RHO: A Software Tool for Targeting and Design of Re finery Hydrogen Networks

Wang Wence; Liao Zuwei; Cui Jiutao; Sun Jingyuan; Jiang Binbo;Wang Jingdai; Yang Yongrong; Feng Baolin

(1. Zhejiang Proνincial Key Laboratory of Adνanced Chemical Engineering Manufacture Technology,College of Chemical and Biological Engineering, Zhejiang Uniνersity, Hangzhou 310027;2. State Key Laboratory of Chemical Engineering, College of Chemical and Biological Engineering, Zhejiang Uniνersity, Hangzhou, 310027;3. CNPC Northeast Re fining & Chemical Engineering Co., Ltd, Dalian 116085)

Abstract: Hydrogen management is important for re fineries to improve their business efficiency. Various approaches such as pinch analysis and mathematical programming have been employed in the management of hydrogen system. However, it is not easy for site engineers to implement these techniques, due to the complicated procedures. At this point, it is necessary to develop a software that can implement the proposed methodologies automatically, which is really the goal of this work.The presented re finery hydrogen system optimization software (RHO) is a web based system. It is developed in the Java Web environment, where the subroutines of mathematical model developed in GAMS software can be easily called. RHO can generate graphics of both the hydrogen pinch diagram and the hydrogen distribution network. The puri fiers as well as the physical distances between units are considered in the optimization model. In addition, there is a special module for the calculation of membrane separation, which is very important in the hydrogen network. The functions and the interfaces of the software are illustrated via practical cases. Case studies show the effectiveness of the RHO software.

Key words: re finery hydrogen management; software; mathematical programming; pinch analysis

1 Introduction

Nowadays, the oil re fining industry is tending to produce cleaner fuels because of restriction on gas emission, which makes the consumption of hydrogen increase rapidly.Both the industry and the academic society around the world have been focusing on decreasing the cost of hydrogen consumption inside refineries. Actually, there are mainly two ways in re finery hydrogen management[1],viz.: the optimization of hydrogen allocation network and the placement of purifiers. They can be applied to both grassroots design and retro fit problems. After decades of development, hydrogen network optimization includes two major methods[2]: the pinch analysis[3-4]and the mathematical programming[5-7].

Mathematical programming is an important method for process synthesis. It is widely used in many areas, such as heat exchanger network, water network, and reactor network. For re finery hydrogen networks, the key factor is to build a model that includes all possible connections between unit operations and finds the optimal structure by algorithms[8]. It usually takes the following four steps to build mathematical model[9], viz.: ① Find the decision variables according to factors that affect object variable. ② Acquire the object function according to the relationship between decision variable and object variable. ③ Develop a network superstructure that includes all possible connections between sources and sinks. ④ Establish a mathematical model for the network superstructure.

Hydrogen pinch analysis is another approach to integrate the hydrogen network. It is based on the concept of heat pinch[10]for heat exchanger networks. In the late 1990s,Alves proposed the concept of hydrogen pinch[11-12]and the corresponding optimization method. At this point,pinch analysis is no longer limited to the optimization of heat exchange networks. After numerous researches and developments, a variety of methods for solving pinch points have been developed. Some of the methods have been widely applied to calculate the hydrogen pinch.The biggest advantage of the pinch analysis method demonstrates that it can determine the hydrogen utility target before a detailed network structure is obtained. The graphical interpretation is clear, easy to understand and easy to solve, and also simple in parameter adjustment.As a result, it is widely applied. However, due to the limitation of the graphical dimension, it cannot include the practical constraints and practical costs.

In recent years, the hydrogen network optimization tools have also been developed, including the H2-int from PIL[13], the Hydrogen network design tools[14], the Hydrogen Pinch and Petro-Sim from KBC[15], and the Hydrogen Management Services from Honeywell[16].

Among these software, H2-int employed the advanced mathematical modeling method and optimization algorithm to do simulation and optimization of the refinery hydrogen network system. Both the pinch analysis and the network design procedures can be carried out in this software.

Hydrogen network design tools[14]can help re finers to find the optimum solution using its advanced methodologies,which include the knowledge of the available option for hydrogen production, supply and recovery, and a suite of tools based on the advanced LP modeling and expertise on equipment cost estimation.

