Minbo Yang ,Xiao Feng, *,Liang Zhao

1 Shaanxi Key Laboratory of Energy Chemical Process Intensification,School of Chemical Engineering and Technology,Xi’an Jiaotong University,Xi"’an 710049,China

2 Key Laboratory of Advanced Control and Optimization for Chemical Processes,East China University of Science and Technology,Ministry of Education,Shanghai 200237,China

Keywords:Hydrogen Light hydrocarbon recovery Pinch technology Simulation Systems engineering

ABSTRACT In refineries,some hydrogen-rich streams contain considerable light hydrocarbons that are important raw materials for the chemical industry.Integrating hydrogen networks with light hydrocarbon recovery can enhance the reuse of both hydrogen and light hydrocarbons.This work proposes an automated method for targeting hydrogen networks with light hydrocarbon recovery.A pinch-based algebraic method is improved to determine the minimum fresh hydrogen consumption and hydrogen sources fed into the light hydrocarbon recovery unit automatically.Rigorous process simulation is conducted to determine the mass and energy balances of the light hydrocarbon recovery process.The targeting procedures are developed through combination of the improved pinch method and rigorous process simulation.This hybrid method is realized by coupling the Matlab and Aspen HYSYS platforms.A refinery hydrogen network is analyzed to illustrate application of the proposed method.The integration of hydrogen network with light hydrocarbon recovery further reduces fresh hydrogen requirement by 463.0 m3·h-1 and recovers liquefied petroleum gas and gasoline of 1711.5 kg·h-1 and 643 kg·h-1,respectively.A payback period of 9.2 months indicates that investment in light hydrocarbon recovery is economically attractive.

1.Introduction

Hydrogen is an essential utility in refineries to produce highquality fuel oil and other petroleum products.In recent years,the stricter environmental regulations,growing demands for fuel oil,and increasing processing of heavy and sour crude oil have led to sharp increases in hydrogen consumption in refineries [1].The refinery hydrogen demand is expected to rise continuously as the International Maritime Organization Marpol Annex VI regulation took effect on January 1,2020,which reduces the sulfur limit of bunker fuel from 3.5%to 0.5%[2].The increasing demand of hydrogen poses great challenges for most refineries.In refineries,there are many hydrogen-rich streams,such as off-gases from hydrocrackers and hydrotreaters,where hydrogen is often recovered and reused to mitigate the incremental hydrogen demand.Hydrogen network integration is an effective way to fully reuse the existing hydrogen and reduce the fresh hydrogen consumption,so it has been of great interest in both academia and industry [3].In addition to hydrogen,some streams also contains considerable light hydrocarbons[4],which are valuable fuels and raw materials for the chemical industry.Therefore,integration of hydrogen network with light hydrocarbon recovery can not only enhance the hydrogen recovery and reuse but also make better use of light hydrocarbons,offering more economic benefits for refineries.

Hydrogen pinch analysis is one of the most studied technologies for hydrogen network integration.The pinch-based methods are usually developed on the basis of either graphical representations or algebraic calculation.Alves and Towler [5] first proposed a graphical method that takes advantage of the purity profiles and hydrogen surplus diagram to determine the maximum direct reuse of hydrogen sources,setting a target for the minimum fresh hydrogen consumption.This method requires iteration to calculate the hydrogen surplus.El-Halwagiet al.[6] presented a noniterative graphical technique in the impurity load versus flow rate diagram for identifying the pinch and the minimum fresh resource usage.A graphical method based on the hydrogen load versus flow rate diagram was also developed[7].Alwiet al.[8]introduced a graphical tool known as network allocation diagram for targeting and design of gas networks.Additionally,graphical methods were developed for hydrogen network integration with purification reuse.Zhanget al.[9] represented the mass balance of a hydrogen purification process as a triangle or polygon and combined it with the hydrogen load versus flow rate diagram[7],proposing a graphical method for targeting the minimum fresh hydrogen consumption,pinch locations,and hydrogen sources for purification.Subsequently,this graphical method was improved for studies on hydrogen networks with purification reuse,e.g.,consideration of purifiers’performance and types[10,11]and targeting the optimal purification process [12].These graphical methods are characterized with clear conceptions and vivid procedures,but they are subject to visual resolution that may lead to inaccurate results.

