Bai Haobo; Li Shiyu; Wang Xupeng; Liu Chang

(School of Chemical Engineering and Technology, Tianjin University, Tianjin 300350)

Abstract: In an integrated re fining and petrochemical complex, a centralized utility system (CUS) is introduced to integrate the steam demands of production plants. Besides, two sub-utility systems (SUSs) located inside the alkene and re finery plants, respectively, can satisfy the shaft demands. It is difficult to determine the steam production of the CUS because the steam demands of the alkene and re finery plants also depend on the design and operation of the SUSs. To explore the complicated interaction between the CUS and SUSs, we proposed a mixed-integer nonlinear programming (MINLP) model for the design and optimization of multiple interconnected utility systems to minimize the total annualized cost (TAC). An extended superstructure was suggested to contain multiple inter-plant connected steam pipe alternatives between the CUS and SUSs. A more accurate model of the complex steam turbine was proposed. Then the proposed MINLP framework is applied to a new integrated refining and petrochemical complex. Two scenarios are investigated in the case study to explore the effect of steam main temperatures on system configurations and operating parameters. By optimizing the main temperatures, a TAC of $2.7 million can be saved. Judging from the results of the two scenarios, the feasibility and effectiveness of the proposed framework for the design and optimization of multiple interconnected utility systems have been demonstrated.

Key words: MINLP model; multiple interconnected utility systems; complex steam turbines; steam main temperatures

1 Introduction

Due to the surplus refining capacity and increasing demand for petrochemicals in China, petroleum re fineries and petrochemical companies have been integrating into complexes[1-2]. In the complex, a CUS is introduced to distribute steam among the production plants. Moreover,two SUSs located inside the alkene and refinery plants can satisfy the respective shaft demands. Because of their large capital and operating costs, the utility systems can offer the opportunity of accomplishing significant economic benefits by improving their design and operation. However, given the complicated interactions between the CUS and SUSs, minimizing such expenditure represents a challenging task.

The design and optimization of utility systems can be achieved through thermodynamic and mathematical programming approaches. The thermodynamic approach is based on heuristic rules to minimize thermodynamic losses of the system[3-4]. The results obtained from this approach are often not economically competitive in practice, because the trade-off between investment and energy is not considered. Papoulias and Grossmann[5]developed mixed-integer linear programming (MILP)framework to minimize the TAC of utility system.Subsequently, Iyer and Grossmann[6-7]extended this framework to be a multi-period MILP model.Nevertheless, most of the above-mentioned frameworks were only applied to a single utility system. Recently,Luo, et al.[8]suggested a MILP model for operational optimization of multiple interconnected utility systems,where a CUS provides steam for several SUSs by inter-plant connected pipes. However, the structural optimization of multiple interconnected utility systems has been rarely reported because of the extensive number of potential combinations of the pipeline alternatives between the CUS and SUSs. Therefore, to design the economically attractive utility systems in an integrated refining and petrochemical complex, it is necessary to formulate a mathematical programming model to optimize the CUS and SUSs simultaneously.

The complex turbines (i.e., multiple extraction turbines),generating shaft power while distributing steam to mains at different pressure levels or condensers simultaneously,are important equipment in utility systems. The complex turbine was modeled by applying a cascade of single ones[4]and the corresponding isentropic enthalpy difference models[9-10]. Many accurately nonlinear models[11-13]were developed to predict single turbine performance. However, the optimization procedure becomes more complicated while adopting these nonlinear models for design problems. To overcome this shortcoming, Aguilar, et al.[14]proposed a linear single turbine model and further developed a linear model of the complex turbine. In this model, the decomposed single turbines are assumed to attain a uniform isentropic efficiency to estimate the corresponding isentropic enthalpy difference. Nevertheless, the isentropic efficiency generally changes with turbine sizes and loads, which ranges from 0.42 to 0.85 at full load[15]. Medina-Flores and Picón-Núñez[16]found that although this isentropic efficiency changed in a narrow range, the maximum relative error (MRE) of the shaft power prediction may exceed 8.0%. Thus, it is necessary to propose a new method to estimate the isentropic enthalpy difference more accurately instead of assuming that directly.

