the distillation process does not guarantee the global optimum because of the stochastic solution. However, the solution indicates that SPC can be lowered about 36% as compared with C3 splitter currently operating in South Korea which has SPC of 29$/ton propylene.

3.2

Specification of membrane separator

Recall that the design specifications of the membrane-based propylene separation are to produce over 99.6% purity of propylene with 97% recovery in regard to the 95:5 (v/v) propylene/propane feed composition at 46,000 kg/hr rate. One of the effective strategies to reduce the energy consumption is to reduce the amount of recycle, i.e., high stage-cut and large membrane area. However, in this study, the case of stage-cut above 0.9 is not considered as it results in overly large membrane areas. In both distillation and membrane separation process, OPEX is higher than CAPEX. Higher compressor work is required in the membrane separation process in every range of stage-cut. However, OPEX of the distillation process is slightly higher because of the large amount of cooling water consumption. The compressor duty at the optimum permeance increases as the stage-cut decreases in spite of smaller pressure ratio due to the high recycle flow rate. The electricity consumption is ca. 5.9 GJ/hr and 6.5 GJ/hr for the distillation process and the membrane process at the optimum permeance with 0.9 stage-cut, respectively. Cooling water consumption of the distillation process can be about 60 times as high as that of the membrane separation process depending on the stage-cut and ΔPpy.

Page 9 of 27

Manuscript Prepared for Chemical Engineering and Processing: Process Intensification Fig. 4 indicates required specifications and economic potential of the membrane separator. Note that the membrane permeance and selectivity resign between 11.3 to 251.5 GPU and 61.8 to 1950 and both of them tend to decrease as the stage cut increases. Sum of OPEX (Fig. 4a) and CAPEX (Fig. 4b) is set to $18.6 per ton propylene production according to the economic evaluation result of the distillation process. Fig. 4a shows that OPEX tends to increase as ΔPpy increases due to the high pressure ratio of both recycle and product compressors. High stage-cut reduces operation cost by decreasing recycle flow rate and increasing propylene concentration in the feed stream. High propylene concentration also leads to reduction on the total pressure drop and consequently, compressor duty decreases. CAPEX shows opposite trends (Fig. 4b). The cost increment comes from the membrane separator cost, while capital and operational cost of other unit processes decrease as increasing stage-cut and decreasing the pressure difference. The high performance membrane is necessary if stage-cut (𝜃) of membrane process is at low level. On the other hand, membrane at higher stage-cut requires relatively mild performance and is also advantageous by downsizing and energy consumption of other unit operations (e.g., compressor, cooler). Total area of the membrane separator shows similar tendency of the CAPEX (Fig. 4c). The decrease of the membrane area is stiffer than CAPEX because the total equipment cost decreases as increasing pressure drop or decreasing stage-cut. Both pressure difference and membrane area are inversely related to permeance of membrane (equation (4)), and the required membrane area is decreased as the pressure difference increases. Therefore, there exists an optimum point in permeance at given stage-cut. For example, permeance of membrane specification is minimized at the pressure drop of 4 bar when stage-cut is 0.6 (Fig. 4d). The optimum permeance of each stage-cut was searched through sensitivity analysis and its result is presented in table 6. Selectivity of the membrane separator can be calculated using equation (7). Required specification decreases as the pressure drop and stage-cut increases (Fig. 4e). Notice that only variable in equation (7) is permeate side pressure at a given stage-cut, thus selectivity is only rational function defined on positive pressure drop. Finally, Fig. 4f indicates total pressure drop of the permeate side. Compositions of permeate and retentate are unchanged with fixed stage-cut by mass balance. As a result, total pressure drop is linearly related to the pressure drop of propylene.

3.3

Sensitivity analysis

Herein, we assumed feed flow rate, membrane cost for unit area, and feed composition based on industrial data or literature data. However, economic potential of these processes may change according to these values. For example, the membrane separation process can be more economically attractive with small feed flow rate because the capital cost of membrane separators is linearly related to membrane area in many cases while distillation process is more expensive in small scale processes. In this section, sensitivity analysis was carried out in order to identify economic influence of change in the flow rate and composition of the feed stream, and the membrane unit cost. The reference point for the membrane process is selected at stage-cut 0.9 with optimum permeance. Fig. 5 shows CAPEX and OPEX changes in accordance with mass flow rate of feed stream. The CAPEX for unit production of membrane process is more gradually decreased because membrane separator cost is linearly related to mass flow rate of the feed stream and it indicates that membrane process can be considered as a feasible option for propylene/propane separation when feed flow rate is not large enough. OPEX of the distillation process is higher regardless of the feed flow rate and it is resulted from high cooling water usage. Even though utility costs of both processes linearly increase

