Design and Economic Evaluation of a Hybrid Membrane Separation Process from Multiple Refinery Gases Using a Graphic Synthesis Method

Abstract

Petrochemical tail gases have various components and many separation methods, thus there are many possible design schemes, making it difficult to determine the optimal scheme. In this work, a graphic synthesis method was used to design a hybrid multi-input refinery gas separation process consisting of membranes, pressure swing adsorption (PSA), shallow condensation (SC), and distillation units for the production of valuable products which include H₂, C₂, LPG, and C₅+. Ten refinery gases with different compositions were visualized and represented with vector couples in a triangular coordinate system. Firstly, according to the characteristics of the refinery gases, the feeds located in the same region of the triangular coordinate system were merged to simplify the number of input streams, then ten original input streams were combined into two mixed streams. Secondly, the optimal separation sequence was determined by using the unit selection rules of a graphic synthesis method. Thirdly, the process was simulated in UniSim Design and the process parameters were determined by sensitivity analysis. Finally, economic assessments were carried out, which led to an annual gross product profit of USD 38.62 × 10⁶ and a payback period of less than 4 months.

1. Introduction

Refinery gas is a mixture of gases generated during refinery processes which are performed to process crude oil into various petroleum products, which can then be traded or sold. They contain profitable components such as hydrogen and light hydrocarbons [1] that are normally burned, together with combustible gases and residual gases in furnaces. It is obvious that these profitable components can be used more reasonably after effective separation rather than direct combustion [2,3,4,5]. The composition of refinery gas varies depending on the composition of the crude oil it originates from and the processes to which it has been subjected. Common components include hydrogen, C1–C4 alkanes, C2–C4 olefins, a small amount of C5 paraffins, and impurities such as hydrogen sulfide and carbon dioxide. Commonly, C2–C4 olefins can be used as raw materials to produce other petrochemical products, while C2–C4 alkanes can be used as feedstock for the production of ethylene by a steam cracking process [6].

Hydrogen is an essential resource used in many processes, such as hydrotreating and hydrocracking. A variety of factors, including increasingly stringent environmental regulations, have increased the demand for higher-quality fuels, and the use of high-sulfur crude oil and heavier oil have forced refineries to process crude oil more thoroughly, causing hydrogen consumption to skyrocket [7,8,9,10]. Although steam methane reforming (SMR) is the dominant industrial process for hydrogen production, environmental concerns related to CO₂ emissions, along with process intensification and energy optimization, are areas that still require improvement [11]. Depending on the H₂ feed composition, product flowrate, reliability, required output purity, and costs, the selection of the most advantageous technique is done [12]. Membrane-based separation processes, a cryogenic process, and pressure swing adsorption (PSA) have been suggested as alternative technologies to reduce the cost and energy requirements of H₂ production as compared to steam methane reforming [13,14,15,16,17].

Light hydrocarbons (LHCs), namely C1-C9 hydrocarbons, which mainly originate from natural or petrochemical gas processing, are essential energy resources and raw materials for the production of some industrially important chemicals [18]. The separation of LHCs into a pure species is a very critical and challenging industrial process and has always highly concerned scientists. Typically, distillation, the use of rubbery organic vapor membranes, and absorption are the three main conventional methods for LHCs purification in the process of industrial production [19]. Refineries may recover LHCs from off-gases, and thus make more profit on these products instead of simply using them as fuel gas [20,21,22,23,24].

