# Non-Dominated Sorting Harmony Search Differential Evolution (NS-HS-DE): A Hybrid Algorithm for Multi-Objective Design of Water Distribution Networks

^{1}

^{2}

^{3}

^{*}

## Abstract

**:**

## 1. Introduction

- (1)
- DE and HS show high efficiency in solving numerous single-objective test problems, especially the least-cost design problems involving WDNs. Moreover, since the proposed algorithm is constituted of DE and HS operators, it is advantageous to include source algorithms in the comparison.
- (2)
- (3)
- NSGA2 and SPEA2 are widely used in representative MOEAs in water resource and environmental engineering. More importantly, NSGA2 was reported to be the superior method relative to several contenders.
- (4)
- MOEA/D is a new approach for solving multi-objective optimization problems (Zhang and Li, [26]), which has a specific characteristic based on the decomposition scheme to separate the original problem into several sub-problems that can be solved collaboratively and simultaneously.

## 2. Methods

#### 2.1. Proposed Algorithm, Non-Dominated Sorting Harmony Search Differential Evolution (NSHSDE)

- (I).
- A mutation vector ${V}_{i}=\left\{{v}_{i,1},{v}_{i,1},\dots ,{v}_{i,n}\right\}$ is computed according to the equation:$${V}_{i}={H}_{C1}+F\times \left({H}_{C2}-{H}_{C3}\right)$$
_{C}_{1}, H_{C}_{2}and H_{C}_{3}are three members randomly selected from HM and F is a scaling factor in (0,1], preferably within the range 0.3–0.7. - (II).
- A pitch adjustment is used to enhance the diversity of the perturbed harmony vector and is considered as new harmony:$${r}_{i,j}=\{\begin{array}{ll}{v}_{i,j}\text{\hspace{0.17em}}+\Delta \text{\hspace{0.17em}};& \mathrm{if}\text{\hspace{0.17em}}(ran{d}_{j}\le PAR)\\ {v}_{i,j}\text{\hspace{0.17em}};& \mathrm{if}\text{\hspace{0.17em}}(ran{d}_{j}>PAR)\end{array}$$
_{w}is the fret width (or band width) parameter with a value considered as a small percentage of the range of decision variable j, and N(0,1) is a normal random number with mean 0 and variance 1. r_{i,j}is the decision variable j of the new harmony, R_{i}.

_{1}, F

_{2}, …, F

_{l}according to their non-domination sort order, where l is the index of the last front. Harmonies in front 1 dominate the harmonies in higher fronts; similarly, harmonies in front 2 dominate the harmonies in fronts 3, 4, …, l, but are dominated by those in front 1, and so on.

_{1}, F

_{2}, …, F

_{l}are sorted according to the crowding distance [27].

_{l}is the number of non-dominated fronts in the current iteration.

_{1}) and proceeding with the subsequently ranked non-dominated fronts (F

_{2}, F

_{3}, …, F

_{l}) until the size exceeds the full capacity. It is necessary to reject some of the lower-ranked non-dominated solutions to reduce the total number of the non-dominated solutions to render it equal to the HMS. This is done by using the crowding distance comparison operator; using this procedure, the elements in the HM are updated.

_{min}and Fw

_{max}are the minimum and maximum values of “fret width” respectively, and MaxIt is the number of iterations (generations).

#### 2.2. Algorithms Implemented for Comparison

#### 2.2.1. Non-Dominated Sorting Genetic Algorithm 2 (NSGA2)

#### 2.2.2. Non-Dominated Sorting Harmony Search (NSHS) Algorithm

#### 2.2.3. Non-Dominated Sorting Differential Evolution (NSDE) Algorithm

#### 2.2.4. Improving the Strength Pareto Evolutionary Algorithm (SPEA2)

#### 2.2.5. Multi-Objective Evolutionary Algorithm Based on Decomposition (MOEA/D)

#### 2.3. Performance Measures

#### 2.3.1. Generational Distance (GD)

_{i}is the distance (measured in objective space) between each of these and the nearest member of the global Pareto-optimal set. The parameter P stands for the Pth norm of the distance which is assumed equal 2–i.e. Euclidean distance, in this research. An ideal value of GD = 0 indicates that all elements generated are in the global Pareto-optimal set. Thus, any other value indicates how “far” the generated elements are from the global Pareto front. The lower GD, the algorithm’s performance better in terms of convergence.