The algebraic methods are particularly useful in the cases of numerous sources and sinks,scaling problems,and broader design task.Almutlaqet al.[13]transformed observations from the graphical targeting approach into algebraic insights and introduced an algebraic procedure for targeting of material-recycle networks.Agrawal and Shenoy[14]developed the composite table algorithm to target the minimum fresh resource consumption and presented an algorithm to design networks based on the principle of nearest neighbors.The composite table algorithm was later extended to locate flowrate targets for interplant hydrogen networks with direct reuse/recycle and purification reuse [15].According to the hydrogen surplus and pinch conceptions,Liuet al.[16] developed an algebraic approach for targeting of hydrogen networks.This algebraic approach was used to determine the limiting and optimal purification feed flow rate [16,17] and quantitatively analyze effects of purification reuse on fresh hydrogen consumption [18].Borgeset al.[19]presented the hydrogen source diagram to calculate the minimum fresh hydrogen consumption.Based on the definition of relative flow rate,Yanget al.[20] developed a noniterative algebraic procedure to determine the minimum fresh hydrogen consumption and pinch location by transforming graphical procedures into algebraic calculation.These two publications also took hydrogen purification reuse into account.

Superstructure-based mathematical programming is another useful tool for hydrogen network integration.Hallale and Liu [21]developed a superstructure including the placement of compressors and purifiers and proposed the first mathematical programming approach for hydrogen network integration.In a later contribution,the systematic selection of appropriate purifiers from pressure swing adsorption,membranes,or hybrid systems for hydrogen recovery was also incorporated into the hydrogen network superstructure[22].Based on the pioneer works,many other mathematical programming approaches have been developed from various viewpoints,including state-space superstructure [23],multi-components[24],multi-period operation[25,26],uncertainties [27],minimum number of compressors [28],minimum compression costs [29],intermediate hydrogen header [30,31],and inter-plant hydrogen network integration [32].Compared with pinch-based methods,mathematical programming methods are able to handle more complex problems but show less insights in hydrogen network integration.

Although pinch-based methods and mathematical programming methods have been well developed for hydrogen network integration,none of the aforementioned publications considers detailed light hydrocarbon recovery.This is because these publications focus on reuse of hydrogen and lump all light hydrocarbons as methane[3].At present,there are very limited studies involving light hydrocarbon recovery when investigating hydrogen network integration.Wuet al.[33] constructed a superstructure that contains light hydrocarbon recovery process,but the detailed process and case study were not presented.Denget al.[34] introduced a pinch-based approach for hydrogen network retrofit with light hydrocarbon recovery on the basis of process simulation in Aspen HYSYS.The hydrogen sources to the light hydrocarbon recovery process are determined by the improved problem table manually and keep constant without optimization.

In order to show insights in the integration of hydrogen network with light hydrocarbon recovery,this work aims to develop a targeting method by combining hydrogen pinch analysis and process simulation.First,the relative flow rate-based algebraic method is improved and programmed in the Matlab platform to automatically identify the minimum fresh hydrogen consumption and the hydrogen sources for light hydrocarbon recovery.A gasoline and diesel-based light hydrocarbon recovery process is modelled in the Aspen HYSYS platform.Next,such two platforms are coupled for data interaction and the targeting procedures are introduced in detail.The proposed method has a high level of practical relevance for industrial applications.

2.Problem Statement

The problem studied in this work can be expressed as below.In a refinery,there are a number of hydrogen consuming processes.The outlet streams of these hydrogen consuming processes are regarded as a set of process hydrogen sources S R={i|i=1,2,...,I},while the inlet streams are treated as a set of hydrogen sinks SK={j|j=1,2,...,J}.For a hydrogen sourcei,it has a known flowrateFiand a hydrogen concentrationyi.For a hydrogen sinkj,it has a required flow rateFjand a lower bound of hydrogen concentrationyj.A fresh hydrogen source with purity ofyfreshis available for supplementary.The surplus process hydrogen sources that are rich in light hydrocarbons can be treated to recover light hydrocarbons and lift the hydrogen purity simultaneously.The resulting hydrogen-rich stream is treated as a new hydrogen source that can be used by any hydrogen sink.The purpose of this work is to determine the minimum fresh hydrogen consumption of hydrogen network with light hydrocarbon recovery.