In the recovery process of profitable compositions in refinery tail gas, due to the large number of streams, complex components, and various separation methods, it is very difficult to determine the optimal separation sequence. Many heuristic methods [25] based on engineering experience, in addition to evolutionary methods [26] by way of enhancing preliminary processes, have been proposed to determine separation sequence for the process. For example, Bausa et al. [27] presented a shortcut design method for hybrid membrane-distillation applicable to multi-component, non-ideal mixtures, and Kao et al. [28] obtained the optimal separation sequence according to modified cost measures, in contrast to an easy-to-use, matrix-based method for the systematic synthesis of distillation configurations as described by Shenvi et al. [29]. However, heuristic methods can only obtain the near-optimal sequence without checking all possible sequences, and they make it difficult to obtain the global optimal solution. Mathematical programming methods and intelligent algorithms for separation sequence synthesis, such as the mixed integer nonlinear programming (MINLP) model and genetic algorithm (GA), are too complex and difficult to calculate the optimal sequence because professional mathematics skills are required. Thus, a graphic synthesis method presents an opportunity to save equipment investment and operation costs by sharing equipment and facilities, along with saving time and effort in synthesis, thereby ensuring better separation sequences [30].

The information required to generate configurations in a multi-input separation process includes the number of components in feed and their compositions, as well as the number of distinct compositions in the final product stream. Separation unit operations are the basic elements used to construct the separation process. In this context, a typical multi-input component separation process consists of fractionating the feedstock into desirable multi-product at minimum operation costs.

Integrated processes for executing refinery gas separation tasks could be regarded as the typical multi-input and multi-component problem. Together with the development of novel membrane materials with improved separation performances, process design becomes a key issue in the economics of a membrane-based separation process [31,32]. In this context, optimization of both the membrane module and the whole membrane-based process is the main concern for improving the performance of this separation technology. Certainly, they can easily be used for testing and providing valuable information about the sizes of the process units, as well as the operating conditions of the entire process flowsheet in a short time [33,34,35]. In conjunction with material science and the process system engineering field, this study widens an important research area towards the improvement of a membrane-based separation processes.

In this study, a graphic synthesis method was used to design a novel appended process of H₂/LHCs production based on the dominant regions of the common technologies for treating refinery gases. The method, which is a shortcut based on the representation of operations by vectors in the composition of space, was proposed in this research to give a rapid response to a multi-input feed system. Moreover, attention was particularly paid to the influence of key parameters such as membrane area, stage cut, and permeate pressure of the process design. A further economic assessment stipulates that the novel process’s total investment and yearly total cost of production brings in an annual product profit of USD 38.62 × 10⁶, and thus is more economically feasible.

2. Process Design

2.1. Problem Description

A hybrid process of membranes and PSA, coupled with shallow condensation (SC), was designed to recover hydrogen and LHCs from 10 streams of refinery gases (from a SINOPEC Petrochemical Company in Luoyang, China), as shown in

Table 1. Data of the main tail gases in a petrochemical company in China.

Streams	1	2	3	4	5	6	7	8	9	10
Temperature (°C)	40	40	40	40	40	40	40	40	40	40
Pressure (MPag)	2.50	2.40	2.50	0.75	0.67	0.86	1.20	1.15	0.58	0.7
Flowrate (Nm3·h−1)	600	2788	2415	1326	2053	1229	1034	600	900	1200
Mole fraction %
H2	98.43	63.34	79.95	31.23	34.42	13.45	35.01	75.78	3.49	69.45
N2	0.00	0.00	0.00	2.04	0.00	0.00	0.00	0.00	0.00	0.41
O2	0.00	0.00	0.00	0.23	0.00	0.00	0.00	0.00	0.00	0.50
CH4	0.88	14.25	8.92	12.17	10.19	2.60	5.64	4.27	1.19	3.35
C2H6	0.23	11.15	4.98	20.21	15.97	39.90	24.04	2.44	5.31	23.94
C3H8	0.08	4.73	2.51	14.45	14.61	38.79	20.23	0.69	30.92	2.38
i-C4H10	0.02	1.09	0.55	3.91	4.03	2.39	6.86	0.19	16.66	0.12
n-C4H10	0.02	1.11	0.99	3.71	7.78	2.31	4.45	1.31	40.36	0.20
C5+	0.07	1.02	1.11	3.29	8.16	0.53	3.56	14.48	0.00	0.12
H2S	0.10	3.31	1.00	8.78	4.78	0.00	0.04	0.85	1.95	1 μL/l

. The designed process uses H₂ selective membranes (GPM) and PSA to recover and purify hydrogen, respectively, by appending rubbery organic vapor membranes (RPM) and shallow condensation to the design for LHCs recovery.