#### 2.3.2. Diversity (D)

_{j}is the vector of jth objective function values of the Pareto front solutions and n is the number of objective functions. The higher the value of D metric, the better the diversity of MOEA. There is no specific lower and upper bound for this metric and its value is problem-dependent.

#### 2.3.3. Hypervolume (HV)

_{R}), was used to reduce the bias arising out of the magnitudes of different objective functions and to evaluate the quality of solutions found by each MOEA. HV

_{R}varies between zero and one with the ideal value of one.

#### 2.3.4. Coverage Set (CS)

#### 2.4. Multi-Objective Design of WDNs

#### 2.4.1. Mathematical Formulation

_{i}= unit cost of pipe i of diameter D

_{i}, L

_{i}= length of pipe i, I

_{n}= network resilience, nn = number of demand nodes, C

_{j}, Q

_{j}, H

_{j}, and ${H}_{j}^{req}$ are uniformity, demand, actual head and minimum required head respectively, of node j; nr = number of reservoirs, Q

_{k}and H

_{k}are discharge and actual head respectively, of reservoir k, npu = number of pumps, P

_{i}= power of pump i, γ = specific weight of water, npj = number of pipes connected to node j, and D

_{i}= diameter of pipe i connected to demand node j. The constraints of the optimization problem are as follows:

_{in}and Q

_{out}are in-flow and out-flow of the node, respectively, and Q

_{e}is the external flow rate or demand at the node.

_{k}is the head loss in pipe k and Nl is the total number of loops in the system. The head loss in each pipe is the difference in head between connected nodes, and is a function of discharge, pipe diameter, and roughness coefficient of the pipe. Head loss is usually calculated using empirical equations, such as the Darcy-Weisbach or the Hazen-Williams equation.

_{j}is the pressure head at node j, ${H}_{j}^{l}$ is the minimum required pressure head at node j, ${H}_{j}^{u}$ is the maximum allowed pressure head at node j, and nn is the number of demand nodes in the network.

_{i}is the diameter of pipe i, {A} denotes the set of commercially available pipe diameters, and np is the number of pipes.

_{i}Is velocity in pipe i, ${v}_{i}^{l}$ and ${v}_{i}^{u}$ are the minimum and maximum allowed velocity in pipe i, respectively.

^{6}in this study. Cp

_{1}is the summation of the penalties of all nodes with pressure violation, Cp

_{2}is the summation of the penalties of all pipes with velocity violation and Cp is the total penalty. Therefore, the total cost of the network is the sum of network cost C and penalty cost Cp in nodes and pipes with pressure and velocity violation, respectively.