For complex re finery hydrogen systems, KBC developed two tools, Soft Hydrogen Pinch and Petro-SIM[15], for the management of refinery hydrogen. Through the combination of the two software, the refinery hydrogen network is optimized and integrated to reduce the cost of equipment and depreciation, which can greatly improve the efficiency of the refinery. The first step is to use the Hydrogen Pinch software to simulate a hydrogen network and identify the possibility of hydrogen network optimization. Then through the sensitivity analysis, the economic cost of parameter shift for each supplier and consumer is determined.

Hydrogen management services software[16]optimizes the hydrogen resources throughout the refinery’s hydrogen network and uses the hydrogen pinch technology to analyze hydrogen balance in the hydrogen network. The improvement of the hydrogen network is defined by the refinery hydrogen network model and the hydrogen generation and purification process model by setting a minimum hydrogen demand. This software can optimize the hydrogen utilization depending on the detailed hydrotreating model and the LP model of the entire re finery range.

In general, the optimization analysis of the refinery requires constant adjustment of the hydrogen utility supply. The user has to draw a pinch diagram and residual hydrogen curve which multiple times to find a pinch point with the same amount of hydrogen supply and demand[17].The above mentioned software could solve this problem.However, there is a need for the analysis and optimization of the purifiers. In this article, based on the hydrogen network optimization model in the GAMS optimization software, we provide a software for optimization of hydrogen network in re finery where the pinch point can be directly calculated without multiple adjustments.The detailed network structure including puri fiers could also be obtained. Besides, this software also contains an additional function of calculating the parameters for purifiers. The remaining part of this article is organized as follows: In section 2 we introduce the structure of this software including the interface for hydrogen data input and membrane calculation data input, then we show the output of pinch diagram, the flowsheet of hydrogen network, and the membrane calculation. In section 3 we detail the process of drawing pinch diagram. Furthermore, we also elucidate the formula and rationale about membrane calculation, which can simulate and figure out the parameters of membrane.In section 4 the process of drawing the flow rate chart is described. In section 5 we select several cases to show the effectiveness of the developed software, one of which is for the pinch analysis and another is for the membrane calculation. Finally, we summarize the performance of the developed software in section 6.

2 Structure of Software

We develop the interface of the software via the Java Web language, which is an open source and can be run on many systems and platforms without modification.Moreover, the Java environment can easily call the programs in GAMS software, where we develop the mathematical models. Since the software is developed in Java Web, users can visit it directly through the browser. The structure of the software is shown in Figure 1: After logging in, the user can choose the hydrogen data input interface to input data for a new project or read historical record which can be corrected and recalculated. Alternatively, the user can directly go to the membrane calculation module. For the hydrogen network optimization, all data will be automatically inputted into the program written with GAMS software for optimization. Then the optimization model in GAMS will be invoked to figure out an optimal scenario. The result of optimization for project user input will be passed to and stored in the database so that the user can read it in the historical record. Then the software will invoke data in the database to draw the pinch diagram and network structure. In membrane calculation after inputting data required for membrane calculation, the GAMS model will figure out the permeation coefficient of membrane and makeup of outlet gas to show user.

Figure 1 Software flow chart

The main input interface of RHO mainly consists of four parts, including the input of hydrogen supply, the hydrogen demand, the distance between supply and demand, as well as the constant parameters. In addition to the input of the flowrate and purity parameters, we have to input the name of the hydrogen supply and its distance to the purifier and the fuel system, respectively. New items of hydrogen supply can be added by pressing the blue circle bottom or deleted by pressing the red circle bottom. Similarly, for the hydrogen demand, we also need to input the name of the hydrogen supply and its distance to the puri fier and the fuel system, respectively.New hydrogen demand can also be added by pressing the blue circle bottom or deleted by the red circle bottom. For input of constant parameters, all the constants describing conditions of PSA have been preset and they could also be changed by the user.

User can press the submit button, after all the data have been inputted. Then the background program calls the optimization software GAMS according to the received front-end data and calculates the hydrogen supply amount of the hydrogen utility project under the current conditions in order to achive a lowest overall cost.

Calculation results are stored in the back-end database,and the pinch diagram is automatically drawn using the result data. The pinch diagram is shown in Figure 2. The red line represents the hydrogen source, the blue line represents the hydrogen sink, and the first segment of the red line is the hydrogen supply flow rate of the hydrogen utility and the hydrogen pinch concentration is also indicated in this diagram.