Due to the combinative nature of process design, the difficulty in synthesizing an optimum, multi-input production process is inadequately exasperated, along with the increase in the available separation methods and number of components. In the case of using M methods to separate a stream of N- component mixture to yield N streams of pure products, the number of candidate production sequences can be given by [25,30]:

$$\frac{\left[2\left(N-1\right)\right]!}{N!\left(N-1\right)!}{M}^{N-1}$$

(1)

Therefore, for the synthesis of a separation sequence for 10 streams and 4 separation methods, the number of candidate schemes is 1.27 × 10⁹. Although most schemes can be removed according to empirical rules, this sequence synthesis is still a complex combinatorial arrangement problem.

2.2. Process Design Based on a Graphic Synthesis Method

2.2.1. Representation and Merging of Feed Gas Streams Based on a Triangular Coordinate System

A steady, complex selection that combines similar refinery gases into various input streams is proposed to reduce the stream number involved in the design. Before blending the feed streams, a satisfactory standard is demanded to grade refinery gas. The dominant ranges of separation techniques can be used as the ready-made criterion demonstrated by Ruan et al. [30]. According to the concept for dividing these ranges, refinery gases stationed in the same dominant range can be treated proficiently by the correlating technique, and therefore they can be graded together as a cluster. The refinery gas territory for incorporating the multi-input process and the 10 gas streams, each located in the region of separation, can be seen in Figure 1. The designed process converts the 10 streams into 2 streams, where streams 1, 2, 3, 8, and 10, in line with the component composition of each stream, combine to form the mixed stream F1, and streams 4, 5, 6, 7, and 9 form the mixed stream F2. This is because streams 1, 2, 3, 8, and 10 are in the separation advantage area of PSA, whereas streams 4, 5, 6, 7, and 9 are in the SC separation advantage area. Therefore, the complex problem of 10 input streams is simplified to 2 mixed streams, which greatly reduces the difficulty of the process. Suitable separation unit operations are selected, then converted into a reasonable process sequence, and with these procedures, the complicated work to select the optimum sequence from numerous candidates can be greatly reduced.

2.2.2. Separation Sequence Determination and Process Design

The synthesis starts by placing these input streams in the coordinate system, as shown in Figure 2, in which P and R relate to the product and the residue, respectively. The two inputs (F1 and F2) are located in the ranges of two different separation techniques (PSA and SC), and then primary operations for separations are determined: PSA separates H₂ in F1, and SC separates LHCs in F2. These primary operations are represented by the vector couples in Figure 2, and the compositions of their residual streams are appraised by the approximate technique suggested in Figure 1. In the majority of cases, the valuable species in the residual gases are still substantial and further recoveries are required (or attainable). The desorbed gas from PSA, denoted as R₁ and located in the territory of the glassy polymer membranes (GPM), is mixed with R₂ from the input stream F2, and the new mixed point, denoted M₂, is sent to a GPM, which in this case is polyimide. GPM is efficient in treating refinery gases with a lower H₂ input content in the range of 30 mol% to 60 mol%. The H₂ produced from GPM containing 96 mol%, denoted P₃ and located in the PSA territory, is mixed with F1, forming a new mixed point denoted M₁. The residue rich in LHCs, denoted R₃, with the aid of a SC system, produces R₄, located in the RPM territory, which is sent to the rubbery polymer membrane (RPM). A further separation takes place in this membrane, and the permeate denoted as P₅ is recycled while the retentate denoted R₅ produces fuel gas, as depicted in Figure 3. The membrane description, along with permeance and in addition to the main separation technologies, can be found in Appendix D. The permeate P₂, rich in in LHCs, with a content higher than 95 mol% according to the position in coordinate system, together with P₄ of the second condensation system, form the mixed stream M₃. The point M₃ is then sent to the distillation columns. Finally, the separation flowsheet designed can be seen in Figure 3.

