Transport ISSN 1648-4142 / eISSN 1648-3480 1 2017 Volume 00: 1–8 53 2 http://dx.doi.org/10.0000/00000000.0000.000001 54 3 55 4 56 5 OPTIMIZATION OF NATURAL GAS TRANSPORT PIPELINE NETWORK LAYOUT: 57 6 A NEW METHODOLOGY BASED ON DOMINANCE DEGREE MODEL 58 7 59 8 60 Zhenjun ZHU 1,∗,ChaoxuSUN2,JunZENG1 and Guowei CHEN 3 9 61 10 1 Jiangsu Key Laboratory of Urban ITS, School of Transportation, Southeast University, Nanjing, China 62 11 2 Zhejiang Provincial Natural Gas Development Co. Ltd, Hangzhou, China 63 12 3 Zhejiang Electric power construction Co. Ltd, Ningbo, China 64 13 65 Submitted 23 January 2017; resubmitted 13 April 2017; accepted 20 April 2017 14 66 15 67 Abstract. Atthephaseof13thfive-yearplaninChina,naturalgaswillplayanimportantroleinenergyrevolution.With 16 the growth of consumption, natural gas infrastructures will become hot spots of future investment and pipeline network 68 17 construction will also usher in a period of rapid development. Therefore, it is of great theoretical and practical significance 69 18 to study layout methods of transport pipeline network. This paper takes natural gas transport pipeline network as a research 70 19 object, introduces dominance degree to analyse benefits of pipeline projects. Then, this paper proposes Dominance Degree 71 Model (DDM) of transport pipeline projects based on Potential Model (PM) and Economic Potential Theory (EPT). Ac- 20 72 cording to DDM of gas transport pipeline projects, layout methods of pipeline network are put forward, which is simple 21 and easy to obtain the overall optimal solution and ensure maximum comprehensive benefits. What’s more, construction 73 22 sequences of gas transport pipeline projects can be also determined. Finally, the model is applied to a real case of natural gas 74 23 transport pipeline projects in Zhejiang Province, China. The calculation results suggest that the model should deal with the 75 24 transport pipeline network layout problem well, which have important implications for other potential pipeline networks 76 not only in the Zhejiang Province but also throughout China and beyond. 25 77 26 Keywords: natural gas, transport pipeline network, dominance degree model, potential model, economic potential theory, 78 27 layout method 79 28 80 29 81 30 82 1. Introduction or expanded. So when we make the pipeline network lay- 31 83 out, market supply and demand, benefits and construction 32 As an efficient and cleaner energy, natural gas is an im- 84 costs of pipeline projects’ investments, etc. – should be con- 33 portantbridgetomakeenergyconsumptiontransittolow- 85 sidered to determine a reasonable layout scheme. 34 carbon (Gillingham et al., 2009). According to 13th five- 86 A number of studies have been conducted on layout opti- 35 year plan and National Energy Development Strategy Plan 87 mization with respect to natural gas transport pipeline net- 36 88 of China, with respect to the proportion of primary energy work, where most of this research has focused on mathe- 37 89 consumption, natural gas demand will continue to increase, matical optimization. For example, Edgar et al. (1978)used 38 90 its share will reach 10% by the year of 2020. Continued Generalized Reduced Gradient Method to optimize natu- 39 growth in consumption is bound to promote the construc- ral gas transport pipeline network system. Bhaskaran and 91 40 tion of natural gas infrastructures, coverage area of trunk Salzborn (1979) studied the optimal design problem of gas 92 41 pipelines will in further expansion. Thus, gas transport and gathering pipeline network, which was divided into three 93 42 distribution network will be improved and facilities in dif- sub-problem: system layout, node locations, and the diam- 94 43 ferent pipe network will also achieve interoperability grad- eter distribution. Pedrycz et al. (1992)examinedthede- 95 44 ually. sign and application of neural networks in the context of a 96 45 Natural gas transport pipeline network shoulders the specific decision-making problem, the selection of an ap- 97 46 task of gas supply, which plays a significant role in im- propriate layout for natural gas distribution system. Singh 98 47 proving the overall socio-economic benefits for the region. and Nain (2012), Sanaye and Mahmoudimehr (2013)ap- 99 48 However, pipeline projects’ investment is large and pay- plied genetic algorithm to pipeline system optimization 100 49 back period is long, which cannot be easily reconstructed and simulated natural gas transport pipeline network. Üster 101 50 and Dilaveroğlu (2014) developed an integrated large-scale 102 51 103 *Corresponding author. E-mail: [email protected]. mixed-integer nonlinear optimization model to determine 52 104