In utility systems, the mains collect steam from turbines or letdown valves at the upper-pressure level and meanwhile distribute steam to them at the lower level.Thus, the main temperatures are affected by the turbine load and steam flow through letdown valves at the upper level and also affect the power production at the lower level[17]. Chen and Lin[18]reported that the main temperatures can also affect the system configurations for design problems. Thus, it is necessary to analyze the in fluence of main temperatures on system con figurations and operating conditions.

In this paper, an MINLP model for the design and optimization of multiple interconnected utility systems was proposed. The performance of the proposed model is demonstrated by being applied in a new integrated refining and petrochemical complex. The effect of the main temperatures on system con figurations and operating conditions is also explored to accomplish TAC saving.

2 Problem Statement

Typical multiple interconnected utility systems comprise a CUS located outside the production plants of the complex,and several SUSs located inside the production plants.Figures 1 and 2 indicate the superstructure of multiple interconnected utility systems extended from the single utility system as proposed by Papoulias and Grossmann[5].The suggested superstructure includes multiple interplant connected pipeline alternatives between the CUS and SUSs to transport steam as well as many possible equipment alternatives and their connections.

Figure 1 The superstructure of the CUS

The configuration alternatives of the CUS are shown in Figure 1. In the CUS, there are four steam mains at very high (VHP), high (HP), medium (MP), and low (LP)pressure levels, respectively. VHP steam can be produced by fired boilers (FB) consuming fuel mixture. In each main, steam can be utilized by turbines, transported to the production plants by inter-plant connected pipelines,or distributed to letdown valves (LTD) where the water can be added. It is necessary to mention that there are no steam turbine (ST) alternatives in the LP level, but a stream of steam is extracted to a deaerator (Dea). The deaerator is used to remove dissolved gases and supplies water to the boiler feedwater pumps, and meanwhile the demineralized water is added to compensate for steam losses. The fired boiler feedwater (FBFW) is further heated by MP steam and the condensate returns to the deaerator. Two types of driver alternatives (i.e., steam turbines and electric motors) are considered to drive utility pumps (e.g., boiler feedwater pumps and draft fans). Electricity demands can be satis fied by gas turbines(GT) and steam turbines, or can be imported from the local power grid. The exhausts from gas turbines can be further exploited by heat recovery steam generators(HRSG) to raise VHP steam.

In plants 1 and 3, due to the absence of steam turbine alternatives, the utility pumps are just driven by electric motors (EM). Thus, the steam exports or imports to/from the CUS directly through the inter-plant connected pipelines at the corresponding pressure level. For the case of performing this work, plant 1 consists of air separation,coking coal gasification, and waste heat recovery units,while plant 3 corresponds to the polyole fin unit. In plant 1, there are three pipelines operating at HP, MP, and LP levels, respectively, to export steam. In plant 3, there are three pipelines at the same levels to import steam.Additionally, a deaerator is included to supply the water for waste heat boilers in plant 1.

The configuration alternatives of the SUS are shown in Figure 2. In the SUS, shaft demands can be satisfied with either direct-drive steam turbines (DDSTs) or electric motors. Thus, there are four inter-plant connected pipeline alternatives to import steam because the steam at lower-pressure levels can be produced by DDSTs.Each main collects steam from the CUS, the process units, and DDSTs or letdown valves at the upper-level,and meanwhile distributes steam to process users, and DDSTs or letdown valves at the lower-level. A deaerator is installed to provide the water for waste heat boilers.For the case of performing this work, there are two SUSs located inside the alkene (plant 2) and re finery (plant 4)plants, respectively.