3.Improved Algebraic Method

The relative flow rate-based algebraic method for hydrogen networks is summarized as below [20].First,hydrogen sources and sinks are ranked in the descending order of hydrogen purities,respectively.Second,the relative flow rates of all hydrogen sinks and sources(RFjandRFi)are calculated by Eqs.(1)and(2),respectively.The next step is to calculate the relative flowrate of a hydrogen source in an interval (RFi,j),which performs one interval by another and needs designer’s judgement.Based onRFi,j,the flowrate of a hydrogen source in an interval (Fi,j) is calculated by Eq.(3).In each interval,the flow rate difference of sources and sink(ΔFj)is calculated by Eq.(4),and the surplus fresh hydrogen source(ΔFfresh,j) is obtained by Eq.(5).Among all surplus fresh hydrogen sources,the minimum one represents the maximum fresh hydrogen saving.The minimum fresh hydrogen consumptionis given by Eq.(6),and the surplus gasis given by Eqs.(7)or (8).

As aforementioned,the step to determineRFi,jneeds designer’s judgement,which limits the algebraic method for automatic solving.In order to overcome this drawback,the step to determineRFi,jis improved as presented in Fig.1.To be clear,the improved step is illustrated based on a hydrogen network with hydrogen sources and sinks shown in Tables 1 and 2.The hydrogen network includes hydrocracking unit (HCU),diesel hydrogenation unit (DHU),kerosene hydrogenation unit (KHU),and gasoline hydrogenation unit(GHU).

Referring to the algebraic method[20],relative flow rates of all hydrogen sinks and sources are calculated by Eqs.(1) and (2),as listed in the second row and third column in Table 3.According to Fig.1,the detailed procedures for interval 1 are exampled in Table 4.Forj=1 andi=1,ΔRFj=1(=20812.8)＞ΔRFi=1(=0).The calculation follows the left side in Fig.1,giving thatRF1,1=ΔRFi=1=0.Thus,ΔRFj=1and ΔRFi=1are unchanged.Next,iincreases from 1 to 2.ΔRFj=1(=20812.8) ＞ΔRFi=2(=1079.1),soRF2,1=ΔRFi=2=1079.1,ΔRFj=1=20812.8-1079.1=19733.7,and ΔRFi=2is changed to 0.Wheni=4,ΔRFj=1is reduced to 493.7 and smaller than ΔRFi=4(=6226.7).The calculation follows the right side in Fig.1,soRF4,1=ΔRFj=1=493.7,ΔRFi=4=6226.7-493.7=5773.0,and ΔRFj=1is reduced to 0.Afterwards,jis changed from 1 to 2,and the calculation goes to interval 2.A relative flow rate table is lastly tabulated as Table 3.Since the total relative flow rate of all sources is larger than that of all sinks,a portion of SR10locates outside of the four intervals and its relative flow rate is equal to 939.5.

Based on Table 3,the flow rate table is tabulated as Table 5.The maximum fresh hydrogen saving is identified as 11116.6 m3·h-1,which means the fresh hydrogen consumption can be reduced from 32,000 to 20883.4 m3·h-1.All hydrogen sinks are above the hydrogen pinch and the hydrogen source SR10locates across the pinch.According to Eq.(8),the waste gas is calculated as 1403.4 m3·h-1of SR10,which locates below the pinch.

4.Light Hydrocarbon Recovery Process

Fig.2 shows the flowsheet of a light hydrocarbon recovery process using gasoline and diesel as absorbents.It is a commonly used process to recover C3+hydrocarbons in refineries.

The feed stream (S1) after compression is mixed with stream(S16) from a deethanizer (T-102) and cooled in a cooler (E-100).The mixed stream (S4) enters a two-phase separator (V-101) to remove the hydrocarbon condensate.The gas product (S5) enters a gasoline absorber (T-101) where lean gasoline (S7) absorbs the most C3and C4components and a portion of C5+components.The resulting gas stream(S8)then flows into a diesel absorber(T-104)to enhance the recovery of C5+components by lean diesel.Finally,C3+hydrocarbons are recovered and a new hydrogen source (S10)is generated with a higher purity than the feed stream.

