Sei sulla pagina 1di 3

Discussion of Optimization of Water Distribution Networks Using Integer and Linear Programming by Hossein M. V.

Samani and Alireza Mottaghi


May 2006, Vol. 132, No. 5, pp. 501509.

DOI: 10.1061/ ASCE 0733-9429 2006 132:5 501

Dragan Savic1 and Maria da Conceio Cunha2


1

Professor, Univ. of Exeter, Centre for Water Systems, Exeter, Devon EX4 4QJ U.K. E-mail: D.Savic@exeter.ac.uk 2 Professor, Civil Engineering Dept., Univ. of Coimbra, Polo II, 3030-290 Coimbra, Portugal. E-mail: mccunha@dec.uc.pt

The authors are commended for tackling the practical and difcult problem of optimal design of water distribution networks WDN . Abundant evidence of interest in this topic can be found in the research literature because WDN are an important element of urban infrastructure, whose construction, operation, and maintenance costs are often immense. However, so far no exact optimization method has been found for this type of mathematical problem. Therefore, it is important that appropriate tools are developed and used to help arrive at the best possible decisions regarding planning, design, and operation of WDN. The authors make a case for the use of the integer linear programming ILP technique to solve a design problem, but by applying it to several small problems only, they do not convincingly demonstrate that a signicant improvement to already available and proven techniques has been made. In fact, the proposed approach may be viewed as a step backward, both in terms of research and practical application of optimization to WDN design. Classical optimization methods, like linear and nonlinear programming, were tried in the past, usually at the cost of drastic simplications to the problem. The challenge in solving this problem, which has been proved to be an NP-hard nonlinear mixed integer optimization problem, has been a reason for a multitude of contributions to the literature in the last twenty-ve years. The complexity associated with 1 the treatment of discrete variables and extremely large search spaces; 2 the need to take into account multiple loading conditions, water quality, and extended period simulation; 3 the need for methodologies able to cope with conicting objectives, reliability, risk, and uncertainty associated with WDN models; and 4 complex behavior of valves, pumps, and other network elements, have caused problems to many optimization techniques. The proposed iterative methodology based on ILP does not appear to have improved upon the current state of the art in the area. The authors acknowledge that when the hydraulic analysis is performed with poorly selected variables, even on very small examples, the methodology fails to converge to a feasible solution. A more detailed analysis of the literature shows the reasons for this behavior and provides impetus for further research in this eld. When considering pipe diameters as the main decision variables three approaches can be found in the literature: the split pipe, the continuous, and the discrete diameter approaches. Alperovits and Shamir 1977 published one of the earliest attempts, which was followed a few years later by other authors Quindry et al. 1981; Fujiwara et al. 1987; Kessler and Shamir 1989; Fujiwara and Khang 1990 . They used the split pipe approach in what has become known as the linear programming gradient method. The solutions provided by this method have some signicant drawbacks, mainly that convergence was not always guaranteed and the algorithm would often end up with an impracticable solution.
1024 / JOURNAL OF HYDRAULIC ENGINEERING ASCE / JULY 2008

