Beruflich Dokumente
Kultur Dokumente
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 7 February 2016
Received in revised form
3 May 2016
Accepted 4 May 2016
Available online 12 May 2016
The synchronization of the minimum risk loss and total cost of natural gas pipeline networks at the
planning stage is discussed in this article. Herein, new procedures for optimizing layout are proposed to
minimize the investment cost, operation expense, and the risk loss of the pipeline network. The procedures include two crucial steps: the rst step is tting two risk cost functions (i.e., leakage risk cost
function and corrosion risk cost function), and the second one is achieving the optimal layout by using
the risk cost functions as the edge weight of the minimum spanning tree algorithm. The suggested
method is applied in three different real cases, leading to three distinct optimal layouts, which are more
suitable than that calculated using intelligent algorithms for practical engineering. Then, two optimal
strategies for pipeline network layouts are presented. Different applications that respectively focus on
the leakage risk cost for urban areas and the corrosion risk cost or leakage risk cost for suburban areas are
shown in the above two strategies. These strategies realize a 6.9e21% greater economic benet than that
of the shortest layout.
2016 Elsevier B.V. All rights reserved.
Keyword:
Natural gas network planning
Layout optimization
Nonlinear tting
Risk cost function
Minimum spanning tree
1. Introduction
The investment and operation costs of natural gas pipeline
networks are commonly considered important aspects of the total
investment. Moreover, environmental and social economic losses
should not be ignored at the planning stage of the network design
because the losses caused by natural gas leakage are so huge during
construction and operation stages. Hence, guaranteeing the gas
transmission to individuals with minimum risk loss and effective
cost necessitates optimizing the pipeline network. Safety and reliability at the operation stage of pipeline networks have been
extensively treated previously for their important roles in risk loss
of natural gas transportation (Rios-Mercado and Borraz-Sanchez,
2015).
Pipe diameter and layout are calculated based on the ow horizon and node pressure horizon (Hanbing et al., 2015), respectively, which is a common optimization method in natural gas
network design that has led to a great many optimization models
and algorithms in recent years. The optimization method usually
consists of two steps: rstly, the shortest pipeline layout or
* Corresponding author.
E-mail address: 450044789@qq.com (J. An).
http://dx.doi.org/10.1016/j.jngse.2016.05.017
1875-5100/ 2016 Elsevier B.V. All rights reserved.
256
J. An, S. Peng / Journal of Natural Gas Science and Engineering 33 (2016) 255e263
gross national economies (Gangya et al., 2007). Every year, economic losses worth 2.8 1011 RMB are caused by corrosion, which
has become a critical factor in inuencing the safety and reliability
of pipeline operation (Borraz-Sanchez and Haugland, 2013). Additionally, the corrosion has even brought about a large amount of
social issues (Ossai et al., 2015).
Synchronizing the layout, risk loss and planning stages in one
model is challenging because the mathematical optimization of
natural gas pipelines is difcult to develop. Previous approaches
usually minimized the risk loss by employing network security
assessment technology and integrity management technology,
which are based on the deterministic layout (Stalheim, 2011;
Suresh et al., 2008; Reitenbach et al., 2015; Wu et al., 2007), and
therefore synchronization of the risk loss and the planning stage
could not be achieved.
Overall, three limitations exist in previous studies according to
the above discussion: 1) common total optimization or operation
optimization is created only after the layout is created; 2) Optimal
layout is obtained by using intelligent algorithms or conventional
optimization algorithms, which are not consistent with the actual
pipeline laying path; 3) Common total optimization does not
consider economic loss caused by environmental disruption or
casualties derived from the risk factors.
Synchronizing total cost, risk loss and the layout at the planning
stage could not be found in any of the literature. The objective of the
present work is to propose the problem of new layout optimization,
in which new procedures are presented and the two risk costs are
developed based on the above three limitations. In addition, two
strategies for optimizing the pipeline network layout are obtained
by applying the procedures.
(1)
(2)
J. An, S. Peng / Journal of Natural Gas Science and Engineering 33 (2016) 255e263
257
Table 1
Actual cost of medium pressure pipe network in the city of Chongqing.