Figure 2 Pinch diagram

Upon clicking on the flowchart button to draw a hydrogen network flow chart based on the calculation results stored in the database, the flow chart will be shown later.

Click the membrane calculation button to launch a membrane calculation. The membrane section, the membrane area, the input pressure, the permeate side pressure, as well as the flow rate comprise the input parameters of membrane. The name of the input stream and its components are specified by the user. More components can be added by clicking the plus button.

In general, the software structure can be divided into three parts, viz.: the database, the data input interface,and the GAMS optimization software. After receiving the data, the background program calls GAMS to run the optimization programme prior to storing the calculation result and the data user inputs in the database, and then calls the data in the database to draw the pinch diagram and the hydrogen network flow rate.

3 Pinch Diagram Generation and Membrane Separation Calculation

3.1 Pinch diagram

The input data used for pinch diagram includes the hydrogen flow rate of hydrogen source and demand, the hydrogen purity of each hydrogen source and hydrogen demand, and the hydrogen flow rate derived from calculation of GAMS model. The drawing of pinch diagram includes several steps as shown below:

1) GAMS calculation: Software program receives data from the user and passes data to the GAMS model.GAMS will figure out the value of the flow rate of hydrogen utility under optimal condition[18]. In this step the optimal result depends on an objective function:

where the two constants of 83333.342 and 14748 is due to the unit conversion,Fhsrepresents the flow rate of hydrogen utility, Fregmeans flow rate of PSA product,Fc10means the cost of thermal energy per unit,Fc8represents the standard heat of combustion for hydorgen,Sw(i)means the off-gas discharged by sourcei,andd(i)means the source location relative to the purity levelk.

2) Pinch concentration calculation.

3) Ranking of hydrogen sources and demands: Rank the hydrogen source and hydrogen demand respectively in descending order of hydrogen purity.

4) Drawing of pinch diagram: After ranking the hydrogen source and hydrogen demand, this software will draw pinch diagram with the hydrogen purity as the ordinate and the hydrogen flow rate as the abscissa. Lay the hydrogen utility as a part of hydrogen source ahead of rest of hydrogen source, with the hydrogen source being marked with red line. Draw the hydrogen demand with blue line in the same way as the hydrogen source.

3.2 Membrane separation calculation model

The calculation principle is based on the cross-flow membrane module model[19], as shown in the following figure (the upper side is the leakage residual side and the lower side is the permeate side):

Figure 3 Model of membrane calculation

The leakage residual side membrane module is equally divided intoNstages according to the area with a decreasing permeate gas concentration. The mass balance equations for each component in each stage can be written as:

whereFR,nis the total flow rate in residual side of thenth stage, andxR,n,mstands forthe molar percentage of componentmin the residual side of thenth stage.Fp,nrepresents the total flow rate in permeation side of thenth stage, andxp,n,mmeans the mole percentage of component m in the permeation side of thenth stage.Lmis the permeation constant of componentm, whileAnmeans the area of thenth stage.

The above membrane model can be used under the following two conditions:

1) Based on the given flow rate, the pressure values of inlet gas and outlet gas, and the known membrane area,we can calculated the permeation coefficient of each component in the membrane.

2) Based on the given flow rate of the inlet gas, the given pressure at inlet and outlet, the membrane area, and the permeation coefficient, we can calculated the makeup of outlet gas.

By means of this membrane calculation model we can predict the performance parameters such as the permeation coefficient for the membrane. We can also forecast the composition of the outlet streams.

4 Hydrogen Network Structure Generation

The optimal result will be stored in database after GAMS software calculates the optimal scenario. Upon drawing a hydrogen network structure diagram, it is at first necessary to derive the stored results in a certain order of the hydrogen source and the hydrogen sink. In a Java program, we use a generic class to receive all the data exported from the database, which forms a data set that aggregates all the data we need for the structure diagram,and passes it to the next step where we begin to draw the hydrogen network structure diagram. Then in a reverse order the layers are unrolled so that each hydrogen source and hydrogen sink are presented in sequence. The main steps in drawing a flowchart include:

Obtain all hydrogen source information and all hydrogen sink information from the expanded data set, and then lay each hydrogen source on the left side and hydrogen sink on the top from left to right. Acquire and label the data of hydrogen flow and purity of each hydrogen source and hydrogen sink.