Page 10 of 27

Manuscript Prepared for Chemical Engineering and Processing: Process Intensification as flow rate changes, nonlinear behavior of fixed charge, labor cost, and plant overhead cost results in exponential decrease of OPEX as flow rate change. The total production cost and corresponding outlet pressure changes in accordance with the feed composition are presented in fig. 6. Note that the inlet propylene mole fraction of the membrane separator should be greater than 61% in order to prevent negative permeate stream pressure and it corresponds to 84.3 propylene mol% of the feed stream. The permeate stream pressure is decreased as more diluted feed stream is introduced to the process, and it requires high pressure ratio in product compressor. The high pressure ratio results exponential increases in both CAPEX and OPEX; and consequently, total production cost increases dramatically as the feed stream propylene concentration approaches to theoretical limit, 84.3 mol%. A low ΔPpy tremendously reduces the flux of propylene in permeate stream. Meanwhile, the permeance is also governed by temperature due to the Langmuir-sorption type interaction of carrier and propylene in FTM. The relatively low temperature can increase the propylene permeance, because the propylene binding equilibrium constant of the fixed carrier increases and desorption rate constant decreases with decrease in temperature. Fig. 7 indicates the total propylene production cost with respect to membrane unit cost. As one may easily imagine both CAPEX and OPEX are linearly depending on the membrane unit cost. Note that OPEX also shows the similar behavior and it is resulted from fixed charge which is a function of fixed capital investment. The result indicates that the total cost can be reduced by 32% when membrane unit cost is decreases to 10% of this study.

4

Conclusions and Outlook

This article highlights the techno-economic feasibility of one-stage membrane-based propylene/propane separation, which produces 99.6% purity of propylene with 97% recovery. In order to propose feasible membrane specification over wide range, we firstly optimized C3 distillation process with direct vapor recompression. The specifications of the membrane process are then calculated to satisfy the same propylene production cost of the distillation process with one stage membrane separation process modeling. The results reveal that the operations between 0.5 and 0.9 stage-cut and under 10 bar ΔPpy condition can have the same economic potential of propylene production using distillation column. The required membrane permeance and selectivity ranges between 11.3 to 251.5 GPU and 61.8 to 1950 depending on stage-cut. With very high stage cut, membrane separation may have higher economic potential than the distillation with reasonable membrane specifications (e.g., 11.3 GPU and selectivity of 68). Further improvements of the membrane separation process can be made with low membrane cost, sweep gas flowing method. In general, olefin separation FTMs have silver or sometimes copper as a carrier center. Thus, in order to reduce the cost of membrane, further efforts are needed to involve copper as an olefin carrier or search a cost attractive carrier. In industrial gas separation processes, an adjustment of maximizing ΔPpy is difficult. Therefore, it is challenging to develop the sweep gas flowing method in olefin/paraffin separation. This sweep gas must be easily removed when liquefaction of propylene or other simple manipulation.

Page 11 of 27

Manuscript Prepared for Chemical Engineering and Processing: Process Intensification

Appendix Table A 1 Energy and mass balance table of the VRC distillation process

Stream Name

1

2

3

5

6

4

7

8

9

13

12

Temperature (°C)

40.4

40.3

40.3

40.0

40.0

40.0

44.7

45.0

49.8

49.8

45.0

Pressure (bar)

16.7

16.7

16.7

16.7

16.7

16.7

16.7

16.7

20.8

20.8

16.7

Vapor Frac

1.0

1.0

0.0

1.0

0.0

1.0

0.0

1.0

0.1

1.0

0.0

Mass Flow (kg/hr)

46000.0

405585.7

363026.5

443642.5

21939.4

21939.4

373256.1

373256.1

421703.0

42549.1

3440.5

PROPANE

0.050

0.044

0.049

0.004

0.004

0.004

0.611

0.611

0.004

0.004

0.641

PROPYLENE

0.950

0.956

0.951

0.996

0.996

0.996

0.389

0.389

0.996

0.996

0.359

Mole Frac

Page 12 of 27

Manuscript Prepared for Chemical Engineering and Processing: Process Intensification