Pipe section number
Resistivity (U m)
1
2
3
4
5
6
7
8
9
10
11
8.82
10.0
14.110
17.030
21.200
23.630
24.192
29.250
48.680
87.630
200.000
12.19
14.91
15.06
15.73
16.47
18.61
19.03
20.08
23.39
22.00
25.00
0.014421
0.024726
0.033159
0.037172
0.040689
0.041171
0.042542
0.042947
0.046142
0.055419
0.058798
0.033650
0.074178
0.131791
0.099479
0.163596
0.109395
0.144262
0.123515
0.152270
0.142506
0.140196
0.048071
0.098904
0.16495
0.136651
0.204285
0.150566
0.186804
0.166462
0.198412
0.197925
0.198994
P
yy
y y b
R q
2
P b
P
y y2
yy
(4)
cb
cX b
cX
b
yb
0
1 1
k k
(5)
258
J. An, S. Peng / Journal of Natural Gas Science and Engineering 33 (2016) 255e263
parties, and the surrounding construction, relate to the soil chemistry, which contains soil resistivity, redox potential, soil pH value,
soil moisture content, soil salt content, and tube potential. Herein,
two elements, resistivity and moisture content, are selected to be
the independent parameters of the risk cost function to calculate
the values of edge weight.
3. Case study
The pipeline networks of the three cases are medium pressure
of urban areas, medium-high pressure A of suburban areas, and the
medium-high pressure B of suburban areas. The main differences
are as follows.
Table 2
Actual cost of testing pipelines.
Pipe section number
1
2
3
4
5
6
7
8
9
10
11
0.041902889
0.046715006
0.023411245
0.035308873
0.035085928
0.046280185
0.045220648
0.064144646
0.038245711
0.04545533
0.026107685
0.163934132
0.19026419
0.118398515
0.161472747
0.175518197
0.189309901
0.178814871
0.24665886
0.115152762
0.170932966
0.118014426
0.122031244
0.143549184
0.09498727
0.126163874
0.140432269
0.143029716
0.133594224
0.182514215
0.076907051
0.125477636
0.091906741
0.109395
0.15227
0.099479
0.123515
0.140196
0.144262
0.131791
0.163596
0.074178
0.123515
0.074178
J. An, S. Peng / Journal of Natural Gas Science and Engineering 33 (2016) 255e263
test point
two cost
fitting
functions
program
minimum
spanning tree
259
Optimal
layout
actual cost
Fig. 4. Procedure for solving optimal layout of pipe network.
Table 3
The two costs of the rst natural gas pipelines case.
Starting node
Termination node
Pit no.
1
1
1
2
2
2
2
3
3
3
4
4
4
5
5
5
6
6
7
7
7
8
8
9
10
10
4
2
13
15
3
5
15
16
6
14
16
5
7
13
6
9
14
11
12
8
9
13
9
10
12
11
12
15
0.825
1.48
0.979
1.00
1.14
0.862
0.503
0.511
0.446
0.837
0.289
1.32
0.549
0.691
0.781
0.702
1.27
1.16
0.657
0.733
0.832
0.843
0.276
0.539
0.810
0.648
0.183
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
0.022461458
0.072329366
0.050821858
0.041902889
0.053255107
0.020180493
0.017760363
0.017928909
0.020640963
0.037849682
0.018537803
0.050484339
0.024954976
0.01804041
0.054264378
0.031825148
0.070539887
0.029105163
0.041615141
0.035483075
0.050647299
0.034487709
0.014111916
0.032425641
0.030979026
0.029455054
0.009131175
0.091683253
0.30290319
0.188138189
0.163934132
0.216901177
0.10205952
0.081220792
0.089689799
0.084432216
0.149668047
0.071284411
0.152001646
0.093842198
0.081547968
0.237328175
0.153583257
0.214610017
0.13648955
0.135332818
0.140560768
0.193279671
0.149856269
0.058295257
0.128157321
0.093273738
0.110764562
0.03745357
Table 4
Physical and chemical index of soil test point.
Pit no.