Copyright © 2017 The Author(s) Published by VGTU Press

This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unre- stricted use, distribution, and reproduction in any medium, provided the original author and source are credited. [vgtuart2x 2017/10/19 v0.0.1] tran000001.tex 2017/10/20 14:40 p. 2/8

2 Z. Zhu et al. Optimization of natural gas transport pipeline network layout: A new methodology based on dominance degree model

1 Table 1. The methodology analysis 56 2 Methods Advantages Limitations 57 3 Graph theory ∙ easy access to computer program processing; ∙ only solving the network layout among known fixed 58 4 ∙ higher computing efficiency; points; 59 5 ∙ solving the shortest tree problem within a large scale ∙ project investment costs are not considered; 60 network effectively ∙ solutions can only be used as the initial network layout 6 61 7 Dynamic ∙ solving optimization problems with multiple ∙ not suitable for dealing with large-scale network 62 programming decision-making variables systems, ‘dimension obstacles’ exist in solving 8 63 Neural network ∙ solving optimization problems with multiple ∙ only obtaining the local optimal solution 9 64 method decision-making variables 10 65 Genetic algorithm ∙ higher adaptability, which can overcome the ∙ lower computing efficiency; 11 difficulties in solving nonlinear optimization ∙ no effective quantitative analysis concerning 66 12 algorithm precision, feasibility and computational 67 13 complexity 68 14 Complex method ∙ the algorithm is simple and suitable for dealing with ∙ unable to deal with multi-variable, multi-constraint 69 15 constrained optimization problems optimization problems 70 16 71 17 pipelines in the network. Schmidt et al. (2015)presented paper, which can not only determine the optimal layout 72 18 stationary neuro-linguistic programming type models of scheme, but also obtain construction sequences of gas 73 19 gas networks that are primarily designed to include detailed transport pipeline projects, which ensure maximum com- 74 20 nonlinear physics in the final optimization steps for mid- prehensive benefits of pipeline network. Specifically, we 75 21 term planning problems. An and Peng (2016)usedriskcost combine Potential Model (PM) and Economic Potential 76 22 functions to study the synchronization of the minimum Theory (EPT) to establish the Dominance Degree Model 77 23 risk loss and total cost of natural gas pipeline networks at (DDM) for pipeline projects, and use dominance degree to 78 24 the planning stage. However, these methods are essentially measure comprehensive benefits of pipeline projects. Then, 79 25 solving network structure to satisfy a given criterion, a local a layout method based on the DDM for natural gas trans- 80 26 optimal solution can be only obtained. port pipeline network is proposed, which can search out the 81 27 Minimum Spanning Tree Method (MSTM) and Dy- most advantageous pipeline project successively through 82 28 namic Programming (DP) are the most commonly used so- cyclecalculations.Thecalculationprocessissimpleand 83 29 lutions. MSTM abstracts pipeline network into undirected easy to obtain the overall optimal solution. To verify valid- 84 30 network, including classic solutions of Dijkstra Algorithms, ity of the method, natural gas transport pipeline projects in 85 31 Kruskal Algorithms, Steiner Algorithms (Graham and Hell, Zhejiang Province are taken as a case study. 86 32 1985). Compared to solutions of traditional graph theory, 87 33 these algorithms adopt computer programs for processing, 2. Dominancedegreeofpipelineprojects 88 34 whose operation efficiency is relatively high, Steiner Al- 89 35 gorithms effectively solve to the shortest path problem of Dominance generally refers to the advantageous form that 90 36 large-scale network (Han and Lim, 2010;Sniedovich,2006; one party can overcome or overwhelm its counterpart. 91 37 Zhou, 2004). However, the above three algorithms do not Dominance degree is an index, which can be used to re- 92 38 consider investment costs of pipeline projects, whose re- flect and compare the degree of the pros and cons among 93 39 sults can be only regarded as the initial pipeline network system elements (Shamsie, 2003). Due to differences in lo- 94 40 layout. DP can deal the optimization problem of multiple cation, economy and geographical conditions, transport 95 41 decision-making variables, while ‘dimension obstacles’ ex- pipeline projects have made a tremendous impact on social 96 42 ist in solving, that is computation will increase exponen- and economic benefits and influenced dominance degree of 97 43 tially with growth of the number of variables. When the network layout. Therefore, this paper uses dominance de- 98 44 dimension of this problem increases to a certain degree, it gree to reflect comprehensive benefits of transport pipeline 99 45 cannot be solved (Bellman and Dreyfus, 2016). Through projects and compare their comparative dominance. 100 46 theaboveanalysis,wecanseethatcurrentcommonlyused As for regional natural gas transport pipeline network, 101 47 methods have certain advantages and limitations, as shown dominance degree of transport pipeline projects are closely 102 48 in Table 1. related to influential factors, such as functional orientation, 103 49 In summary, layout optimization of regional natural gas scale, economic development level of cities and towns, con- 104 50 transmission pipeline network is a multi-objective nonlin- struction costs of transport pipeline projects, etc. 105 51 ear programming problem; the above methods are difficult 106 52 to solve the problem effectively in the practical application. 2.1. Functional orientation of cities and towns 107 53 Compared to the existing methods, a new methodology Functional orientation of cities and towns is the core of 108 54 considering practical application is proposed to simplify their development and competition, cities and towns that 109 55 computing and obtain the overall optimal solution in this play a dominant role in socio-economic development are 110 [vgtuart2x 2017/10/19 v0.0.1] tran000001.tex 2017/10/20 14:40 p. 3/8