Set 1/2/3 comprises a series of turbines. The pressure levels entering turbines in a set are the same but the number of extractions and the corresponding pressure levels are different. As indicated in Figure 3, the inlet steam of ST at set 2 is the HP level, which includes seven types of turbines (t1—t7).

3 Mathematical Model Formulation

3.1 Objective functions

Figure 2 The superstructure of the SUS

Figure 3 Diagram illustrating all types of turbines in ST set 2

The objective of this study is to minimize the TAC of the utility systems, given as Equation (1). The purchase cost functions of steam turbines with/without condensers[12]and the rest of equipment[14]are adopted to calculate the capital cost (Cpc), as shown in Equation (2). Equation (3)represents the operating cost (Opc) comprising the cost of fuel, demineralized water, and electricity import, and the revenue of electricity export.

where Fann, Fcepi, Finst, and AOT are parameters for the annualization factor, chemical engineering plant cost index factor, installation factor, and annual operating time, respectively; PC, Cst, M, and We are variables for the equipment purchase cost, specific cost, mass flow,and electricity, respectively; n and f are indices for the equipment and type of fuels respectively while imp and exp are superscripts for import and export, respectively.

3.2 Equipment limits

In this work, Equations (4) and (5) limit the size range for the equipment alternatives. Equation (6) defines a minimum partial load for the equipment so that the actual load does not go below the allowed limits.

where Lod is a parameter for the equipment load; Outp is a variable for the equipment output; yselis a binary variable to specify the selections of equipment; and D,low, and up are superscripts for the design state, lower and upper limit, respectively.

3.3 Main equipment models

Compared with a single utility system, much larger amounts of structural and operational alternatives are evaluated in the design and optimization of multiple interconnected utility systems. Thus, the linear models of equipment proposed by Aguilar[14]are adopted, including multi-fuel boilers, gas turbines with HRSGs, single steam turbines, and electric motors. The power of boiler draft fans is estimated by the correlation proposed by Varbanov[12].

As mentioned above, the complex turbine model proposed by Aguilar[14]is not accurate enough, so it is necessary to modify the model to improve the accuracy. As shown in Figure 4, the complex turbine t is decomposed into z single turbines in series with fixed inlet and outlet pressures. Equation (7) shows that the shaft power Wtof complex turbine t is the summation of that of decomposed single ones. The isentropic enthalpy difference is correlated against the speci fic enthalpy in steam entering the single turbines, shown as Equation(8). The MRE of these regression models is no more than 0.30%.

where p and T are parameters for pressure (MPa) and temperature (°C) respectively; h and Δh are variables for the speci fic enthalpy (kJ/kg) and their difference; and is,in, out, and L are superscripts/subscripts for the isentropic expansion, input, output, and saturated water at the corresponding pressure, respectively.

To calculate the isentropic enthalpy difference of decomposed single turbines, the specific enthalpy in steam leaving the turbine at the upper-pressure level needs to be determined. The turbine machine efficiency is generally in the range from 0.97 to 0.99[19]and does not change substantially with actual loads. Thus, with a value of the turbine machine efficiency ηm, the speci fic enthalpy of the turbine exhaust can be calculated by Equation (9).

3.4 Mechanical driver selection

In this work, each shaft demand (index i) can be satis fied by two types of drivers (i.e., electric motors and steam turbines), as shown in Equation (10). The constraint (11)ensures that only one of the turbines (index j) or one electric motor can be on.

where W, Wtur, and Wem are variables for the shaft power, that of steam turbine and electric motor,respectively; yturopand yemopare binary variables to specify the operating status (on/off) of steam turbines and electric motors respectively; and dem is superscript for the demand.

3.5 Electrical balance

To precisely determine the cost or revenue due to electricity import or export, it is necessary to build an equality constraint between electricity providers and potential consumers, as presented in Equation (12).

where gen is superscript for the generation of electricity or steam.