Basic modules, such as compressors, heat exchangers, distillation columns, etc., were included in the simulation process in this work, which was done in UniSim Design. In the case of the membrane module, our former groupmate programmed the precompiled dynamic link library (DLL) file, which included the algorithms and external definition file (EDF) defining the portable document format (PDF) icon and user interface in UniSim Design, allowing for the easy and quick simulation of gas membrane separation [36]. The simulation model was constructed in UniSim Design and was based on the design criteria, as well as the basic parameters, and the Peng–Robinson equation was chosen as the state equation. This equation was chosen as there is no other simple Van der Waals equation of state that has demonstrated such broad and consistent applicability to the calculation of vapor–liquid equilibria (VLE) for systems containing light hydrocarbons, permanent gases, carbon dioxide, and hydrogen sulfide [22,30]. The simulation process flow diagram of the illustrative multi-input separation process in UniSim Design is illustrated in Figure 4.

In Figure 4, streams 8 and 10, with pressures of 1.15 and 0.7 MPag, are initially mixed together, then pressurized to 2.4 MPag by compressor K-1 and cooled to 40 °C, as there will be a significant pressure loss if mixed immediately with streams 1, 2, and 3. The compressors run with an isentropic efficiency equal to 75% in simulation. The newly formed stream 11 is mixed with streams 1, 2, and 3 utilizing V-1 flash separator to produce the first mixed stream F1, whereas streams 4, 5, 6, 7, and 9, with similar pressures of 0.75, 0.65, 0.86, 1.30, and 0.58 MPag, respectively, are mixed using a flash separator V-2 to produce the second mixed stream F2. The mixed streams F1 and F2 are first introduced to the adsorption towers (A-1 and A-2, respectively) with the aim of removing H₂S impurities. Secondly, the F1 and F2 H₂S free streams (13 and 18, respectively) are sent to their primary unit operations, as designed. F1 is conveyed to the PSA (PSA data can be found in

Table A2. PSA separation performance.

	Feed Gas	Product	Tail Gas
Flow (kmol.h−1)	494.4	317.5	176.9
Pressure (kPa)	2500	2000	650
Temperature (°C)	39.78	39	38.65
Mole fraction %
H2	80.2	99.91	44.83
CH4	7.74	0.09	21.48
C2H6	7.13	0	19.93
C3H8	2.33	0	6.5
C4H10	1.09	0	3.07
C5+	1.38	0	3.85

) to procure high purity H₂ (99.9 mol%), while F2 enters the condensation system, with the temperature of E-1 being −40 °C for H₂ and LHCs production. The gaseous phase stream 20 from the condensation with approximately 50 mol% H₂ mixes with the residual stream 16 of the PSA containing 40 mol% H₂. The newly formed stream 22 is pressured by K-2, then introduced into the GPM. Approximately 96 mol% of H₂ is recovered from the permeate stream of the GPM, re-pressurized by K-3, cooled, and then mixed with stream 13, which enters the PSA for further H₂ purification. The residual stream of GPM containing 63.53 mol% LHCs is fed to the second condensation system, E-3, at a temperature of −30 °C, and the vapor stream after passing through flash V-4 is heated by E-4 to 40 °C before being introduced into the second membrane unit, RPM. In this membrane, the LHCs produced via the permeate stream 33 are first pressurized to about 2.4 MPag by compressor K-4 and then cooled and recycled, whereas the residue is collected as fuel gas. Finally, the liquid phases (the bottom streams 21 and 31 from V-1 and V-2) of the first and second condensation systems are mixed together and then conveyed to the distillation systems.

3. Sensitivity Analysis

3.1. Hydrogen and Light Hydrocarbon Production

The graphic synthesis design consists of establishing the optimal unit operations and their interconnections. The work done here was able to produce ≥ 99.91 mol% H₂ and 99.8 mol% of LHCs, as shown in

Table 2. Overall mass balance of the illustrative multi-input recovery process.