Transport, 2017, 00: 1–8 3

1 important strategic nodes of regional natural gas trans- 3.2. Economic potential theory 56 2 57 port pipeline network layout. They have different develop- EPT is a manifestation trend of economic energy (Chang, 3 58 ment advantages, such as location, resource and policy, etc. 2008). Natural gas transport pipelines undertake the impor- 4 59 They can also maximize the optimal resource configura- tant task of transporting natural gas, which plays a signifi- 5 60 tion, which offers good conditions for expanding natural cant role in optimizing regional energy consumption struc- 6 61 gas consumption market. ture, promoting economic development and improving en- 7 62 vironment quality. There, we consider natural gas transport 8 2.2. Scale and economic development level of cities 63 pipelines will generate strong economic potential in its gas 9 and towns 64 supply region. 10 65 Natural gas consumption is closely related to scale and eco- According to the electromagnetic theory, current-car- 11 66 nomic development level of cities and towns, which de- rying conductors will generate a magnetic field in its sur- 12 67 termine consumption capacity and growth speed of nat- roundings. Thus, each point in vicinity of current-carrying 13 68 ural gas. In order to make natural gas consumption mar- conductors owns the potential condition (Novoselov et al., 14 69 ket have more potential, transport pipeline projects require 2004).Pipelineprojectsarejustlikecurrent-carryingcon- 15 70 cities and towns with a certain concentration of consumer ductors,theimpactoncitiesandtownsisrelatedtogas 16 71 groups and economic strength. Therefore, scale and eco- transport flow, the distance between cities and off-take sta- 17 72 nomic development level are the main factors that affect tions, development level. In this paper, pipeline projects are 18 73 natural gas consumption. simulated into current-carrying conductors, gas transport 19 74 flow is simulated into electric current, the city 𝑃 and off- 20 75 2.3. Construction costs of pipeline projects take station 𝐴 are simulated into nodes. Field intensity can 21 76 Construction costs of pipeline projects are related to its be used to describe economic potential of node 𝑃,which 22 77 length and costs of crossing obstacles, which directly affect can be repressed as follows: 23 78 24 the investment and benefits of projects. Natural gas trans- 79 port pipeline projects have a certain service life, the more 𝜇⋅𝐼 25 𝐵= , (2) 80 𝑎 26 construction costs we input, the longer payback period and 81 the weaker profitability it will be. Therefore, construction 27 where: 𝜇 is undetermined coefficient; 𝐼 is natural gas trans- 82 costsshouldbeconsideredtoensureprofitabilitywhen 28 port flow from node 𝐴 to 𝑃; 𝑎 is the distance between node 83 making natural gas transport pipeline network layout. 29 𝑃 (cities or towns) and node 𝐴 (off-take stations). 84 30 85 31 3. Related theories and models 4. Methodology 86 32 87 3.1. Potential model 33 Comprehensive benefits of pipeline projects are related to 88 34 PM is a commonly used model in the empirical study of re- functional orientation, scale, economic development level 89 35 gional economy, which is mainly used to measure a point of cities or towns. Through the analysis of these influential 90 36 influenced by a set of given points in the space. Firstly, inter- factors, we find that PM can be used to analyse the domi- 91 37 action of a node with each other node is calculated. Then, nance degree of pipeline projects. The specific construction 92 38 comprehensiveinfluenceisobtainedbysummingupinter- method is shown in Fig. 1. 93 39 action of a node with each other point (Mátyás, 1998;Lu 94 40 and Taur, 2006). Total potential of node 𝑖 can be expressed 4.1. Model construction 95 41 as follows: Based on PM and EPT, this paper proposes following as- 96 42 sumptionsandmethodswhenconstructingDDM: 97 𝑚 43 𝑗 98 𝑉𝑖 =𝐾⋅∑ , (1) ∙ pipeline projects will generate economic potential, 44 𝑗 𝑑𝑖𝑗 99 which can be regarded as an index reflecting city scale; 45 100 ∙ construction costs of pipeline projects will affect their 46 where: 𝐾 is the dielectric constant; 𝑚 is the scale of point 101 𝑗 economical benefits directly, therefore, construction 47 𝑗, according to different research questions, population and 102 costs are taken as parameter reflecting impedance in 48 economy scale can be selected; 𝑑 is the impedance between 103 𝑖𝑗 PM; 49 𝑖 and 𝑗. 104 ∙ the city’s economic potential influenced by pipeline 50 In terms of cities and towns within the study area, PM 105 projects is closely related to its development level, 51 can be used to analyse potential social and economic ben- 106 pipeline projects will directly affect the transformation 52 efits generated by transport pipeline projects and compare 107 of economic potential. 53 their advantages and disadvantages. Therefore, this paper 108 54 introduces PM to analyse dominance degree of natural gas Therefore, some specific indicators can be selected to re- 109 55 transport pipeline projects. flect these influential factors. Above all, the DDM of gas 110 [vgtuart2x 2017/10/19 v0.0.1] tran000001.tex 2017/10/20 14:40 p. 4/8