3.6 Mass balance

Many distinct streams are mixed in numerous nodes of utility systems. For a specified node, the total mass flows entering must be equal to the summation of the flows leaving. Equations (13) and (14) show the mass equilibrium around mains in the CUS and SUSs,respectively. Equation (15) corresponds to the mass balance of the deaerator, in which several water flows are heated by LP steam.

where k, hr, and pn are indices for the steam mains,HRSGs, and production plants, respectively; and stm,wat, preh, vnt, cnd, ret, mkp, and bfw are superscripts/subscripts for the steam, water, FBFW preheat, vent,condensate, condensate from the process, demineralized water makeup, and boiler feedwater, respectively.

3.7 Heat balance

Many distinct streams under various conditions (e.g.,temperature and pressure) are adiabatically mixed. For a speci fied node, the total enthalpy entering must be equal to the one leaving. Equations (16) and (17) ensure that enough steam is supplied to SUSs and the production plants, respectively. Equation (18) ensures that enough LP steam is merged into the deaerator so that the water leaving is under the saturated liquid conditions.

Figure 4 Decomposition of the complex turbine

3.8 Selection of inter-plant connected steam pipelines

The nominal diameter of steam pipelines is generally more than 50 mm[20]in petrochemical plants, so we need to limit the lower steam flow through the interplant connected pipelines between the CUS and SUSs,as given in Equation (19). Since the superheated steam velocity is usually more than 20 m/s[21]and the maximum steam density at each pressure level is easily obtained,so that the lower steam flow through the pipelines can be determined. Equation (20) is introduced to activate the flowrates that are consistent with those of the final pipelines.

Thus, the optimization procedure of multiple interconnected utility systems is a MINLP model comprising the objective functions of Equations (1)—(3)and the constraints of Equations (4)—(20).

4 Case Study

In this section, a case study based on a new integrated refining and petrochemical complex is given to prove the capabilities of the proposed MINLP model. The site conditions are given in Table 1, and the base flowrate of fuel gas generated from plant 1 is 74.3 t/h. Besides,natural gas can be consumed in either GTs or fired boilers while fuel gas is only consumed in the latter. A set of steam and electricity demands is given in Table 2,and the corresponding steam conditions are presented in Table 4. Table 3 shows the shaft demands, and those of utility pumps are calculated by the optimization procedure.

This MINLP model is formulated in GAMS 23.8 and is implemented by the ANTIGONE solver[22]in NEOS Server. The annualization factor adopted is 0.1.As mentioned above, the main temperatures play a signi ficant role in the system con figurations and operating conditions, so two scenarios are considered in the case study.T

4.1 Scenario 1: Fixed steam main temperatures

In this scenario, steam temperatures in mains or from the process units are given in Table 4. To match the fixed main temperatures, water can be injected into letdown valves.

Table 2 Steam and electricity demands data for this case

Table 3 Shaft demands data for this case

Table 4 Steam conditions for this case

The final design and the corresponding operating parameters are shown in Figures 5—7 for the CUS,SUS in plant 2 (SUSP2), and SUS in plant 4 (SUSP4),respectively, the TAC of which is $701.17 million. In the CUS, a 59.4 MW GT with a supplementary-fired (SF)HRSG of 204.8 t/h and two fired boilers (with one rated at 450.0 t/h and the other rated at 377.8 t/h) supplying steam to the VHP main. It is necessary to mention that the number 450.0 t/h is set as the maximum output for fired boilers. A 35.8 MW complex turbine is selected to generate electricity. The remaining electricity demand of 601.0 MW is imported from the local power grid. Two backpressure turbines are installed to drive the boiler feedwater pumps (BFP) and boiler draft fans (BDF). The flowrates of steam discharged from VHP, HP, and MP mains through letdown valves are 1.7 t/h, 3.5 t/h, and 0.3 t/h, respectively. There are two inter-plant connected pipelines at different pressure levels to supply 31.7 t/h of VHP steam and 149.2 t/h of HP steam for SUSP2. There is one pipeline to supply 558.4 t/h of VHP steam for SUSP4.