Obtain specific hydrogen dispatching data from the data set. In the proposed model, there is a set of binary variables that denote the existence of the connection between each hydrogen source and hydrogen demand.The value of these variables will be passed to the drawing program to complete the network structure. According to these values, the program draws connections between the sources and demands. In addition, the flowrate data between the existing connections will be extracted and labeled on beside the connection.Obtain the data of connections between the hydrogen sources and the fuel system and connections between the hydrogen utility and the hydrogen sinks. Draw the connections and label the flowrate parameters as the previous two steps.

5 Case Study

Case 1

Case 1 is from a certain refinery in northern China. It contains four hydrogen sources and six hydrogen sinks,the process data of which are shown in Table 1.

Table 1 Hydrogen supply and demand data of case 1

After clicking the submit calculation button, the calculation results are stored in the database. The pinch diagram and network structure are illustrated in Figure 4.In Figure 4(a) the calculated pinch concentration is 99.0 mol%. The red line represents the hydrogen supply and the blue line means the hydrogen demand. The first segment of red line is derived from the hydrogen provided by the hydrogen utility which is equal to 1 663.52 m3/s.For ease of representation, we use simplified symbols a1—a4 and b1—b6 to represent the actual units. The correspondence of these symbols are shown in Table 1.In Figure 4 (b) the hydrogen utility is connected to three hydrogen demand units and only one hydrogen supply has connection with the fuel system.

In the original project the flow rate of the hydrogen utility is equal to 2473.61 m3/s. The HCU of hydrogen supply has three connections to HCU, NHT, and DHT for hydrogen demand, which is equal to 444.4 m3/s, 13.88 m3/s, and 5600 m3/s, respectively. The DHT related hydrogen supply has only one connection to SRU for hydrogen demand, which is equal to 778.7 m3/s. Upon comparing the optimized plan with the original project plan of case 1, it can be found that the optimized project meets the requirements of each device, and the overall cost is cut by 32.7% compared with the original scenario.

Case 2

This case contains 9 kinds of gases―H2, H2S, CH4,C2H6, C3H8, butane, iso-butane, pentane, and CO2, which account for 87.82%, 0.034%, 5.94%, 3.53%, 1.83%,0.37%, 0.37%, 0.04%, and 0.04%, respectively, in the original makeup, and account for 15.97%, 0.13%,39.4%, 25%, 13.28%, 2.73%, 2.73%, 0.31%, and 0%, respectively, in the residual side. The constants for membrane attributes include a membrane area of 1 600 m2, and a number of membranes equating to 10.

It can be seen from Table 2 that the inlet stream is divided into two streams by the membrane. One is the permeate stream, the other one is the residual. The membrane model figures out the value of permeation coefficient according to the change in gas composition before and after they are in contact with the membrane. In this case the calculated permeation coefficient of these 8 kinds of gases is equal to 0.104 18, 0.005 18, 0.001 7, 0.001 05, 0.000 81, 0.000 56,0.000 56, 0, and 8.015 46, respectively.

Figure 4 Result output of case 1

Table 2 Output result of case 2

6 Conclusions

The developed RHO software can receive the basic data of the hydrogen network from the user and optimize the hydrogen network through the optimization model developed in the GAMS optimization software, and generate a more cost-effective solution. The hydrogen network structure diagram and the pinch analysis diagram can be automatically drawn according to the optimization scheme.

In the membrane separation calculation module, an idea of dividing the membrane into N equal parts for iterative calculation is provided. It can be used to calculate the permeation coefficient of each component in the membrane module according to the inlet and outlet pressure, the composition, the flow rate, and the membrane area. It can also calculate the outlet composition and flow rate according to the flow rate at the inlet, the pressure on both sides of the membrane module,and the membrane area, as well as the membrane module permeability coefficient.

Acknowledgements:The financial support provided by the Project of National Natural Science Foundation of China(21822809 & 21978256), the National Science Fund for Distinguished Young (21525627), and the Fundamental Research Funds for the Central Universities(2019XZZX004-03),and Ningxia Collaborative Innovation Center for Value Upgrading of Coal-based Synthetic Resin (2017DC57) are gratefully acknowledged.