4 Z. Zhu et al. Optimization of natural gas transport pipeline network layout: A new methodology based on dominance degree model

1 56 2 57 3 58 4 59 5 60 6 61 7 62 8 63 9 64 10 65 11 66 12 67 13 68 14 69 15 70 16 71 17 72 18 73 19 Fig. 1. Flow chart of model construction. 74 20 75 Table 2. Calibration for the value of 𝐾𝑖 21 transport pipeline projects is established as follows: 76 𝑃 [10000 people] 𝐾 22 𝑖 77 (𝜇𝑖 ⋅𝐼𝑖)/𝑎𝑖𝑗 <20 1 23 𝐴 =𝐾 ⋅ , (3) 78 𝑖𝑗 𝑖 𝑓 24 𝑖𝑗 20–50 1.25 79 25 50–100 1.5 80 where: 𝐴𝑖𝑗 is the dominance degree of pipeline projects be- 26 100–300 2.0 81 tween city 𝑖 and gas transport station 𝑗; 𝐾𝑖 is the dielec- 27 300–500 2.5 82 tric constant, which is a constant within a specific period 28 500–1000 3.0 83 for specific cities; 𝜇𝑖 is the sensitivity coefficient and reflects 29 84 sensitive degree that cities put on pipeline projects; 𝐼𝑖 is the 30 85 𝐼𝑖 is the gas demand of city 𝑖,whichisrelatedtolocalen- gas demand of city 𝐼; 𝑎𝑖𝑗 is the distance between city 𝑖 and 31 gas transport station 𝑗, which can be expressed by straight- ergy consumption structure and policies. In order to facili- 86 32 tate analysis of this question, energy consumption per unit 87 line distance; 𝑓𝑖𝑗 is construction costs of pipeline projects 33 between city 𝑖 and natural gas transport station 𝑗. GDP is selected to reflect energy consumption and saving 88 34 status. To some extent, it reflects current gas demand and 89 35 4.2. Parameter calibration development potential of future natural gas. 90 36 91 Parameter calibration is critical for model application, this 𝑓𝑖𝑗 is construction costs of pipeline projects between city 𝑖 37 paper proposes following calibration methods for model and gas transport station 𝑗,whichwilldirectlyaffectinvest- 92 38 parameters. ment income. Natural gas transport pipeline projects will 93 39 94 𝐾𝑖 is the dielectric constant, As for DDM of natural gas cross rivers, roads, railways, etc. Thus, construction costs 40 transport pipeline projects, which is closely related to urban are mainly determined by pipeline length and investment 95 41 functional orientation of cities and towns. Functional ori- for crossing obstacles. Therefore, the calculation method of 96 42 entation is a manifestation of government functions, which 97 𝑓𝑖𝑗 is shown as follows: 43 guides many planning system, such as urban policies, in- 98 44 99 dustry structure, etc. City scale is of great significance for 4 45 100 functional orientation and relatively stable in a long period 𝑓𝑖𝑗 =𝑉⋅𝐺⋅𝛽⋅𝑎𝑖𝑗 + ∑ 𝑚𝑘 ⋅𝑤𝑘, (4) 46 of time. Therefore, in order to simplify this question, ur- 𝑘=1 101 47 ban scale 𝑃 is selected to reflect this parameter. With the 102 48 reference to notification about adjusting classification cri- where: 𝑉 is pipeline price per unit weight; 𝐺 is pipeline 103 49 weight per unit length (the value of 𝑉 and 𝐺 can take 104 teria for city scale of 2014, 𝐾𝑖 canbecalibratedasshownin 50 Table 2. pipelines of the same diameter, material, thickness as stan- 105 51 106 𝜇𝑖 is related to the richness of urban resources, market