Soil texture
The value of pH
Resistivity(U m)
Natural potential(-V)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
Loam
Loam
Clay
Clay
Clay
Clay
Sandy
Sand
Clay
Clay
Clay
Loam
Loam
Sandy
Clay
Clay
Sandy
Sandy
Loam
Clay
Clay
Loam
Loam
Clay
Clay
Clay
Loam
Loam
Loam
Sandy
Sandy
Sandy
8.30
6.46
5.52
6.58
6.99
6.99
5.95
5.50
6.27
7.24
7.05
8.26
8.30
8.25
5.35
8.02
4.84
8.38
8.15
5.52
5.70
7.65
6.31
8.33
4.91
5.38
6.30
4.35
6.89
5.84
7.76
6.79
10.000
8.820
200.000
14.110
24.560
24.560
48.680
87.630
21.200
29.250
24.192
17.030
23.630
27.690
38.880
63.370
105.500
27.810
25.110
54.260
241.900
110.770
21.040
88.900
25.900
58.600
40.350
51.360
78.800
105.500
27.810
48.680
16.47
25.00
15.06
20.08
21.84
21.84
22.00
12.19
18.61
19.03
23.39
15.73
14.91
18.79
10.32
9.61
22.94
30.46
12.13
26.76
22.56
18.01
13.12
20.35
18.33
21.13
11.47
33.36
13.23
22.94
30.46
22.00
0.550
0.542
0.550
0.542
0.540
0.540
0.760
0.772
0.748
0.810
0.785
0.787
0.840
0.820
0.880
0.870
0.825
0.861
0.878
0.850
0.829
0.841
0.880
0.815
0.834
0.690
0.818
0.820
0.821
0.825
0.861
0.760
loam soil
loam soil
loam soil
loam soil
loam soil
loam soil
loam soil
260
J. An, S. Peng / Journal of Natural Gas Science and Engineering 33 (2016) 255e263
1 The same area is used for the rst case and second case so that
the form and parameter values of the two cost functions are the
same but the independent variables are not;
2 A different area is used for the third case so that the form of the
cost function is the same as that of the rst and second case but
the parameter values and independent variable are not the
same.
3 The rst case is for urban areas while the second and third cases
depict suburban areas.
Soil data are shown in Table 3, and the node number and the
length of the laying path are shown in Table 4. Data was obtained
from the geographic data of each node.
The test points (Pit.) are selected between each section of the
pipeline, and soil resistivity, moisture content, natural potential, pH
value and soil texture are chosen as the basic soil physical and
chemical index. All of the data of Pit No. are shown in Table 4, and
they were chosen at random. The more test points that are determined, the more accurate the tted cost function and the optimization results become.
3.1. The rst case
The two costs of the rst example are calculated by applying
steps 1 to 5 of the new procedures and are shown in the Table 3.
Then, the results are as follows by using the minimum spanning
tree Kruskal algorithm.
(1) Shortest path layout without focused on soil erosion risk cost
T Columns 1 through 11
4
15
8
10
4
5
3
14
2
16
3
6
9
12
4
13
10
12
7
8
5
6
c 0.3396.
Here, the savings cost is calculated
0.390140213e0.3396 0.050540213 m.
as
follows:
10
12
8
9
4
5
2
16
4
15
3
14
3
6
1
2
9
12
4
13
2
15
Columns 12 through 15
10
11
6
12
7
8
3
16
c 1.4211.
Here, the cost savings are calculated
1.306645949e1.4211 0.114454051 m.
as
follows:
10
12
8
9
4
5
8
10
10
11
3
14
4
13
2
16
3
6
6
12
7
8
Columns 12 through 15
Columns 12 through 15
5
9
10
11
1
2
3
16
7
13
4
15
3
16
1
2
c 0.3731.
Here, the cost savings are calculated
0.371955139e0.3731 0.001144861 m.
as
follows:
T Columns 1 through 11
T Columns 1 through 11
4
5
10
12
8
9
2
16
3
6
4
15
4
5
2
15
3
14
1
2
4
13
6
11
10
12
8
9
3
14
2
16
3
6
1
2
4
13
2
15
2
3
4
15
Columns 12 through 15
Columns 12 through 15
10
11
9
12
5
9
7
13
8
10
5
9
7
8
10
11
J. An, S. Peng / Journal of Natural Gas Science and Engineering 33 (2016) 255e263
261
Table 5
The two costs of the second natural gas pipelines case.
(1) The length of the pipeline is not changed, so the total length of the shortest layout is also 8.5450 km. However, the area of the network changes as the pipelines are
planned further away from the crowded city such that the leakage risk cost is smaller. The leakage risk cost and the corrosion risk cost are 0.371955139 m and
1.621615599 m, respectively.
(2) Layout optimization results focused on leakage risk cost.
Starting node
Termination node
1
1
1
2
2
2
2
3
3
3
4
4
4
5
5
5
6
6
7
7
7
8
8
9
10
10
4
2
13
15
3
5
15
16
6
14
16
5
7
13
6
9
14
11
12
8
9
13
9
10
12
11
12
15
0.825
1.48
0.979
1.00
1.14
0.862
0.503
0.511
0.446
0.837
0.289
1.32
0.549
0.691
0.781
0.702
1.27
1.16
0.657
0.733
0.832
0.843
0.276
0.539
0.810
0.648
0.183
0.057322
0.094934
0.062011
0.060874
0.068581
0.047878
0.026112
0.026127
0.022254
0.040905
0.01399
0.061664
0.025408
0.03141
0.035501
0.031825
0.05743
0.048607
0.026878
0.028034
0.03182
0.029765
0.009684
0.014675
0.021147
0.016259
0.004284
0.091683
0.170426
0.112735
0.117663
0.134536
0.10206
0.081221
0.08377
0.075367
0.143071
0.0494
0.231684
0.097593
0.123561
0.147851
0.133565
0.243536
0.222922
0.134464
0.150019
0.17138
0.178054
0.060383
0.125214
0.192593
0.159835
0.05561
T Columns 1 through 11
c 1.5069.