#### 2.4.2. Experimental Tests on WDNs

## 3. Results and Discussions

## 4. Conclusions

## Acknowledgments

## Author Contributions

## Conflicts of Interest

## References

- Nicklow, J.; Reed, P.; Savic, D.; Dessalegne, T.; Harrell, L.; Chan-Hilton, A.; Zechman, E. State-of-the-art for genetic algorithms and beyond in water resources planning and management. J. Water Resour. Plan. Manag.
**2010**, 136, 412–432. [Google Scholar] [CrossRef] - Keedwell, E.; Khu, S.T. Hybrid Genetic Algorithms For Multiobjective Optimisation of Water Distribution Networks. In Genetic and Evolutionary Computation Gecco; Springer: Berlin/Heidelberg, Germany, 2004. [Google Scholar]
- Montalvo, I.; Izquierdo, J.; Schwarze, S.; Pérez-García, R. Multi-Objective Particle Swarm Optimization Applied to Water Distribution Systems Design: An Approach with Human Interaction. Math. Comput. Model.
**2010**, 52, 1219–1227. [Google Scholar] [CrossRef] - Todini, E. Looped water distribution networks design using a resilience index based heuristic approach. Urban Water
**2000**, 2, 115–122. [Google Scholar] [CrossRef] - Prasad, T.D.; Park, N.-S. Multiobjective Genetic Algorithms for Design of Water Distribution Networks. J. Water Resour. Plan. Manag.
**2004**, 130, 73–82. [Google Scholar] [CrossRef] - Farmani, R.; Savic, D.A.; Walters, G.A. Evolutionary Multi-Objective Optimization in Water Distribution Network Design. Eng. Optim.
**2005**, 37, 167–183. [Google Scholar] [CrossRef] - Farmani, R.; Walters, G.A.; Savic, D.A. Trade-Off between Total Cost and Reliability for Anytown Water Distribution Network. J. Water Resour. Plan. Manag.
**2005**, 131, 161–171. [Google Scholar] [CrossRef] - Perelman, L.; Ostfeld, A.; Salomons, E. Cross-Entropy Multiobjective Optimization for Water Distribution Systems Design. Water Resour. Res.
**2008**, 44, W09413. [Google Scholar] [CrossRef] - Zheng, F.; Simpson, A.R.; Zecchin, A.C. An Efficient Hybrid Approach for Multiobjective Optimization of Water Distribution Systems. J. Water Resour. Res.
**2014**, 50, 3650–3671. [Google Scholar] [CrossRef] - Matos, J.P.; Monteiro, A.J.; Matias, N.M.; Schleiss, A.J. Redesigning Water Distribution Networks Using A Guided Evolutionary Approach. J. Water Resour. Plan. Manag.
**2015**, 142, C4015004. [Google Scholar] [CrossRef] - Bi, W.; Dandy, G.C.; Maier, H.R. Use of Domain Knowledge to Increase the Convergence Rate of Evolutionary Algorithms for Optimizing the Cost and Resilience of Water Distribution Systems. J. Water Resour. Plan. Manag.
**2016**, 142, 04016027. [Google Scholar] [CrossRef] - Geem, Z.W. Particle Swarm Harmony Search for Water Network Design. Eng. Optim.
**2009**, 41, 297–311. [Google Scholar] [CrossRef] - Geem, Z.W.; Kim, J.H.; Loganathan, G.V. A New Heuristic Optimization Algorithm: Harmony Search. Simulation
**2001**, 76, 60–68. [Google Scholar] [CrossRef] - Geem, Z.W.; Cho, Y.H. Optimal Design of Water Distribution Networks Using Parameter-Setting-Free Harmony Search for Two Major Parameters. J. Water Resour. Plan. Manag.
**2011**, 137, 377–380. [Google Scholar] [CrossRef] - Babu, K.J.; Vijayalakshmi, D.P. Self-Adaptive PSO-GA Hybrid Model For Combinatorial Water Distribution Network Design. J. Pipeline Syst. Eng.
**2013**, 4, 57–67. [Google Scholar] [CrossRef] - Cisty, M. Hybrid Genetic Algorithm and Linear Programming Method for Least-Cost Design of Water Distribution Systems. Water Resour. Manag.
**2010**, 24, 24. [Google Scholar] [CrossRef] - Sedki, A.; Ouazar, D. Hybrid Particle Swarm Optimization and Differential Evolution for Optimal Design of Water Distribution Systems. Adv. Eng. Inform.
**2012**, 26, 582–591. [Google Scholar] [CrossRef] - Tolson, B.A.; Asadzadeh, M.; Maier, H.R.; Zecchin, A. Hybrid Discrete Dynamically Dimensioned Search (HD-DDS) Algorithm for Water Distribution System Design Optimization. Water Resour. Res.