For example, a solution could have a number of pipes consisting almost entirely of one diameter with the remaining short length of the same pipes having a different diameter. In the second type of approach Fujiwara and Khang 1990 and 1991; Varma et al. 1997 , the diameter is taken as a continuous variable and nonlinear programming can be used to solve the problem. The solution provided by the continuous diameter approach is even less practical since in reality diameters must be chosen from a set of commercially available discrete value diameters. The conversion of continuous solutions into discrete diameters can worsen the quality of the solution and even induce a loss of feasibility. All this means that diameters must be treated as discrete variables. Early efforts at this treatment were described by Gessler 1981, 1982, 1985 , who used an explicit enumeration methodology to explore all possible diameter combinations. The problem with this approach is that for large-scale problems the method becomes inefcient. Methods capable of handling nonlinear mixed integer models keeping all the realistic features , searching for global optimality, have been developed only recently. Among these are genetic algorithms Goldberg and Kuo 1987; Hadji and Murphy 1990; Murphy et al. 1993; Walters and Cembrowicz 1993; Simpson et al. 1994; Savic and Walters 1997; Walters et al. 1999 , simulated annealing Cunha and Sousa 1999, 2001 , tabu search Cunha and Ribeiro 2004 , harmony search Geem et al. 2001; Geem 2005, 2006 , shufed frog leaping algorithms Eusuff and Lansey 2003 , and ant colony algorithms Zecchin et al. 2006 . The use of such algorithms represents a signicant step forward for solving the water distribution system optimization problems, as it shown by the uptake of these methods by commercial software vendors. It has been demonstrated even on large, real problems that, if appropriately calibrated, they work well Savic et al. 2000; Farmani et al. 2005; Optimatics 2006 . Uncertainty and risk issues demanding more resilient and robust solutions are new research challenges that have been tackled recently Tolson et al. 2004; Babayan et al. 2005, 2006; Kapelan et al. 2005; Tricarico et al. 2006 . It is in the light of all these challenges and past developments that we see the contribution of this paper. For a number of reasons the paper does not appear to advance the state of the art in the eld. The test networks used to demonstrate the effectiveness of the methodology are very small and appear almost trivial in comparison to a benchmark Anytown network set in 1987 by Walski et al. 1987 . Even that network is not large by todays water industry standards, where all-mains models are developed that consist of tens of thousands of elements. The proof of convergence is missing and even for the simplied, single-loading cases presented in the paper, the authors state, When the hydraulic analysis is performed with poorly selected variables, computed discharges in the pipes will be so far from the nal solution that no feasible solution can be found through the optimization process. It is clear that if a large, more realistic problem, i.e., one with a number of complex valves and/or pumps, is attempted using ILP many simplications will be required. Even then the convergence rate will be affected adversely, if a solution can be achieved at all. The claim that the main advantage of the ILP methodology is the use of commercially available pipe sizes would have been welcomed 15 years ago, but since then many soft computing techniques have been able to deal with commercially available diameters and produce veriable near-optimal solutions. Finally, the state of the art in WDN optimal design has moved on to incorporate multiple objectives and risk/reliability/ redundancy measures in the optimization procedure Halhal et al.

1997 . It is difcult to see how ILP could be extended to deal with these issues effectively and efciently.

References
Alperovits, E., and Shamir, U. 1977 . Design of optimal water distribution systems. Water Resour. Res., 13 6 , 885900. Babayan, A., Kapelan, Z., Savic, D., and Walters, G. 2005 . Least cost design of water distribution networks under demand uncertainty. J. Water Resour. Plann. Manage., 131 5 , 375382. Babayan, A. V., Kapelan, Z., Savic, D. A., and Walters, G. A. 2006 . Comparison of two approaches for the least cost rehabilitation of water distribution systems under uncertainty. Eng. Optimiz., 38 3 , 281297. Cunha, M. C., and Ribeiro, L. 2004 . Tabu search algorithms for water network optimization. Eur. J. Oper. Res., 157, 746758. Cunha, M. C., and Sousa, J. 1999 . Water distribution network design optimization: Simulated annealing approach. J. Water Resour. Plann. Manage., 125 4 , 215221. Cunha, M. C., and Sousa, J. 2001 . Hydraulic infrastructures design using simulated annealing. J. Infrastruct. Syst., 7 1 , 3239. Eusuff, E. E., and Lansey, K. E. 2003 . Optimization of water distribution network design using the shufed frog leaping algorithm. J. Water Resour. Plann. Manage., 129 3 , 210225. Farmani, R., Savic, D. A., and Walters, G. A. 2005 . Evolutionary multi-objective optimization in water distribution network design. Eng. Optimiz., 37 2 , 167183. Fujiwara, O., and Khang, D. B. 1990 . A two-phase decomposition method for optimal design of looped water distribution networks. Water Resour. Res., 26 4 , 539549. Fujiwara, O., and Khang, D. B. 1991 . Correction to A two-phase decomposition method for optimal design of looped water distribution networks. Water Resour. Res., 27 5 , 985986. Fujiwara, O., Jenchaimahakoon, B., and Edirisinghe, N. C. P. 1987 . A modied linear programming gradient method for optimal design of looped water distribution networks. Water Resour. Res., 23 6 , 977 982. Geem, Z. W. 2005 . Harmony search in water pump switching problem. Lect. Notes Comput. Sci., 3612, 751760. Geem, Z. W. 2006 . Optimal cost design of water distribution networks using harmony search. Eng. Optimiz., 38 3 , 259280. Geem, Z. W., Kim, J. H., and Loganathan, G. V. 2001 . A new heuristic optimization algorithm: Harmony search. Simulation, 76 2 , 6068. Gessler, J. 1981 . Analysis of pipe networks. Closed-conduit ow, M. H. Chaudhry and V. Yevjevich, eds., Water Resources Publications, 6169. Gessler, J. 1982 . Optimization of pipe networks. Proc. Inst. on Urban Hydr. and Sediment Control, Univ. of Kentucky, Lexington, Ky., 165 171. Gessler, J. 1985 . Pipe network optimization by enumeration. Computer Applications in Water Resources, H. Torno, ed., ASCE Water Resources Planning and Management Division Conference, Buffalo, N.Y.