Fig.1. Improved procedure to determine the relative flow rate of a hydrogen source in an interval.

The rich gasoline (S12) and hydrocarbon condensate (S6) are mixed and heated in a heat exchanger (E-102).The mixture then enters a deethanizer T-102,where the C2and lighter components are removed.The bottom product(S18)is pumped to a debutanizer(T-103)after being heated in a heat exchanger(E-103).The C3and C4hydrocarbons are recovered as liquefied petroleum gas (LPG).Meanwhile,debutanized gasoline (S21) is obtained from the bottom and passes through two heat exchangers (E-103 and E-102)and a cooler (E-104) to reach desired temperature.The obtained gasoline can be recycled as absorbent or piped to downstream units.In this study,recycle of gasoline is not considered in order to make the simulation model easy convergence.In refineries,the diesel used as absorbent is typically taken from other units,such as the diesel hydrotreating unit.The rich diesel is usually treated together with other diesel streams.Therefore,this part is not included in the light hydrocarbon recovery process.

5.Targeting Procedure

The targeting procedure for hydrogen networks with light hydrocarbon recovery is illustrated in Fig.3.It is an automatic approach that combines the improved algebraic method and rigorous process simulation to target the minimum fresh hydrogen consumption.This approach is realized by connecting Matlab and Aspen HYSYS through Component Object Model(COM)technology[35].

The targeting procedure starts in the Matlab platform.For a given hydrogen network,the initial fresh hydrogen saving (Fs0) is set as zero.The first step is to identify the hydrogen sources and sinks.Next,the fresh hydrogen saving (Fs) and surplus hydrogen sources are identified by the improved algebraic method in Section 3.It is noteworthy that the identified surplus sources always have lower hydrogen concentrations than the hydrogen sourcesdirectly reused.The increment of fresh hydrogen saving (ΔFs) is defined as the difference betweenFsandFs0.For a specified small value ε,if ΔFs＞ε,it means there exists a potential for fresh hydrogen saving,and thusFsis assigned toFs0for next iteration.The Matlab platform then transfers the data of surplus hydrogen sources to the Aspen HYSYS platform,where these hydrogen sources are considered as the feed to the light hydrocarbon recovery process.The rigorous process simulation is performed to determine the mass and energy balances.As aforementioned,a new hydrogen source is generated ashydrocarbons are removed.The data of the new hydrogen source is returned to the Matlab platform.As a result,a new hydrogen source set is formed including the new hydrogen source and the original hydrogen sources exclude the feed to the light hydrocarbon recovery process.Afterwards,a new iteration starts.The targeting procedure stops when ΔFs≤ε,which means that the increment of fresh hydrogen saving is quite small and can be ignored.

Table 1 Data of hydrogen sources

Table 2 Requirements of hydrogen sinks

Table 3 Relative flow rate table

Table 4 Example for calculating the relative flow rate of a hydrogen source in an interval (j=1)

Table 5 Flow rate table (m3·h-1)

6.Case Study

Fig.2. Flowsheet of the light hydrocarbon recovery process.

In this section,the proposed method is tested by the hydrogen network studied in Section 3.The compositions of gasoline and diesel considered as absorbents are given in Tables 6 and 7,respectively.Peng-Robinson is selected as the property package in the process simulation [36].For simplification,the mass-based ratios of gasoline to gas and diesel to gas are assumed to be constant and taken as 6 and 10.These two values are determined by simulation in Aspen HYSYS to recover nearly all C3+hydrocarbons.Besides,the value of ε is taken as 0.1 m3·h-1.

The proposed targeting procedure is performed in Matlab R2018b and Aspen HYSYS V10.Table 8 shows the changes of fresh hydrogen saving and its increment along with the feed to the light hydrocarbon recovery process.It can be seen that the targeting procedure stops after 8 iterations with ΔFs＜0.1 m3·h-1,reaching the specified stopping criterion.It is noteworthy that the second column gives the fresh hydrogen saving of 11116.6 m3·h-1when light hydrocarbon recovery is not considered.With light hydrocarbon recovery,the fresh hydrogen saving increases to 11579.6 m3·h-1by 463.0 m3·h-1.As a result,the fresh hydrogen consumption reduces to 20420.4 m3·h-1.A feasible design of the hydrogen network that meets this target is presented in Fig.4 with flowrates given in m3·h-1.