dard); 𝛽 is terrain correction coefficient, which is used to 52 development level, living standards of residents, the exist- modify pipeline route length (different topography condi- 107 53 ing state of industry. In this paper, Gross Domestic Product tions are endowed with different values, which can be de- 108 54 (GDP) per capital of macroeconomic indicators is used to termined by referring to engineering data within the study 109 55 110 represent this parameter. area); 𝑚𝑘 is crossing times of the 𝑘th type’ 𝑤𝑘 is the corre- [vgtuart2x 2017/10/19 v0.0.1] tran000001.tex 2017/10/20 14:40 p. 5/8

Transport, 2017, 00: 1–8 5

1 56 2 57 3 58 4 59 5 60 6 61 7 62 8 63 9 64 10 65 11 66 12 67 13 68 14 69 15 70 16 71 17 72 18 73 19 74 20 75 21 76 22 77 23 78 24 79 25 80 26 81 27 82 28 83 29 84 30 85 31 86 32 Fig. 2. Natural gas transport pipeline network layout process. 87 33 88 34 sponding construction costs (𝑘 = 1, 2, 3, 4 respectively rep- (GIS), their locations in the map can be determined. 89 35 resents the river, grade highway, substandard highway and Then, connect these nodes based on pipeline align- 90 36 railway). ments and obtain spatial structure diagram of regional 91 37 92 pipeline network; 38 4.3. The layout method based on DDM 93 – Step 2: Calculate the dominance degree of pipeline 39 94 DDM of pipeline projects fully embody comprehensive projects.Similarly,abstractcitiesandtownsthatun- 40 socio-economic benefits of the projects. Apply DDM to nat- 95 covered by gas transmission pipeline network into a set 41 ural gas transport pipeline network layout will maximum 96 of nodes 𝑖 = {1,2,...,𝑛}. Calibrate model parameters 42 socio-economic benefits. Through the analysis of DDM of 97 43 and use DDM to calculate dominance degree between 98 pipeline projects, the optimal layout scheme and construc- 44 tion sequences are determined, which will give full play to node 𝑖 and node 𝑗 sequentially. Then, sort calculation 99 45 100 advantages of pipeline projects and promote regional socio- results in descending order. 46 101 economic development. The basic application process is – Step 3:Updatepipelinenetworkintheregionalarea. 47 102 shown in Fig. 2. According to calculation results, take the pipeline 48 103 According to the layout process, layout methods of nat- project of maximum dominance degree and connect 49 104 ural gas transport network based on DDM have four steps, its nodes at both ends to form a new pipeline route. 50 105 which are shown as follows: 51 Then, corresponding nodes 𝑖 is incorporated into a set 106 52 – Step 1: Abstract pipeline network in the regional area. of node 𝑗,thatis𝑗 = {1,2,...,𝑚+.Forexample, 1} 107 53 First, abstract gas transport stations and valve chests assume that 𝑖={1,2,3,4}and 𝑗={1,2},thecalculated 108 54 into spatial nodes, which record a set of nodes 𝑗= maximum dominance degree 𝐴21,whichmeans𝑖=2 109 55 {1,2,...,𝑚}. Using Geographic Information System is removed from a set of node 𝑖,subsumedintoaset 110 [vgtuart2x 2017/10/19 v0.0.1] tran000001.tex 2017/10/20 14:40 p. 6/8