Nomenclature VRC = Direct vapor recompression process NCC = Naphtha Cracking Center FTM = Facilitated transport membranes HIDiC = Heat-integrated distillation column 𝑉𝐹𝐹 = Vapor fraction 𝑁𝐹 = Column Feed Stage 𝐹𝐶𝑜𝑚𝑝 = Compressor inlet stream feed flow rate (ton/hr)) 𝑃𝐶𝑜𝑙 = Column top pressure (bar) 𝜃 = Stage cut 𝑗 = molar flux (mole/cm2) 𝑓

𝑝𝑖 = partial pressure of component i in feed stream (bar) 𝑝𝑖𝑃 = partial pressure of component i in permeate stream (bar) 𝑙 = Membrane thickness (cm) 𝑃𝑖 = permeability 𝛼 = Selectivity of membrane 𝑥𝑖,𝑝𝑒𝑟𝑚 = mole fraction of component i

References [1] Eldridge RB. Olefin/paraffin separation technology: a review. Industrial & engineering chemistry research. 1993;32(10):2208-12. [2] Ravanchi MT, Kaghazchi T, Kargari A. Application of membrane separation processes in petrochemical industry: a review. Desalination. 2009;235(1):199-244. [3] Wang K, Stiefel EI. Toward separation and purification of olefins using dithiolene complexes: An electrochemical approach. science. 2001;291(5501):106-9. [4] Jarvelin H, Fair JR. Adsorptive Separation of Propylene Propane Mixtures. Industrial & Engineering Chemistry Research. 1993;32(10):2201-7. [5] Pitsch F, Krull FF, Agel F, Schulz P, Wasserscheid P, Melin T, et al. An Adaptive Self‐Healing Ionic Liquid Nanocomposite Membrane for Olefin‐ Paraffin Separations. Advanced Materials. 2012;24(31):4306-10.

Page 13 of 27

Manuscript Prepared for Chemical Engineering and Processing: Process Intensification

[6] Baker RW, Low BT. Gas separation membrane materials: A perspective. Macromolecules. 2014;47(20):6999-7013. [7] Robeson LM. The upper bound revisited. Journal of Membrane Science. 2008;320(1):390-400. [8] Staudt-Bickel C, Koros WJ. Olefin/paraffin gas separations with 6FDAbased polyimide membranes. Journal of Membrane Science. 2000;170(2):20514. [9] Nikolakis V, Xomeritakis G, Abibi A, Dickson M, Tsapatsis M, Vlachos DG. Growth of a faujasite-type zeolite membrane and its application in the separation of saturated/unsaturated hydrocarbon mixtures. Journal of Membrane Science. 2001;184(2):209-19. [10] Giannakopoulos IG, Nikolakis V. Separation of propylene/propane mixtures using faujasite-type zeolite membranes. Industrial & engineering chemistry research. 2005;44(1):226-30. [11] Kwon HT, Jeong H-K. In situ synthesis of thin zeolitic–imidazolate framework ZIF-8 membranes exhibiting exceptionally high propylene/propane separation. Journal of the American Chemical Society. 2013;135(29):10763-8. [12] Kwon HT, Jeong H-K. Highly propylene-selective supported zeoliteimidazolate framework (ZIF-8) membranes synthesized by rapid microwaveassisted seeding and secondary growth. Chemical Communications. 2013;49(37):3854-6. [13] Pan Y, Li T, Lestari G, Lai Z. Effective separation of propylene/propane binary mixtures by ZIF-8 membranes. Journal of membrane science. 2012;390:93-8. [14] Ploegmakers J, Japip S, Nijmeijer K. Mixed matrix membranes containing MOFs for ethylene/ethane separation Part A: Membrane preparation and characterization. Journal of membrane science. 2013;428:445-53. [15] Zhang C, Dai Y, Johnson JR, Karvan O, Koros WJ. High performance ZIF8/6FDA-DAM mixed matrix membrane for propylene/propane separations. Journal of Membrane Science. 2012;389:34-42.