In SUSP2, external five shaft demands and BFP are all satis fied with steam turbines, two of which are complex turbines. The letdown steam flowrate from HP and MP mains is 25.4 t/h and 27.3 t/h, respectively. In SUSP4,external eleven shaft demands are all satisfied with turbines except that the BFP is driven by the electric motor, and two of DDSTs are complex turbines. The letdown steam flowrate from VHP and MP mains is 20.3t/h and 1.8 t/h, respectively.

Judging from the results, we can found that a large amount of steam passes through letdown valves in the complex utility systems. The possible reason is that the letdown steam is used to match the main temperature requirements. Moreover, the letdown steam flows in SUSs are much larger than the CUS. The possible reason is that the search space is limited by DDSTs of fixed sizes in SUSs. Therefore, relaxing the limits of main temperatures may accomodate some new feasible structures or operating conditions to reduce the letdown steam flow rates.

4.2 Scenario 2: Optimized steam main temperatures

In scenario 2, the same problem is solved again with variable main temperatures except for the VHP, the range of which is shown in Equation (8). To simplify the optimization procedure, water injection in letdown valves is not considered, and the temperatures of steam from the process units are still fixed.

Figure 5 Flowsheets of the CUS for the fixed main temperatures

Figure 6 Flowsheets of SUSP2 for the fixed main temperatures

Figure 7 Flowsheets of SUSP4 for the fixed main temperatures

Figures 8—10 indicate the optimal configuration and corresponding operating parameters for the CUS, SUSP2,and SUSP4, respectively, and the corresponding TAC is $698.51 million. In the CUS, there are two VHP fired boilers of the same size as scenario 1, and a 60.9 MW GT with SF-HRSG of a slightly larger size so that the fuel cost increases by $1.9 million/a. One more backpressure turbine is installed besides a complex turbine so that the electricity production increases by 4.8 MW. The remaining electricity import reduces by 6.3 MW so that the corresponding cost decreases by $5.4 million/a. There are still two backpressure turbines for driving BFP and BDF separately. The letdown steam flows reduce to zero,because more appropriate main temperatures are found,which are 380.3 °C, 253.9 °C, and 185.4 °C for HP, MP,and LP mains, respectively. An MP inter-plant connected pipeline transports the steam to SUSP2 in addition to the original VHP and HP ones. There is an MP pipeline connected to SUSP4 besides the original VHP one.

In these two SUSs, the main temperatures are 375.8 °C,253.3 °C, and 180.6 °C for SUSP2, and 378.2 °C,258.4 °C, and 183.5 °C for SUSP4. Due to the changes in main temperatures, the final type of drivers is the same as scenario 1 but the DDST configurations are different. In the SUSP2, the shaft demand 1 is satisfied by a HP complex turbine, which replaces the original VHP complex turbine. The shaft demand 5 is satisfied by a VHP complex turbine with two extractions to HP and MP main separately, which replaces the original HP condensing turbine with an extraction. In the SUSP4, the shaft demand 8 is satis fied by a HP backpressure turbine,which replaces the original MP complex turbine. The shaft demand 10 is satis fied by a VHP condensing turbine with two extractions to MP and LP main separately,which replaces the original VHP backpressure turbine.The letdown steam flows from the mains decrease to zero in these two SUSs, besides 1.8 t/h of letdown steam discharging from the HP main in SUSP4.

Compared with scenario 1, we can found that the total letdown steam flowrates from VHP, HP, and MP mains reduce by 22.0 t/h, 27.1 t/h, and 29.3 t/h, respectively,in the complex utility systems. Since the most portion of the reduction is contributed by SUSP2 and SUSP4,the main temperatures play a more important role in the steam distribution for the utility system containing multiple DDSTs. Because of this improvement, the TAC is saved by $2.7 million. Therefore, it is evident that the lower TAC can be accomplished by optimizing the main temperatures. More detailed cost comparisons for these two scenarios are shown in Table 5.