6 Z. Zhu et al. Optimization of natural gas transport pipeline network layout: A new methodology based on dominance degree model

1 56 2 57 3 58 4 59 5 60 6 61 7 62 8 63 9 64 10 65 11 66 12 67 13 68 14 69 15 70 16 71 17 72 18 73 19 74 20 75 21 76 22 77 23 78 24 79 25 80 26 81 27 Fig. 3. Natural gas transport pipeline network. 82 28 83 of node 𝑗 and denoted as 𝑗=3,then𝑖 = {1,3,4}and 5.1. Pipeline network and model parameters 29 84 𝑗={1,2,3}. Finally, updated spatial structure diagram 30 There are three natural gas transport stations and two valve 85 31 of regional pipeline network is obtained. chests, GIS is used to determine their locations. Then, ab- 86 32 – Step 4: Calculate cyclically and output the results. Re- stract pipeline network in the regional area as shown in 87 33 peat Step 2 and Step 3 until the set of nodes 𝑖= Fig. 3. 88 34 {1,2,...,𝑛}is an empty set, while the set of nodes 𝑗= According to Statistical Yearbook of Lishui City of 2015 89 35 {1,2,...,𝑚+𝑛}.Atthistime,allcitiesandtownshave and energy consumption statistics of each county of Lishui 90 36 been covered by pipeline network and layout scheme city (Mátyás, 1998), input parameters are shown in Table 3. 91 37 of pipeline network is obtained. Take related data of Lishui section in the Jinhua–Lishui– 92 38 93 Wenzhou pipeline engineering as reference, use L450M 39 94 steel for standard pipelines. The value of 𝛽 is 1.032; 𝑉 is 0.77 40 5. Case study 95 ten thousand yuan/tons; 𝐺 is 222.3 tons/km; 𝑤 , 𝑤 , 𝑤 , 𝑤 41 1 2 3 4 96 In this paper, Lishui region along Jinhua–Lishui–Wenzhou are respectively 250, 15, 4, 25 thousand yuan/time. 42 97 43 gas transport trunk pipeline engineering in Zhejiang 98 5.2. Model calculation results 44 Province,Chinaistakenasacasestudy.DDMisusedto 99 45 determine the optimal layout scheme and construction se- Take relevant parameters into the DDM, through cycle cal- 100 46 quences. culation, the results are obtained as shown in Table 4. 101 47 102 Table 3. Input parameters of cities and towns 48 103 Name 𝐾𝑀[10000 yuan/person] 𝐼 [tons of standard coal/10000 yuan] 49 104 50 Suiyang County 1.25 46.596 0.470 105 51 Songyang County 1.25 44.037 0.456 106 52 Yunhe County 1 45.627 0.532 107 53 Jingning County 1 38.541 0.458 108 54 Longquan City 1.25 39.546 0.491 109 55 Qinyuan County 1.25 43.943 0.490 110 [vgtuart2x 2017/10/19 v0.0.1] tran000001.tex 2017/10/20 14:40 p. 7/8