Page 14 of 27

Manuscript Prepared for Chemical Engineering and Processing: Process Intensification

[16] Dong G, Li H, Chen V. Challenges and opportunities for mixed-matrix membranes for gas separation. Journal of Materials Chemistry A. 2013;1(15):4610-30. [17] Suda H, Haraya K. Alkene/alkane permselectivities of a carbon molecular sievemembrane. Chemical Communications. 1997(1):93-4. [18] Ma X, Williams S, Wei X, Kniep J, Lin Y. Propylene/Propane Mixture Separation Characteristics and Stability of Carbon Molecular Sieve Membranes. Industrial & Engineering Chemistry Research. 2015;54(40):9824-31. [19] Li H, Song Z, Zhang X, Huang Y, Li S, Mao Y, et al. Ultrathin, molecularsieving graphene oxide membranes for selective hydrogen separation. Science. 2013;342(6154):95-8. [20] Li Y, Wang S, He G, Wu H, Pan F, Jiang Z. Facilitated transport of small molecules and ions for energy-efficient membranes. Chemical Society Reviews. 2015;44(1):103-18. [21] Sunderrajan S, Freeman B, Hall C, Pinnau I. Propane and propylene sorption in solid polymer electrolytes based on poly (ethylene oxide) and silver salts. Journal of Membrane Science. 2001;182(1):1-12. [22] Jose B, Ryu JH, Kim YJ, Kim H, Kang YS, Lee SD, et al. Effect of plasticizers on the formation of silver nanoparticles in polymer electrolyte membranes for olefin/paraffin separation. Chemistry of materials. 2002;14(5):2134-9. [23] Chae IS, Kang SW, Kang YS. Olefin separation via charge transfer and dipole formation at the silver nanoparticle–tetracyanoquinoid interface. RSC Advances. 2014;4(57):30156-61. [24] Chae IS, Kang SW, Park JY, Lee YG, Lee JH, Won J, et al. Surface Energy‐Level Tuning of Silver Nanoparticles for Facilitated Olefin Transport. Angewandte Chemie. 2011;123(13):3038-41. [25] Song D, Kang YS, Kang SW. Highly permeable and stabilized olefin transport membranes based on a poly (ethylene oxide) matrix and Al (NO 3) 3. Journal of Membrane Science. 2015;474:273-6.

Page 15 of 27

Manuscript Prepared for Chemical Engineering and Processing: Process Intensification

[26] Huang Y, Merkel TC, Baker RW. Pressure ratio and its impact on membrane gas separation processes. Journal of Membrane Science. 2014;463:33-40. [27] Sridhar S, Khan A. Simulation studies for the separation of propylene and propane by ethylcellulose membrane. Journal of membrane science. 1999;159(1):209-19. [28] Motelica A, Bruinsma OS, Kreiter R, den Exter M, Vente JF. Membrane Retrofit Option for Paraffin/Olefin Separation A Technoeconomic Evaluation. Industrial & Engineering Chemistry Research. 2012;51(19):6977-86. [29] Alcántara-Avila JR, Gómez-Castro FI, Segovia-Hernández JG, Sotowa KI, Horikawa T. Optimal design of cryogenic distillation columns with side heat pumps for the propylene/propane separation. Chemical Engineering and Processing: Process Intensification. 2014;82:112-22. [30] Chen D, Yuan X, Xu L, Yu K. Comparison between different configurations of internally and externally heat-integrated distillation by numerical simulation. Industrial & engineering chemistry research. 2013;52(16):5781-90. [31] Olujić Ž, Sun L, De Rijke A, Jansens P. Conceptual design of an internally heat integrated propylene-propane splitter. Energy. 2006;31(15):3083-96. [32] Ho T-J, Huang C-T, Lin J-M, Lee L-S. Dynamic simulation for internally heat-integrated distillation columns (HIDiC) for propylene–propane system. Computers & Chemical Engineering. 2009;33(6):1187-201. [33] Yee TF, Grossmann IE. Simultaneous optimization models for heat integration—II. Heat exchanger network synthesis. Computers & Chemical Engineering. 1990;14(10):1165-84. [34] Navarro-Amorós MA, Caballero JA, Ruiz-Femenia R, Grossmann IE. An alternative disjunctive optimization model for heat integration with variable temperatures. Computers & Chemical Engineering. 2013;56:12-26. [35] Burri JF, Wilson SD, Manousiouthakis VI. Infinite DimEnsionAl Statespace approach to reactor network synthesis: application to attainable region construction. Computers & chemical engineering. 2002;26(6):849-62.