**2009**, 45, W12416. [Google Scholar] [CrossRef] - Vrugt, J.A.; Robinson, B.A. Improved Evolutionary Optimization from Genetically Adaptive Multimethod Search. Proc. Natl. Acad. Sci. USA
**2007**, 104, 708–711. [Google Scholar] [CrossRef] [PubMed] - Raad, D.; Sinske, A.; van Vuuren, J. Robust Multi-Objective Optimization for Water Distribution System Design using a Meta-Metaheuristic. Int. Trans. Oper. Res.
**2009**, 16, 595–626. [Google Scholar] [CrossRef] - Wang, Q.; Guidolin, M.; Savic, D.; Kapelan, Z. Two-objective Design of Benchmark Problems of a Water Distribution System via MOEAs: Towards the Best-known Approximation of the True Pareto Front. J. Water Resour. Plan. Manag.
**2014**, 141, 04014060. [Google Scholar] [CrossRef] - Moosavian, N.; Lence, B.J. Nondominated Sorting Differential Evolution Algorithms for Multiobjective Optimization of Water Distribution Systems. J. Water Resour. Plan. Manag.
**2016**, 143, 04016082. [Google Scholar] [CrossRef] - Yazdi, J.; Yoo, D.G.; Kim, J.H. Comparative study of multi-objective evolutionary algorithms for hydraulic rehabilitation of urban drainage networks. Urban Water J.
**2017**, 14, 483–492. [Google Scholar] [CrossRef] - Zitzler, E.; Laumanns, M.; Thiele, L. SPEA2: Improving the Strength of Pareto Evolutionary Algorithm—Evolutionary Methods for Design, Optimisation and Control. In Proceedings of the 2001 International Center for Numerical Methods in Engineering (EUROGEN2001), Athens, Greece, 19–21 September 2001. [Google Scholar]
- Coello, C.A.; Toscano Pulido, G.; Salazar Lechuga, M. Handling Multiple Objectives with Particle Swarm Optimization. IEEE Trans. Evol. Comput.
**2004**, 8, 256–278. [Google Scholar] [CrossRef] - Zhang, Q.; Li, H. Multiobjective Evolutionary Algorithm based on Decomposition. IEEE Trans. Evol. Comput.
**2007**, 11, 712–731. [Google Scholar] [CrossRef] - Yazdi, J.; Sadollah, A.; Lee, E.H.; Yoo, D.G.; Kim, J.H. Application of Multi-objective Evolutionary Algorithms for Rehabilitation of Storm Sewer Pipe Networks. J. Flood Risk Manag.
**2015**. [Google Scholar] [CrossRef] - Mahdavi, M.; Fesanghary, M.; Damangir, E. An Improved Harmony Search Algorithm for Solving Optimization Problems. Appl. Math. Comput.
**2007**, 188, 1567–1579. [Google Scholar] - Deb, K.; Agrawal, S.; Pratap, A.; Meyarivan, T. A Fast Elitist Non-dominated Sorting Genetic Algorithm for Multi-Objective Optimization: NSGA II. In Proceedings of the International Conference on Parallel Problem Solving From Nature, Paris, France, 18–20 September 2000; Springer: Berlin/Heidelberg, Germany; pp. 849–858. [Google Scholar]
- Kougias, I.P.; Theodossiou, N.P. Multiobjective Pump Scheduling Optimization using Harmony Search Algorithm (HSA) and Polyphonic HSA. Water Resour. Manag.
**2013**, 275, 1249–1261. [Google Scholar] [CrossRef] - Storn, R.; Price, K. Differential Evolution: A Simple and Efficient Adaptive Scheme for Global Optimization over Continuous Space; International Computer Science Institute: Berkeley, CA, USA, 1995. [Google Scholar]
- Suribabu, C.R. Differential Evolution Algorithm for Optimal Design of Water Distribution Networks. J. Hydroinform.
**2010**, 12, 66–82. [Google Scholar] [CrossRef] - Zheng, F.; Zecchin, A.C.; Simpson, A.F. Self-adaptive Differential Evolution Algorithm Applied to Water Distribution System Optimization. J. Comput. Civ. Eng.
**2013**, 27, 148–158. [Google Scholar] [CrossRef] - Yazdi, J. Decomposition based multi objective evolutionary algorithms for Design of Large-Scale Water Distribution Networks. Water Resour. Manag.
**2016**, 30, 2749–2766. [Google Scholar] [CrossRef] - Van Veldhuizen, D.A. Multiobjective Evolutionary Algorithms: Classifications, Analyses, and New Innovations; Air Force Institute of Technology: Wright-Patterson AFB, OH, USA, 1999. [Google Scholar]
- Zitzler, E.; Deb, K.; Thiele, L. Comparison of Multi-Objective Evolutionary Algorithms: Empirical Results. Evol. Comput.
**2000**, 8, 173–195. [Google Scholar] [CrossRef] [PubMed] - Rossman, L.A. EPANET 2 Users Manual; U.S. Environment Protection Agency: Cincinnati, OH, USA, 2000.