Stream	13	18	H2	C2H6	LPG	C5+	FG
Pressure/kPa	2500	680	2000	2150	1400	1450	2100
Temperature/°C	40	31	39	−7.6	56.68	142.3	−3
Rate/kmol·h−1	333.5	281.7	317.4	94.77	126.4	17.23	59.17
Mole fraction %
H2	74.48	26.63	99.91	0	0	0	10.27
N2	0.06	0.43	0	0	0	0	2.42
O2	0.08	0.05	0	0	0	0	0.68
CH4	9.15	7.48	0.09	0.89	0	0	84.58
C2H6	9.77	21.93	0	98.28	0	0	2.04
C3H8	3.02	23.09	0	0.83	58.94	0.03	0.01
i-C4H10	0.63	6.10	0	0	15.26	0.44	0
n-C4H10	0.87	10.26	0	0	24.85	2.82	0
n-C5H12	1.93	4.04	0	0	0.94	96.69	0
H2S	0.00	0.00	0	0	0	0	0

. Importantly, the ethane produced is meant for ethylene production through a steam cracking process.

3.2. Membrane Area

3.2.1. Impacts of RPM Area on the Purity and Recovery of C₂+

Figure 5 shows the effect of the RPM membrane area on the purity and recovery of C₂+. When the RPM membrane area is more than 600 m², the C₂+ recovery is almost at a constant value (99.5 mol%), while at the same time, the C₂+ purity decreases slowly. LHC/H₂ permeation across the RPM membrane was found to be notably affected by the sorption behavior of the permeant, and thus characterized as a sorption-selective process. In addition, the RPM membrane has a very high selectivity with regard to LHC due to diluting the H₂ component, with LHC component selectively absorbed in the membrane, thereby blocking permeation. The area with the best performance was then decided to be 600 m².

3.2.2. Impacts of GPM Area on the Purity and Recovery of Hydrogen

The area of the membrane dictates the total investment for the membrane module and the product standard; therefore, we studied the effect of the hydrogen membrane (GPM) area on H₂ purity and recovery to ascertain the appropriate GPM area. The results are shown in Figure 6. As seen in the graph, an increase in membrane area was followed by a decrease in hydrogen purity accompanied by an increase in hydrogen recovery. The decrease in H₂ purity and increase in its recovery stems from the fact that increasing the membrane area of GPM increases the permeation amount of the fast gas component, which would also lead to an increase in the permeation amount of the slow gas component. Further, when the area is large enough, the recovery will no longer vary with the membrane area. On the basis of Figure 6 and the separation necessity, the GPM optimized area was determined to be 3500 m² in order to keep the purity of hydrogen entering the PSA unit greater than 90 mol%.

3.3. Impacts of Stage Cut on the Purity and Recovery of C₂+ and H₂

Figure 7a,b shows the influence of stage cuts on the traditional GPM and RPM modules on the purity and recovery of C₂+ and H₂ at a pressure of 2400 kPa and operating temperature of 40 °C. It could be observed that both the H₂ and C₂+ purity decrease with increase of stage cuts in the GPM and RPM membranes. At a high stage cut, most of the feed permeated the membranes, which indicates the high purity of H₂ and C₂+. However, low stage-cuts are needed to acquire high recoveries of H₂ and C₂+. Therefore, the driving forces of H₂ and C₂+ permeation was reduced and H₂ and C₂+ purity decreased. The selected stage cuts for GPM and RPM are 0.51 and 0.8, respectively. Alongside the change in concentration and recovery, it is important to mention that a smaller stage-cut reduces the size of further gas-cleaning utilities.