Page 16 of 27

Manuscript Prepared for Chemical Engineering and Processing: Process Intensification

[36] Viswanathan J, Grossmann IE. Optimal feed locations and number of trays for distillation columns with multiple feeds. Industrial & engineering chemistry research. 1993;32(11):2942-9. [37] Yeomans H, Grossmann IE. A systematic modeling framework of superstructure optimization in process synthesis. Computers & Chemical Engineering. 1999;23(6):709-31. [38] Raman R, Grossmann IE. Symbolic integration of logic in mixed-integer linear programming techniques for process synthesis. Computers & Chemical Engineering. 1993;17(9):909-27. [39] Lee U, Mitsos A. Two Stage Optimization of Mixed Working Fluid Selection for Low Temperature Organic Rankin Cycle Energy. 2016. [40] Lee U, Burre J, Caspari A, Kleinekorte J, Schweidtmann AM, Mitsos A. Techno-economic Optimization of a Green-Field Post-Combustion CO2 Capture Process Using Superstructure and Rate-Based Models. Industrial & Engineering Chemistry Research. 2016;55(46):12014-26. [41] Trespalacios F, Grossmann IE. Review of Mixed‐Integer Nonlinear and Generalized Disjunctive Programming Methods. Chemie Ingenieur Technik. 2014;86(7):991-1012. [42] Zimmermann H, Walzl R. Ethylene. In: Wiley I, editor. Ullmann's Encyclopedia of Industrial Chemistry. Weinheim, Germany: Weinheim, Germany : Wiley-VCH; 2012. [43] Zimmermann H. Propene. In: Wiley I, editor. Ullmann's Encyclopedia of Industrial Chemistry. Weinheim, Germany: Weinheim, Germany : Wiley-VCH; 2013. [44] Neagu M, Cursaru DL, MOVILEANU DL, DINU F. Thermal Energy Approach of a Conventional Propylene/propane Splitter. REVISTA DE CHIMIE. 2013;64(8):880-5. [45] Olujić Ž, Sun L, Gadalla M, De Rijke A, Jansens P. Enhancing thermodynamic efficiency of energy intensive distillation columns via internal heat integration. Chemical and Biochemical Engineering Quarterly. 2008;22(4):383-92.

Page 17 of 27

Manuscript Prepared for Chemical Engineering and Processing: Process Intensification

[46] Biegler LT, Grossmann IE, Westerberg AW. Systematic methods for chemical process design. 1997. [47] Hasan MF, Baliban RC, Elia JA, Floudas CA. Modeling, simulation, and optimization of postcombustion CO2 capture for variable feed concentration and flow rate. 1. Chemical absorption and membrane processes. Industrial & Engineering Chemistry Research. 2012;51(48):15642-64. [48] Douglas JM. Conceptual design of chemical processes: McGraw-Hill New York, 1988. [49] Peters MS, Timmerhaus KD, West RE, Timmerhaus K, West R. Plant design and economics for chemical engineers: McGraw-Hill New York, 1968. [50] Seider WD, Seader JD, Lewin DR. PRODUCT & PROCESS DESIGN PRINCIPLES: SYNTHESIS, ANALYSIS AND EVALUATION, (With CD): John Wiley & Sons, 2009. [51] Lee U, Burre J, Caspari A, Kleinekorte J, Schweidtmann A, Mitsos A. Techno-economic Optimization of a Green-Field Post-Combustion CO2 Capture Process using Superstructure and Rate-Based Models. Industrial & Engineering Chemistry Research. 2016. [52] Lee U, Mitsos A, Han C. Optimal retrofit of a CO 2 capture pilot plant using superstructure and rate-based models. International Journal of Greenhouse Gas Control. 2016;50:57-69. [53] Na J, Kshetrimayum KS, Lee U, Han C. Multi-objective optimization of microchannel reactor for Fischer-Tropsch synthesis using computational fluid dynamics and genetic algorithm. Chemical Engineering Journal. 2017;313:1521-34. [54] Alabdulkarem A, Mortazavi A, Hwang Y, Radermacher R, Rogers P. Optimization of propane pre-cooled mixed refrigerant LNG plant. Applied Thermal Engineering. 2011;31(6–7):1091-8.

Page 18 of 27

Manuscript Prepared for Chemical Engineering and Processing: Process Intensification

Figures

Fig. 1. The schematic of the Naphtha Cracking Center(NCC) plant Page 19 of 27

Manuscript Prepared for Chemical Engineering and Processing: Process Intensification

Fig. 2. Process flow diagram of the VRC distillation process.

Page 20 of 27

Manuscript Prepared for Chemical Engineering and Processing: Process Intensification

4

5

Recycle to feedstocks

Rec Comp 3 6 Cooler

7 From Propyne Converter

1

2

Membrane Separator

8

Prod Comp 9 To Propene storage

Fig. 3. Process flow diagram of the membrane separation process.