**Figure 1.**Layout of two-loop network, TLN [21].

**Figure 2.**Layout of Hanoi network (HAN) [20].

**Figure 3.**Layout of Fossolo network (FOS) [21].

**Figure 4.**Layout of Balerma network (BIN) [21].

**Figure 5.**Pareto front (PF) of benchmark problems, (

**a**) Two-loop network (TLN) problem; (

**b**) Hanoi network (HAN) problem.

**Figure 6.**Pareto front (PF) of benchmark problems, (

**a**) Fossolo network (FOS) problem; (

**b**) Balerma network (BIN) problem.

**Figure 7.**A close-up view of Pareto fronts (PFs) generated by Non-Dominated Sorting Harmony Search (NSHS) and nondominated sorting genetic algorithm 2 (NSGA2), (

**a**) FOS problem; (

**b**) BIN problem.

Diameter (in.) | Unit Cost ($/m) | Diameter (in.) | Unit Cost ($/m) |
---|---|---|---|

1 | 2 | 12 | 50 |

2 | 5 | 14 | 60 |

3 | 8 | 16 | 90 |

4 | 11 | 18 | 130 |

6 | 16 | 20 | 170 |

8 | 23 | 22 | 300 |

10 | 32 | 24 | 550 |

Diameter (in.) | Unit Cost ($/m) | Diameter (in.) | Unit Cost ($/m) | Diameter (in.) | Unit Cost ($/m) |
---|---|---|---|---|---|

12 | 45.73 | 20 | 98.39 | 30 | 180.75 |

16 | 70.40 | 24 | 129.33 | 40 | 278.28 |

NI | Pmax (m) | NI | Pmax (m) | NI | Pmax (m) | NI | Pmax (m) | NI | Pmax (m) | NI | Pmax (m) |
---|---|---|---|---|---|---|---|---|---|---|---|

1 | 55.85 | 7 | 53.1 | 13 | 59.1 | 19 | 58.1 | 25 | 56.6 | 31 | 56.6 |

2 | 56.6 | 8 | 54.5 | 14 | 58.4 | 20 | 58.17 | 26 | 57.6 | 32 | 56.8 |

3 | 57.65 | 9 | 55.0 | 15 | 57.5 | 21 | 58.2 | 27 | 57.1 | 33 | 56.4 |

4 | 58.5 | 10 | 56.83 | 16 | 56.7 | 22 | 57.1 | 28 | 55.35 | 34 | 56.3 |

5 | 59.76 | 11 | 57.3 | 17 | 55.5 | 23 | 56.8 | 29 | 56.5 | 35 | 55.57 |

6 | 55.60 | 12 | 58.36 | 18 | 56.9 | 24 | 53.5 | 30 | 56.9 | 36 | 55.1 |

Diameter (mm) | Unit Cost (€/m) | Diameter (mm) | Unit Cost (€/m) | Diameter (mm) | Unit Cost (€/m) | Diameter (mm) | Unit Cost (€/m) |
---|---|---|---|---|---|---|---|