Here, the cost savings are calculated
1.621615599e1.5069 0.114715599 m.
as
follows:
8
9
10
12
10
12
8
10
2
3
4
5
2
16
2
15
1
2
3
14
3
6
4
15
2
16
4
5
2
3
3
14
2
15
1
2
3
6
Columns 12 through 15
7
13
8
9
8
10
10
11
4
13
7
9
2
5
c 1.2728.
Here,
cost
savings
are
calculated
1.367118542e1.2728 0.094318542 m.
Columns 12 through 15
as
follows:
4
13
10
11
c 0.3087.
Here, the cost savings are calculated
0.390021523e0.3087 0.081321523 m.
7
9
as
follows:
262
J. An, S. Peng / Journal of Natural Gas Science and Engineering 33 (2016) 255e263
Table 6
The two costs of the third natural gas pipelines case.
(1) The length of the pipeline is not changed, so the total length of the shortest layout is also 8.5450 km. However, the area of the network changes such that the moisture
content is different. The leakage risk cost and corrosion risk cost is equal to 0.390021523 m and 1.367118542 m.
(2) Layout optimization focus on leakage risk cost.
Starting node
Termination node
1
1
1
2
2
2
2
3
3
3
4
4
4
5
5
5
6
6
7
7
7
8
8
9
10
10
4
2
13
15
3
5
15
16
6
14
16
5
7
13
6
9
14
11
12
8
9
13
9
10
12
11
12
15
0.825
1.48
0.979
1.00
1.14
0.862
0.503
0.511
0.446
0.837
0.289
1.32
0.549
0.691
0.781
0.702
1.27
1.16
0.657
0.733
0.832
0.843
0.276
0.539
0.810
0.648
0.183
0.021457049
0.09309708
0.070573764
0.018404647
0.023840308
0.021251292
0.021136758
0.023538682
0.021490006
0.045411386
0.019833192
0.053766664
0.02590588
0.027725225
0.049054738
0.041064991
0.070925767
0.061850816
0.04640089
0.043801986
0.030892375
0.025834587
0.007131375
0.011428614
0.03293063
0.027983748
0.008593352
0.091683253
0.370539631
0.267117646
0.085804739
0.112260523
0.091470661
0.084116133
0.096737359
0.089511536
0.215683657
0.085721558
0.155142863
0.122649225
0.064960059
0.157190087
0.148558934
0.313940691
0.282300194
0.192369924
0.185521459
0.124300126
0.105431089
0.031162877
0.051051451
0.093273738
0.110764562
0.037433211
Table 7
The two costs and their benets in three cases.
Name of cost
Case 1
Case 2
Case 3
0.390140213
1.306645949
0.3396
0.050540213(13%)
1.4211
0.114454051
0.371955139
1.621615599
0.3731
0.001144861
1.5069
0.1147155(7.1%)
0.390021523
1.367118542
0.3087
0.0813215(21%)
1.2728
0.094318(6.9%)
5. Conclusions
In the present paper, new efcient optimization procedures are
built to enhance safety and reliability of natural gas pipeline
transportation. The main thrust of this study is the two cost function that was dened based on unique independent parameters to
be used as the edge weight. The new optimization procedures are
applied successfully to three different cases, and then the result
shows that the regular layout cost with the shortest total length is
not less than the optimal layout cost of minimizing pipeline
corrosion risk. As expected, choosing between corrosion risk cost
and leakage risk cost for the deciding factor is crucial to ensure the
best economic and social environmental benets. Finally, leakage
risk cost could be suitable for low pressure pipelines of urban areas,
as shown in case 1; corrosion risk cost could be suitable for higher
pressure pipelines of suburban areas, as shown in case 2; and both
could be suitable for suburban areas, as in case 3. In summary,
leakage risk cost is used as the deciding factor for urban areas, and
corrosion risk cost is suitable for suburban areas, especially because
both corrosion risk cost and leakage risk cost can be used as the
deciding factor for the latter when the two benets values are
positive. This study is distinguished by the problem of the new
layout optimization, in which new efcient optimization
J. An, S. Peng / Journal of Natural Gas Science and Engineering 33 (2016) 255e263
263
gas liquefaction process using the particle swarm paradigm with nonlinear
constraints. Energy 49 (1), 146e155.