Page 21 of 27

Manuscript Prepared for Chemical Engineering and Processing: Process Intensification

Fig. 4. CAPEX, OPEX and specifications of membrane process with respect to stage-cut and ΔPpy in order to meet the same economic potential of the C3 distillation process.

Page 22 of 27

Manuscript Prepared for Chemical Engineering and Processing: Process Intensification

Fig. 5. Cost change of membrane and distillation process in accordance with the process capacity

Fig. 6. Total cost of the membrane process with different propylene feed concentration Page 23 of 27

Manuscript Prepared for Chemical Engineering and Processing: Process Intensification

Fig. 7. Total cost of membrane process with respect to membrane unit cost

Page 24 of 27

Manuscript Prepared for Chemical Engineering and Processing: Process Intensification Tables Table 1. Input specifications of the VRC C3 splitter Parameters Feed composition (propylene) Top product composition (propylene) Top product propylene recovery Feed flow rate Feed pressure Total number of stage

Value 95.0 99.6 97 46 20 200

Unit mol% mol% mol% ton/h Mpa stage

Table 2. Optimization result of the distillation process Variable

Value

Unit

SPC y (Feed stage)

18.6 165

$/ton propylene

x1 (Feed vapor fraction)

1

x2 (Compressor pressure ratio)

1.24

x3 (Compressor inlet flow rate)

422

Table 3. Equipment cost of VRC distillation process Type Of Equipment Distillation Column1 Distillation Column2 Propylene Flash Drum Propane Flash Drum Overhead Cooler MVR Heat exchanger Compressor

Cost (M$) 2.26 2.48 0.18 0.10 0.16 1.69 1.14

Total

8.00

Page 25 of 27

ton/h

Manuscript Prepared for Chemical Engineering and Processing: Process Intensification Table 4. Total capital cost investment of VRC distillation process Percentage of FCI

Used

Cost (M$)

0.60

12.00

Direct cost ISBL Purchased equipment

20-40%

0.40

8.00

Purchased equipment installation

7.3-26%

0.12

2.40

Instrumentation and control

2.5-7.0%

0.03

0.50

Piping

3-15%

0.03

0.60

Electrical

2.5-9.0%

0.03

0.50

0.20

4.00

OSBL Building and building services

6-20%

0.07

1.38

Yard improvements

1.5-5.0%

0.02

0.48

Services facilities

8.0-35.0%

0.09

1.78

Land

1-2%

0.02

0.38

0.80

16.01

Total direct cost Indirect cost Engineering

4-21%

0.06

1.23

Construction expenses

4.8-22.0%

0.07

1.43

Contractor’s fee

1.5-5.0%

0.02

0.30

Contingency

5-20%

0.05

1.03

0.20

4.00

Total indirect cost Fixed capital investment (FCI)

1.00

1.00

20.01

Working Capital

10-20%

0.10

2.00

Start-Up Cost

8-10%

0.08

1.60

Total Capital Investment (CAPEX)

23.61

Page 26 of 27

Manuscript Prepared for Chemical Engineering and Processing: Process Intensification Table 5. Total operating expenses of VRC distillation process Range

Used

Cost (M$)

Local Taxes

1-4% of FCI (Peter et al.)

1.0%

0.24

Insurance

0.4-1% of FCI (Peter et al.)

0.4%

0.09

Production Cost Fixed Charge

Direct Production Cost Raw Material

0.00

Cooling Water

0.64

Electricity

0.97

Maintenance

30% of Total Labor Cost

0.25

Operating Labor (OL)

Aspen Process Economic Analyzer

0.83

Supervision and support labor

30% of Total Labor Cost

0.36

Operating Supplies

15% of Maintenance

15.0%

0.04

Laboratory charges

15% OL

15.0%

0.12

50-70% of (M+S+OL)

50.0%

0.72

Administrative Cost

15-20 %of OL

15.0%

0.12

Distribution and Marketing

2-20% Production Cost

2.0%

0.09

R&D Cost

2-5%Production Cost

2.0%

0.09

Plant Overhead Cost General Expenses

Total Production Cost (OPEX)

4.58

Table 6. Optimum permeance of propylene and propane with respect to stage-cut Stage-Cut

Permeate Pressure (bar)

Ppropylene (GPU)

Ppropane (GPU)

0.5 0.6 0.7 0.8 0.9

5.91 5.84 4.97 4.41 4.01

90.0 40.8 24.3 15.8 11.3

0.046 0.095 0.102 0.117 0.166

Page 27 of 27