16 | 0.38 | 61.4 | 4.44 | 147.20 | 24.78 | 290.6 | 99.58 |

20.4 | 0.56 | 73.6 | 6.45 | 163.6 | 30.55 | 327.4 | 126.48 |

26 | 0.88 | 90 | 9.59 | 184.00 | 38.71 | 368.2 | 160.29 |

32.6 | 1.35 | 102.2 | 11.98 | 204.6 | 47.63 | 409.2 | 197.71 |

40.8 | 2.02 | 114.6 | 14.93 | 229.2 | 59.7 | ||

51.4 | 3.21 | 130.80 | 19.61 | 257.8 | 75.61 |

Diameter (mm) | Unit Cost (€/m) | Diameter (mm) | Unit Cost (€/m) |
---|---|---|---|

113 | 7.22 | 226.2 | 28.6 |

126.6 | 9.10 | 285 | 45.39 |

144.6 | 11.92 | 361.8 | 76.32 |

162.8 | 14.84 | 452.2 | 124.64 |

180.8 | 18.38 | 581.8 | 215.85 |

Problem | NFE ^{a} | Population Size | DV ^{b} | PD ^{c} | Search Space Size |
---|---|---|---|---|---|

Two-loop Network | 20,000 | 40 | 8 | 14 | 1.48 × 10^{9} |

Hanoi Network | 50,000 | 60 | 34 | 6 | 2.87 × 10^{26} |

Fossolo Network | 200,000 | 100 | 58 | 22 | 7.25 × 10^{77} |

Balerma Irrigation Network | 1,000,000 | 400 | 454 | 10 | 1.0 × 10^{454} |

^{a}NFE = number of function evaluations,

^{b}DV = number of decision variables,

^{c}PD = number of pipe diameter options.

Algorithm | Parameter | Value |
---|---|---|

NSGA2 | Mutation rate | 1/(no. variables) |

Crossover prob. | 0.9 | |

Tournament size | 2 | |

Mutation step size | 0.1 × Variable range | |

SPEA2 | Mutation rate | 1/(no. variables) |

Crossover prob. | 0.9 | |

Tournament size | 2 | |

Mutation step size | 0.1 × Variable range | |

NSHS | HMCR | 0.98 |

PAR | 0.4 | |

Fw | (0.05–0.005) × Variable range | |

NSDE | F (scaling factor) | 0.5 |

Crossover prob. | 0.7 | |

NSHSDE | F (scaling factor) | 0.5 |

PAR | 0.4 | |

Fw | (0.05–0.005) × Variable range | |

MOEA/D | Mutation prob. | 0.3 |

Mutation rate. | 0.1 | |

T (number of neighbors) | 0.2 × Pop size | |

Z * (Goal point) | The ideal values for objective functions, zero for the cost and one for the resiliency index |

**Table 8.**Comparison of multi-objective algorithms using the average values of generational distance (GD), relative diversity (D), and hypervolume (HV) metrics.

Problem | Algorithm | GD | D | HV_{R} |
---|---|---|---|---|

TLN | NSHSDE | 193.96 | 0.79 | 1.00 |

NSDE | 1317.76 | 0.50 | 0.62 | |

NSGA2 | 3690.13 | 0.43 | 0.49 | |

NSHS | 3362.55 | 0.19 | 0.26 | |

SPEA2 | 3136.40 | 0.51 | 0.72 | |

MOEA/D | 4007.28 | 0.41 | 0.49 | |

HAN | NSHSDE | 1992.61 | 0.80 | 0.98 |

NSDE | 2420.65 | 0.66 | 0.96 | |

NSGA2 | 1987.79 | 0.22 | 0.30 | |

NSHS | 1670.78 | 0.17 | 0.24 | |

SPEA2 | 1866.20 | 0.12 | 0.18 | |

MOEA/D | 5130.38 | 0.81 | 0.86 | |

FOS | NSHSDE | 2253.75 | 0.5 | 0.96 |

NSDE | 10529.41 | 0.67 | 0.98 | |

NSGA2 | 213.13 | 0.02 | 0.03 | |

NSHS | 103.00 | 0.00 | 0.01 | |

SPEA2 | 310.30 | 0.00 | 0.00 | |

MOEA/D | 36463.37 | 0.33 | 0.37 | |

BIN | NSHSDE | 7032.75 | 0.32 | 0.36 |

NSDE | 28295.29 | 0.48 | 0.69 | |

NSGA2 | 2144.96 | 0.06 | 0.06 | |

NSHS | 1697.33 | 0.03 | 0.03 | |

SPEA2 | 1762.07 | 0.01 | 0.01 | |

MOEA/D | 1729.88 | 0.09 | 0.08 |

Problem | Algorithm | NSHSDE | NSDE | NSGA2 | NSHS | SPEA2 | MOEA/D |
---|---|---|---|---|---|---|---|