3.4. Impacts of Permeate Pressure on the Purity and Recovery of C₂+ and H₂

The feed and permeate pressures attainable in most gas separations are restricted by economics. In industrial gas separation processes, adjustment of the pressure ratio, which is the ratio of feed to the permeate, comes at a significant price in terms of compressor capital cost and power usage. However, in this case, the permeate pressure is an important process variable since the efficacy of the H₂ and C₂+ produced by the membrane process depends, in part, on the pressure at which H₂/C₂+ is produced. We should note that increased pressure has one important effect: it is a higher driving force for permeation. Consequently, increasing the pressure has a positive effect in decreasing the total cost. In Figure 8 below, we observe that an increase in permeate pressure leads to a sharp decrease in H₂ and C₂+ recovery, while maintaining a stable but modest increase of H₂ and C₂+ purity. This leads to permeate pressures of 250 and 200 kPa for GPM and RPM, respectively, which is a good compromise.

3.5. Design and Optimization of Distillation Columns

Condensable hydrocarbons obtained by shallow condensation still contain valuable components, including methane, ethane, propane, butane, and pentane. In order to further enhance the added value of refinery gas and improve the economic benefits of the refinery, a distillation method was used in the process of condensable hydrocarbon separation.

According to the differences in the content of different hydrocarbon components in condensable hydrocarbons, the sequential three-tower separation process was adopted in the separation sequence design of condensable hydrocarbon. The feed stream (stream 38) is shown in Figure 4. With a temperature of −34.7 °C and a molar flowrate of 257 kgmole/h, it first enters the de-ethane column T-1 to remove the light components (CH₄, C₂H₆) at the top, which then enters the de-methane column T-2 for further separation. As the tower pressure rises, so does the condenser temperature, allowing us to use less-expensive refrigerants (such as cooling water or air). However, if the pressure is too high, the equipment requirements will rise as well (i.e., a thicker tower wall steel plate will be required). Furthermore, increasing the pressure reduces the relative volatility between components, requiring a higher reflux ratio and higher energy consumption to meet product separation requirements, which is unquestionably uneconomical. We must, therefore, weigh their relationship while choosing the appropriate operating pressure. Fuel gas and ethane products are obtained from the top and bottom of T-2, respectively. The T-1 bottom stream enters the de-pentane column T-3 for further refining to obtain LPG (C₃H₈, C₄H₁₀) and pentane products. It must be noted that the design and initial simulation of the columns in this work follow similar procedures in the literature [37,38], and the economic specifications have been described in Appendix A.

We carried out parameter sensitivity analysis to examine the effects of the significant process parameters on the distillation columns for the high product design. All tower tray efficiencies are 0.8. The results of the sensitivity analysis for the T-1 column are presented in Figure 9. For the chosen parameters (reflux ratio, number of trays), before analyzing T-1, a condenser pressure close to the feed pressure is assumed. The sensitivity analysis showed that the C₃+ mole fraction in the column bottoms of T-1 increase considerably along with the increase in the total number of trays and reflux ratios. When the number of trays and reflux ratio are greater than 20 and 3, respectively, the purity and recovery of C₃+ at the bottom of the tower change very little, and the optimal number of trays and reflux ratios could be chosen as 20 and 3, respectively, where the mole percent of C₃+ in column bottoms is 99.62%.

In the case of the T-2 column, as in shown in Figure 10, the trend in C₂H₆ mole fraction in the column bottom increases as both the total number of trays and reflux ratios increases. However, there is a sharp increase in C₂H₆ mole fraction when the total number of trays exceeds 20. Therefore, the reflux ratio was 6 while total number of trays was 20, and the mole percent of C₂H₆ in the gas distillate is 98.28%.

In the T-3 column, however, as in shown in Figure 11, the mole fraction of LPG in column distillates and the mole fraction of C₅+ in the column bottoms are monotonously increasing along with the increase of total number of trays and reflux ratios. Moreover, when the total number of trays and reflux ratios exceeds 20 and 1.5, respectively, the mole fractions of LPG and C₅+ steadily increase. The total number of trays was then chosen to be 20, while the reflux ratio was 1.5. The detailed column operation parameters can be found in

Table 3. Process parameters for the distillation columns.