Transport, 2017, 00: 1–8 7

1 Table 4. Model calculation results County, Songyang County to Yunhe County, Yunhe County 56 2 Cycle times 𝑖𝑗𝐴𝑖𝑗 to Jingning County, Yunhe County to Longquan City, 57 3 1SongyangTerminal station 6.63 Longquan City to Qinyuan County. Gas transport sta- 58 4 County in Lishui City tions can be also obtained, which is arranged at Songyang 59 5 2SongyangSuichang 18.85 County, Yunhe County and Longquan City. 60 6 County 61 7 3SongyangYunhe 8.69 62 County 6. Conclusions 8 63 4 Yunhe County Jingning 28.52 9 The layout optimization of natural gas transport pipeline 64 5 Longquan City Yunhe County 6.14 10 is a subject with social and economic benefits, but it is 65 6QinyuanLongquan City 5.22 11 also a complex system problem. Based on influential fac- 66 County 12 tors analysis of natural gas transport pipeline projects’ dom- 67 13 inance degree, this paper proposes layout method of nat- 68 14 According to calculation results, update natural gas ural gas transport pipeline network based on DDM. The 69 15 transport pipeline network within the study area succes- model can not only determine the optimal layout scheme, 70 16 sively. Finally, the layout scheme of natural gas transport but also obtain construction sequences of gas transport 71 17 pipeline network is obtained, as shown in Fig. 4. pipeline projects, which ensure maximum comprehensive 72 18 Based on considering economic benefits and costs, the benefitsofpipelinenetwork.Naturalgastransportpipeline 73 19 layout scheme calculated by DDM is the overall opti- projects in Zhejiang Province, China is taken as a case study, 74 20 mal solution. Meanwhile, according to cycle calculation which suggest that the model should deal with the transport 75 21 results, construction sequences of natural gas transport pipeline network layout problem well. The study finding 76 22 pipeline projects can be determined and maximum com- has important implications for other potential pipeline net- 77 23 prehensive benefits of natural gas transport pipeline net- works not only in the Zhejiang Province but also through- 78 24 work are ensured. Under the condition of funding con- out China and beyond. 79 25 straints, the optimal construction sequence in Lishui re- With respect to calculating construction costs of pipeline 80 26 gion is arranged successively: from teminal station in Lishui projects, this paper mainly considers pipeline material con- 81 27 City to Songyang County, Songyang County to Suichang sumption and costs of crossing obstacles, etc. However, 82 28 83 29 84 30 85 31 86 32 87 33 88 34 89 35 90 36 91 37 92 38 93 39 94 40 95 41 96 42 97 43 98 44 99 45 100 46 101 47 102 48 103 49 104 50 105 51 106 52 107 53 108 54 109 55 Fig. 4. Layout scheme of natural gas transport pipeline network. 110 [vgtuart2x 2017/10/19 v0.0.1] tran000001.tex 2017/10/20 14:40 p. 8/8

8 Z. Zhu et al. Optimization of natural gas transport pipeline network layout: A new methodology based on dominance degree model