TLN | NSHSDE | - | 0.87 | 0.85 | 0.92 | 0.82 | 0.94 |

NSDE | 0.84 | - | 0.84 | 0.93 | 0.85 | 0.95 | |

NSGA2 | 0.74 | 0.76 | - | 0.89 | 0.78 | 0.86 | |

NSHS | 0.31 | 0.32 | 0.35 | - | 0.32 | 0.37 | |

SPEA2 | 0.66 | 0.67 | 0.7 | 0.83 | - | 0.85 | |

MOEA/D | 0.58 | 0.61 | 0.62 | 0.87 | 0.63 | - | |

HAN | NSHSDE | - | 0.6 | 0.72 | 0.53 | 0.99 | 0.95 |

NSDE | 0.54 | - | 0.71 | 0.51 | 0.99 | 0.93 | |

NSGA2 | 0.28 | 0.29 | - | 0.34 | 0.92 | 0.78 | |

NSHS | 0.47 | 0.49 | 0.66 | - | 0.92 | 0.85 | |

SPEA2 | 0.01 | 0.01 | 0.08 | 0.08 | - | 0.48 | |

MOEA/D | 0.06 | 0.07 | 0.22 | 0.15 | 0.52 | - | |

FOS | NSHSDE | - | 0.82 | 0.93 | 0.75 | 1 | 1 |

NSDE | 0.21 | - | 0.98 | 0.55 | 0.83 | 1 | |

NSGA2 | 0.07 | 0.02 | - | 0.17 | 0.44 | 0.71 | |

NSHS | 0.25 | 0.45 | 0.83 | - | 0.17 | 0.67 | |

SPEA2 | 0 | 0.17 | 0.566 | 0.83 | - | 0.92 | |

MOEA/D | 0 | 0 | 0.29 | 0.33 | 0.08 | - | |

BIN | NSHSDE | - | 0.59 | 1 | 0.64 | 1 | 0.53 |

NSDE | 0.41 | - | 1 | 0.62 | 0.55 | 0.38 | |

NSGA2 | 0 | 0 | - | 0.48 | 0.4 | 0 | |

NSHS | 0.36 | 0.38 | 0.52 | - | 1 | 0.46 | |

SPEA2 | 0 | 0.45 | 0.59 | 0 | - | 0 | |

MOEA/D | 0.48 | 0.61 | 1 | 0.54 | 1 | - |

© 2017 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## Share and Cite

**MDPI and ACS Style**

Yazdi, J.; Choi, Y.H.; Kim, J.H. Non-Dominated Sorting Harmony Search Differential Evolution (NS-HS-DE): A Hybrid Algorithm for Multi-Objective Design of Water Distribution Networks. *Water* **2017**, *9*, 587.
https://doi.org/10.3390/w9080587

**AMA Style**

Yazdi J, Choi YH, Kim JH. Non-Dominated Sorting Harmony Search Differential Evolution (NS-HS-DE): A Hybrid Algorithm for Multi-Objective Design of Water Distribution Networks. *Water*. 2017; 9(8):587.
https://doi.org/10.3390/w9080587

**Chicago/Turabian Style**

Yazdi, Jafar, Young Hwan Choi, and Joong Hoon Kim. 2017. "Non-Dominated Sorting Harmony Search Differential Evolution (NS-HS-DE): A Hybrid Algorithm for Multi-Objective Design of Water Distribution Networks" *Water* 9, no. 8: 587.
https://doi.org/10.3390/w9080587