Mohammad, V., Mohammad, A., Saeid, E., 2015. Surface irrigation simulation
models: a review. Int. J. Hydrol. Sci. Technol. 5 (1), 51e70.
Ossai, C.I., Boswell, B., Davies, I.J., 2015. Pipeline failures in corrosive environments
e a conceptual analysis of trends and effects. Eng. Fail. Anal. 53, 36e58.
Pfetsch, M.E., Fuegenschuh, A., Geissler, B., Geissler, N., Gollmer, R., Hiller, B.,
Humpola, J., Koch, T., Lehmann, T., Martin, A., Morsi, A., Roevekamp, J.,
Schewe, L., Schmidt, M., Schultz, R., Schwarz, R., Schweiger, J., Stangl, C.,
Steinbach, M.C., Vigerske, S., Willert, B.M., 2015. Validation of nominations in
gas network optimization: models, methods, and solutions. Optim. Methods &
Softw. 30 (1), 15e53.
Reitenbach, V., Ganzer, L., Albrecht, D., Hagemann, B., 2015. Inuence of added
hydrogen on underground gas storage: a review of key issues. Environ. Earth
Sci. 73 (11), 6927e6937.
Rios-Mercado, R.Z., Borraz-Sanchez, C., 2015. Optimization problems in natural gas
transportation systems: a state-of-the-art review. Appl. Energy 147, 536e555.
Sanaye, S., Mahmoudimehr, J., 2013. Optimal design of a natural gas transmission
network layout. Chem. Eng. Res. Des. 91 (12), 2465e2476.
Sanchez-Ubeda, E.F., Berzosa, A., 2007. Modeling and forecasting industrial end-use
natural gas consumption. Energy Econ. 29 (4), 710e742.
Schmidt, M., Steinbach, M.C., Willert, B.M., 2015. High detail stationary optimization
models for gas networks. Optim. Eng. 16 (1), 131e164.
Seyyed, M., Mohammad, E., Mohammad, V., 2012. Parameters estimate of autoregressive moving average and autoregressive integrated moving average models
and compare their ability for inow forecasting. J. Math. Stat. 8 (3), 330e338, 01
January.
Stalheim, D., 2011. Metallurgical optimization of microalloyed steels for oil and gas
transmission pipelines. J. Iron Steel Res. Int. 18, 669e674.
Suresh, K., Ghosh, S., Kundu, D., Sen, A., Das, S., Abraham, A., 2008. Inertia-adaptive
particle swarm optimizer for improved global search. In: Eighth International
Conference on Intelligent Systems Design and Applications, vol. 2, pp. 253e258.
lu, S., 2014. Optimization for design and operation of natural gas
Ster, H., Dilaverog
transmission networks. Appl. Energy 133, 56e69.
Valipour, M., 2016. Optimization of neural networks for precipitation analysis in a
humid region to detect drought and wet year alarms. Meteorol. Appl. 23,
91e100.
Valipour, M., Banihabib, M., Behbahani, S., 2013. Comparison of the ARMA, ARIMA,
and the autoregressive articial neural network models in forecasting the
monthly inow of Dez dam reservoir. J. Hydrol. 476, 433e441.
Valipour, M., Montazar, A., 2012. Sensitive analysis of optimized inltration parameters in SWDC model. Adv. Environ. Biol. August, p. 2574(8).
Wu, X., Li, C., Jia, W., He, Y., 2014. Optimal operation of trunk natural gas pipelines
via an inertia-adaptive particle swarm optimization algorithm. J. Nat. Gas Sci.
Eng. 21, 10e18.
Wu, Y., Lai, K.K., Liu, Y., 2007. Deterministic global optimization approach to steadystate distribution gas pipeline networks. Optim. Eng. 8 (3), 259e275.
Zhiping, Z., Leijing, W., Chaohui, Y., Fayuan, W., Xu, T., 2014. The application of factor
analysis for substation soil corrosivity appraisal. Chin. J. Corros. Prot. 147e152
(02).
Zhu, G.Y., Henson, M.A., Megan, L., 2001. Dynamic modeling and linear model
predictive control of gas pipeline networks. J. Process Control 11 (2), 129e148.