Streams	39	40	FG-2	C2H6	LPG	C5+
Key component	C2H6	C3+	CH4	C2H6	LPG	C5+
Mole fraction of key component	0.8273	0.9962	0.923	0.9828	0.991	0.967
Pressure (kPa)	2300	2350	2100	2150	1400	1450
Temperature (°C)	−9.1	90.67	−84.57	−7.6	56.68	142.3

.

4. Economic Assessment

4.1. Impact of Total Investment and OPEX on Process Design

Across different regions, the membrane cost differs considerably. We therefore took a higher value with the aim of including all the conditions. The literature and industrial data, together with utility and equipment economic parameters, are shown in

Table 4. Cost parameters for the economic assessment.

Item	Unit	Price	Reference
Membrane	$/m2	420/1000 1	[5]
Compressor	$/kW0.95	1620	[5]
Refrigeration	$/kW	1000	[39]
PSA	$	4.33 × 106	Equation (A7)
Heat exchangers	$/m2	1500 × A0.6	[5]
Pump	$/(m3h−1)	1.67 × 103	[5]
Electricity	$/kWh	0.13	[5]
Steam	$/t	14.50	[40]
Refrigeration cost	$/GJ	56.17	[22]
Cold water	$/t	0.10	[5]
Depreciation time	Year	5 for membrane, 5 for PSA 2, 15 for others

¹ USD 420 is the GPM’s unit price, whereas USD 1000 is the RPM’s price. ² The depreciation life of PSA in this study is set at 5 years in a manner corresponding to the replacement period of the adsorbent material. Pumps and compressors are designed with one for working unit and one for backup.

. The annual operating time is 8000 h.

According to the data shown in

Table 5. The capital cost and depreciation cost of the hybrid process.

Item	Capital Cost ($106)	Depreciation Cost ($106/Year)
Compressors	1.91	0.13
Refrigeration	1.80	0.12
PSA	4.33	0.86
Membranes	2.07	0.41
Distillation	1.008	0.07
Adsorption	0.40	0.03
Heat exchangers	0.002	1.4 × 10−4
Pumps	0.029	1.97 × 10−3
Other equipment cost 1	1.73	0.12

¹ Other equipment cost include mixers, pipes and valves, control systems, and flash tanks.

, in the process design, the cost of the PSA equipment accounts for 38.6% of the total investment, and is more than twice the total investment of the membrane module. PSA works better at high pressure, thus it is used to ensure the required hydrogen product quality. The compressor equipment and heat exchanger equipment cost, along with the distillation columns and refrigeration, contributed substantially (42.2%) to the total investment of the hybrid process.

From

Table 6. The operation cost of the hybrid process.

Item	Cost ($106/Year)
Electricity	1.66
Steam	0.55
Refrigeration	2.92
Cold water	0.30

, the total steam (12.87 × 10⁶ kJ·h⁻¹) in the graphic synthesis process represents 10.1% (USD 545.18 thousand/year) of the OPEX, and that of the cold water operation cost represents 5.52% while that of the electricity (4.36 × 10³ kW) happens to be 30.6%. The extra compressors used for the residual gas of the PSA device and the permeate stream of the RPM are the cause of the significantly increased electricity operation costs.

Figure 12 represents total investment breakdown of the hybrid process. The PSA and membrane investment cost calculation can be found in Appendix B and Appendix C, respectively.

The gross profit (GP) from the process design is roughly evaluated by the value of hydrogen, LHCs, and FG (fuel gas) produced, minus the major operating expenses (depreciation of membranes and other equipment, and the added cost of steam, refrigeration, cold water, and electricity), together with the feed streams values as described by Equation (2):

GP = total product yield + value of FG − (total annual cost + feed stream value)

(2)

The annual yield, calculated by multiplying the annual quantity (5.13 × 10³ t in the case of H₂) produced by the unit price (3.02 × 10³ $/t), is approximately 15.47 × 10⁶ $/year. Subsequently, the gross profit, calculated by the yield less the main current expenditures, along with the value of the feed streams, is approximately USD 38.62 × 10⁶ with a payback period of less than 4 months. The payback period expressed as Equation (3):

$$\mathrm{Payback}\mathrm{period}=\frac{\mathrm{TI}}{\mathrm{GP}}$$

(3)

where TI is total investment and GP is the annual gross profit.