1 during pipeline engineering design and construction pro- Gillingham, K., Newell, R. G., & Palmer, K. (2009). Energy ef- 56 2 cess, construction costs of pipeline projects are also influ- ficiency economics and policy. Annual Review of Resource 57 3 enced by proportion of tunnels, topography and policy, etc. Economics, 1, 597–620. http://dx.doi.org/10.1146/annurev. 58 resource.102308.124234. 4 Therefore, how to quantify construction costs of natural gas 59 5 transport pipeline projects comprehensively need to be fur- Han, D.-M., & Lim, J.-H. (2010). Design and implementation 60 of smart home energy management systems based on Zig- 6 ther researched. In addition, the layout scheme of natural 61 Bee. IEEE Transactions on Consumer Electronics, 56(3), 1417– 7 62 gas transport pipeline network can be obtained by DDM. 1425. http://dx.doi.org/10.1109/TCE.2010.5606278. 8 However, distribution of natural gas, pipeline network tech- Lu, H., & Taur, Y. (2006). An analytic potential model for sym- 63 9 nology and changes of airflow direction under different metric and asymmetric DG MOSFETs. IEEE Transactions 64 10 conditions should also be researched to determine the lo- on Electron Devices, 53(5), 1161–1168. http://dx.doi.org/10. 65 11 cation and quantity of natural gas stations and valve chests 1109/TED.2006.872093. 66 12 reasonably and implement rational utilization of natural gas Mátyás, L. (1998). The gravity model: Some econometric consid- 67 13 resources. erations. The World Economy, 21(3), 397–401. http://dx.doi. 68 14 org/10.1111/1467-9701.00136. 69 15 Contribution Novoselov,K.S.,Geim,A.K.,Morozov,S.V.,Jiang,D.,Zhang, 70 16 Y., Dubonos, S. V., Grigorieva, I. V., & Firsov, A. A. (2004). 71 Electric field effect in atomically thin carbon films. Sci- 17 Zhenjun Zhu and Chaoxu Sun conceived and designed the 72 ence, 306(5696), 666–669. http://dx.doi.org/10.1126/science. 18 research; Zhenjun Zhu and Jun Zeng analyzed the data; 73 1102896. 19 Zhenjun Zhu wrote the paper; Chaoxu Sun and Guowei 74 Pedrycz, W., Davidson, J., & Goulter, I. (1992). Neural network 20 Chen helped improve the figures and manuscript. 75 based decision model, used for design of rural natural gas sys- All authors have read and approved the final manuscript. 21 tems. In IEEE International Conference on Fuzzy Systems 1992, 76 22 San Diego, CA, US, 8–12 March 1992 (pp. 1219–1226). http:// 77 23 Disclosure statement dx.doi.org/10.1109/FUZZY.1992.258651. 78 24 79 The authors declare that there are no conflict competing fi- Shamsie, J. (2003). The context of dominance: An industry-driven 25 framework for exploiting reputation. Strategic Management 80 nancial, professional and personal interests from other par- 26 Journal, 24(3), 199–215. http://dx.doi.org/10.1002/smj.291. 81 ties. 27 Sniedovich, M. (2006). Dijkstra’s algorithm revisited: The dy- 82 28 namic programming connexion. Control and Cybernetics, 83 References 35(3), 599–620. 29 84 30 An, J., & Peng, S. (2016). Layout optimization of natural gas net- Sanaye, S., & Mahmoudimehr, J. (2013). Optimal design of a nat- 85 ural gas transmission network layout. Chemical Engineering 31 work planning: Synchronizing minimum risk loss with total 86 cost. Journal of Natural Gas Science and Engineering, 33, 255– Research and Design, 91(12), 2465–2476. http://dx.doi.org/10. 32 87 263. http://dx.doi.org/10.1016/j.jngse.2016.05.017. 1016/j.cherd.2013.04.005. 33 88 Bellman, R. E., & Dreyfus, S. E. (2016). Applied Dynamic Program- Singh, R. R., & Nain, P. K. S. (2012). Optimization of natural gas 34 89 ming. Princeton: Princeton University Press. 390 p. pipeline design and its total cost using GA. International Jour- 35 90 Bhaskaran, S., & Salzborn, F. (1979). Optimal diameter assign- nal of Scientific and Research Publications, 2(8), 1–10. 36 ment for gas pipeline networks. The Journal of the Australian Schmidt, M., Steinbach, M. C., & Willert, B. M. (2015). High de- 91 37 Mathematical Society. Series B. Applied Mathematics, 21(2), tail stationary optimization models for gas networks. Opti- 92 38 129–144. http://dx.doi.org/10.1017/S0334270000002009. mization and Engineering, 16(1), 131–164. http://dx.doi.org/ 93 39 Edgar, T. F., Himmelblau, D. M., & Bickel, T. C. (1978). Optimal 10.1007/s11081-014-9246-x. 94 40 design of gas transmission networks. Society of Petroleum En- Üster, H., & Dilaveroğlu, S. (2014). Optimization for design and 95 41 gineers Journal, 18(2). http://dx.doi.org/10.2118/6034-PA. operation of natural gas transmission networks. Applied En- 96 42 Chang, S. J. (2008). A cultural and philosophical perspective on ergy, 133, 56–59. http://dx.doi.org/10.1016/j.apenergy.2014. 97 06.042. 43 Korea’s education reform: A critical way to maintain Korea’s 98 economic momentum. Academic Paper Series, 3(2), 1–11. 44 Zhou, H. (2004). Efficient Steiner tree construction based on span- 99 ning graphs. IEEE Transactions on Computer-Aided Design of 45 Graham, R. L., & Hell, P. (1985). On the history of the minimum 100 spanning tree problem. Annals of the History of Computing, Integrated Circuits and Systems, 23(5), 704–710. http://dx.doi. 46 101 7(1), 43–57. http://dx.doi.org/10.1109/MAHC.1985.10011. org/10.1109/TCAD.2004.826557. 47 102 48 103 49 104 50 105 51 106 52 107 53 108 54 109 55 110