Figure 8 Flowsheets of the CUS for the optimized main temperatures

Figure 9 Flowsheets of SUSP2 for the optimized main temperatures

Figure 10 Flowsheets of SUSP4 for the optimized main temperatures

Table 5 Major economic parameters for this case study

5 Conclusions

This study proposed an MINLP model for the design and optimization of multiple interconnected utility systems.To determine the optimal configuration, an extended superstructure was presented where multiple inter-plant connected pipeline alternatives are included between the CUS and SUSs. A more accurate isentropic enthalpy difference model was developed to improve the accuracy of the complex turbine, the MER of which is no more than 0.3%. Then the proposed MINLP framework was applied to a new integrated refining and petrochemical complex. Two scenarios were studied in the case study to explore the in fluence of main temperatures on system structure and operating conditions. For the fixed main temperatures, the TAC is $701.17 million in scenario 1. By optimizing the main temperatures, there is better performance on steam distribution in scenario 2 so that the TAC can be saved by $2.7 million. Judging from the results of the case study, the applicability of the proposed framework to specify the optimal con figuration for power generating equipment and to determine the optimal main temperatures has been proved. Therefore, adopting the proposed framework can obtain an economically attractive design and operation of multiple interconnected utility systems.

Three Green Chemical Technologies Developed by SINOPEC RIPP Help to Optimize and Upgrade the Industry

Recently three technologies, viz.: “The PDP of an 120 kt/a unit for manufacture of hydrogen peroxide from anthraquinone in slurry bed reactor”, “The package PDP of a 300 kt/a unit for manufacture of propylene oxide from propylene and H2O2(HPPO process)”, and “The alkanol ether recovery technology”, which were developed by the SINOPEC Research Institute of Petroleum Processing(RIPP), have passed the review and evaluation organized by the SINOPEC Science and Technology Division.These three technological achievements have obvious effects on helping the enterprises to reduce production cost and increase economic bene fits, optimize product mix and upgrade product quality, and promote environmental protection. Among these technologies the HPPO process has been on a par with the internationally advenced level.The technology for manufacture of H2O2from anthraquinone in slurry bed reactors is a package technology independently developed by SINOPEC, which in August 2019 had constructed the first in China commercial demonstration unit at its Baling Branch Company with independent intellectual property rights. This unit, which was put on stream with success at the first attempt, has been operating smoothly for one year. On this basis the researchers of RIPP further worked on process improvements and engineering scale-up to develop a PDP of an 120 kt/a unit for production of H2O2from anthraquinone in slurry bed to enhance the economics and reliability of the process that can meet the needs of the asssociated 300 kt/a caprolactam unit to unravel the bottlenecks of downstream units.

The HPPO process is the advanced technology supported by the state policy and urgently needed by the development plan, which can be conducive to the utilization of China’s propylene resource and adjustment of petrochemical industry structure to enhance the competitive edge of petrochemical industry. RIPP，while tackling gut issues for many years in collaboration with other institutions, had construted in 2014 the first in China 100 kt/a commercial demonstration unit for manufacture of propylene oxide from propylene and H2O2(HPPO process) to break the overseas monopoly and fill up the blank gap inside China, which has become the third country possessing the package HPPO technology. Currently this unit has been working smoothly with long operating cycle. Based on the experiences learned from running the 100 kt/a commercial demonstration unit, the research team of RIPP with in flexible devotion in grappling with gut issues related with technical optimization of process flow diagram, reaction compartment and control system has developed the first in China PDP of a 300 kt/a package unit for production of propylene oxide, which is expected to become SINOPEC’s Triton among the minnows in the domain of petrochemical industry to prop up the technical progress and development of the high-end new materials prduction.

Furthermore, RIPP in collaboration with the Tianjin University has developed the technology for recovering alkanol ethers, which not only can effectively dispose of the watewater generated in the HPPO process, but also can recover the high value-added propylene glycol ether and propylene glycol to reduce the production cost along with protecting the environment.