It is evident that the hybrid process validated by Ruan et al. [30] is a viable and profitable technique to produce hydrogen and LHCs. The annual product yield of H₂ was 16.7% (USD 15.47 × 10⁶), while that of the LHCs was 82.9% (USD70.8 × 10⁶), with LPG representing the highest at 55.5%, as can be illustrated in

Table 7. Feed and product yield.

Component	Flowrate (103t/Year)	Unit Price ($103/t) a	Amount ($106/Year)
H2	5.1296	3.015	15.47
C2H6	22.224	0.72	16.00
LPG	49.368	0.98	48.38
C5+	11.416	1.08	12.28
FG	7.248	0.50	3.62
Feed gases (F1 + F2)	99.944	0.50	49.97

^a Unit price is from industrial data.

.

4.2. Break-Even Analysis

We ran a sensitivity analysis for the entire proposed process to see how the significant process factors affected the H₂ break-even pricing. Figure 13 shows the final results. Because of the strong dependence of H₂ production on membrane area, the GPM cost has the greatest impact on the hydrogen break-even price. The membrane unit cost range [5,8,39,41] (300 to 1000 $/m²) was utilized to reflect the variance in H₂ break-even price, and increasing the membrane unit cost to 1000 $/m² improves the H₂ price. As a result, area and cost are important considerations, and upgrading membranes will improve the economics. The hydrogen unit production cost compared with other studies [8,35,41] was relatively lower, demonstrating economic superiority. When compared to the cost of GPM, the cost of compressors and heaters had a weak impact on price. This emphasizes the importance of GPM cost in the overall proposed process’s economics. In the case of H₂ production from refinery gas using the graphic synthesis method, the membrane and its efficient operation, rather than other unit operations, determine the hydrogen break-even price.

5. Conclusions

Graphic synthesis design strategy has been presented for H₂ and LHC production from refinery gases using membranes, PSA, and shallow condensation. The design strategy uses shortcut methods by merging similar components, then representing operations into a visible triangular coordinate system. In this work, 10 streams containing valuable components, organized according to their concentrations, are merged into 2 mixed streams. The first mixed stream (F1) is sent to a PSA unit for high purity H₂ (> 0.99). The second (F2) mixed stream makes use of condensation, and with the aid of a membrane unit, H₂ is recovered from the residual gas of the PSA and redirected back to the PSA. Finally, the products of the second stream containing high amounts of LHCs, together with those of a second shallow condensation system, are sent into distillation columns where LHCs are produced. The design sensitivity was probed by changing the operating conditions and the membrane properties. The research findings of the factors influencing the membrane process showed that the optimized areas of the GPM and RPM membranes were determined to be 3500 m² and 600 m², respectively. The permeate pressures of the GPM and RPM membranes were set to be 250 and 200 kPa, respectively. Further economic analysis was done according to the optimized membrane properties, and the scheduled products include 0.641 × 10³ kg·h⁻¹ of H₂, 2.778 × 10³ kg·h⁻¹ of C₂H₆, 6.171 × 10³ kg·h⁻¹ of LPG, and 1.427 × 10³ kg·h⁻¹ of C₅+, showing product yield of USD 15.47 million, 16.00 million, 48.38 million, and 12.28 million per year for H₂, C₂H₆, LPG, and C₅+, respectively. The product yield brought in an annual gross profit of USD 38.62 × 10⁶ with a payback period of less than 4 months, thus indicating the potential of this graphic synthesis method and hybrid process.