Goldberg, D. E., and Kuo, C. H. 1987 . Genetic algorithms in pipeline optimization. J. Comput. Civ. Eng., 1 2 , 128141. Hadji, G., and Murphy, L. J. 1990 . Genetic algorithms for pipe network optimization. 4th Year Student Civil Engineering Research Rep., Univ. of Adelaide, Australia. Halhal, D., Walters, G. A., Ouazar, D., and Savic, D. A. 1997 . Water network rehabilitation with structured messy genetic algorithm. J. Water Resour. Plann. Manage., 123 3 , 137146. Kapelan, Z., Savic, D. A., Walters, G. A., and Babayan, A. V. 2005 . Risk and robustness based solutions to a multiobjective water distribution system rehabilitation problem under uncertainty. Water Sci. Technol., 53 1 , 6175. Kessler, A., and Shamir, U. 1989 . Analysis of the linear programming gradient method for optimal design of water supply networks. Water Resour. Res., 25 7 , 14691480. Murphy, L. J., Simpson, A. R., and Dandy, G. C. 1993 . Design of a network using genetic algorithms. Water, 20, 4042. Optimatics. 2006 . http://www.optimatics.com/projects.htm Oct. 29, 2006 . Quindry, G., Brill, E., and Liebman, J. 1981 . Optimization of looped water distribution systems. J. Envir. Engrg. Div., 107 4 , 665679. Savic, D. A., and Walters, G. A. 1997 . Genetic algorithms for leastcost design of water distribution networks. J. Water Resour. Plann. Manage., 123 2 , 6777. Savic, D. A., and Walters, G. A., Randall-Smith, M., and Atkinson, R. M. 2000 . Large water distribution systems design through genetic algorithm optimisation. Joint Conf. on Water Resources Engineering and Water Resources Planning and Management, ASCE, Minneapolis. Simpson, A. R., Dandy, G. C., and Murphy, L. J. 1994 . Genetic algorithms compared to other techniques for pipe optimization. J. Water Resour. Plann. Manage., 120 4 , 423443. Tolson, B. A., Maier, H. R., Simpson, A. R., and Lence, B. J. 2004 . Genetic algorithms for reliability-based optimization of water distribution systems. J. Water Resour. Plann. Manage., 130 1 , 6372. Tricarico, C., Gargano, R., Kapelan, Z., Savic, D., and de Marinis, G. 2006 . Economic level of reliability for the rehabilitation of hydraulic networks. Civ. Eng. Environ. Syst., 23 3 , 191207. Varma, K. V. K., Narasimhan, S., and Bhallamudi, S. M. 1997 . Optimal design of water distribution systems using an NLP method. J. Environ. Eng., 123 4 , 381388. Walski, T. M., et al. 1987 . Battle of the network models: Epilogue. J. Water Resour. Plann. Manage., 113 2 , 191203. Walters, G. A., and Cembrowicz, R. G. 1993 . Optimal design of water distribution networks. Water supply systems, state of the art and future trends, E. Cabrera and F. Martinez, eds., Computational Mechanics Publications, Southampton, 91117. Walters, G. A., Halhal, D., Savic, D., and Quazar, D. 1999 . Improved design of Anytown distribution network using structured messy genetic algorithm. Urban Water, 1 1 , 2328. Zecchin, A. C., et al. 2006 . Application of two ant colony optimisation algorithms to water distribution system optimization. Math. Comput. Modell., 44 5-6 , 451468.

JOURNAL OF HYDRAULIC ENGINEERING ASCE / JULY 2008 / 1025

Potrebbero piacerti anche