Table 9 gives the input and output of the light hydrocarbon recovery process.It can be seen that 99.9% of C3components and all C4+components are recovered from the feed.With a further insight,it is found that the LPG product (S20) takes over 98% of the C3and C4components in the feed (S1).Besides,98% of the C5components and 67% of the C6components exist in gasoline(S24).The light hydrocarbon recovery process has very limited influences on hydrogen,C1and C2components,and thus almost all of them remain in the new hydrogen source (S10).

To be a worthwhile investment,a venture for the installation of a new process must be profitable.Therefore,economic analysis is performed for the light hydrocarbon recovery process.Aspen Process Economic Analyzer V10 is employed to evaluate the total capital investment.The annual production cost is estimated as the sum of utilities,operating labor,maintenance,and supervision.The annual revenue is evaluated by summing the benefit of fresh hydrogen saving and the sales of LPG and incremental gasoline.Table 10 gives the parameters for economic analysis.The breakdowns of the total capital investment,annual production cost,and annual revenue are presented in Fig.5.It can be seen that steam consumption dominates the total annual production cost.The steam at 1.0 MPa and steam at 3.5 MPa are consumed by the deethanizer and debutanizer,respectively,meaning that the flow rate of gasoline highly affects the total annual production cost.In terms of the annual revenue,it mainly comes from LPG and incremental gasoline,rather than the fresh hydrogen saving.The payback period is calculated as about 9.2 months,indicating that the introduction of light hydrocarbon recovery into the hydrogen network is economically attractive.

Fig.3. Targeting procedure of the hybrid approach for hydrogen networks.

Table 6 Composition of gasoline

Table 7 Composition of diesel (ASTM D86)

Table 8 Results of each iteration

Fig.4. A feasible network design with light hydrocarbon recovery (LHR) (unit:m3·h-1).

Table 9 Input and output of the light hydrocarbon recovery process

Table 10 Parameters for economic analysis

7.Conclusions

Fig.5. Techno-economic analysis of the light hydrocarbon recovery process.

In this work,a hybrid method was developed for targeting hydrogen network with light hydrocarbon recovery.This method combined hydrogen pinch analysis with rigorous process simulation and was realized by coupling Matlab and Aspen HYSYS.A hydrogen network was studied to illustrate the effectiveness of the developed method.The results showed that the introduction of light hydrocarbon recovery could not only enhance the reuse of hydrogen but also promote the value of hydrocarbons with a relatively short payback period of 9.2 months.However,this work does not include the optimization of the light hydrocarbon recovery process and hydrogen network configuration,which will be considered in our future work.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

This work was supported by the Fundamental Research Funds for the Central Universities (2020ACOCP04).

Nomenclature

FiFlow rate of hydrogen sourcei,m3·h-1

Fi,jFlow rate of hydrogen sourceiin intervalj,m3·h-1

FjFlow rate of hydrogen sinkj,m3·h-1

FsFresh hydrogen saving,m3·h-1

IThe number of hydrogen sources

JThe number of hydrogen sinks

RFiRelative flow rate of hydrogen sourcei,m3·h-1

RFi,jRelative flow rate of hydrogen sourceiin intervalj,m3·h-1

RFjRelative flow rate of hydrogen sinkj,m3·h-1

SR Set of hydrogen sources

SK Set of hydrogen sinks

yfreshHydrogen concentration of fresh hydrogen source,% (mol)

yiHydrogen concentration of hydrogen sourcei,% (mol)

yjHydrogen concentration of hydrogen sinkj,% (mol)

ΔFfresh,jSurplus fresh hydrogen source in intervalj,m3·h-1

ΔFjFlow rate difference of sources and sink in intervalj,m3·h-1

ΔFsChange of fresh hydrogen saving,m3·h-1

ε Specified small value,m3·h-1