A methodology for optimal design of pipe distribution networks

Abstract

A series of computer programs has been developed to assist in the optimal design of pipe distribution networks. These programs are capable of handling nonstandard network components such as booster pumps, minor-loss devices, reservoirs, check valves, and pressure-reducing valves. All three stages of network problem, namely analysis, design, and optimization, can be solved using the same solution procedure. A known technique for layout design has been adapted to help in the selection of redundant links.The problem is formulated as a nonlinear program using as design variables the diameter, discharge, and shutoff head (where appropriate) for each link. Pump lift is assumcd to be described by a parabolic Q-h,, curve, the coefficients of which may be defined by the user. A powerful, large-scale, nonlinear, sparse-oriented package (MLNOS) is used to solve the problem. The continuous solution obtained by MINOS is modified by a discretization procedure to arrive at the optimal discrete solution. The constraint matrix including the loop, nodal head, and continuity constraints is automatically generated.The user is required to supply a simple and concise data file that will be interfaced by a preprocessor to generate the large and complicated data file required by MINOS. The latter data file includes the nonzero elements of the constraint matrix, ordered and stored in a column-by-column fashion. A postprocessor is also used to convert the mathematical output of the package into easily understood engineering data. Three different network examples are used to demonstrate the different aspects of the model (i.e. nonstandard components, discretization procedure, and layout design).Une sCrie de logiciels a Ct C mise au point afin de faciliter la conception optimale des rCseaux de distribution des tuyaux. Ces logiciels sont en mesure de tenir compte de composantes spCciales tclles quc les pompes de surpression, les accessoires de tuyauterie, les rtservoirs, les clapets de non-rctour et les rCducteurs de pression. Les trois types dc problkmes du rCseau, soit I'analyse, la conception et I'optimisation, pcuvent etrc rCsolus dc la m&me fagon. Une techniquc connue de conception de rCseaux a CtC adaptCe pour aider a la s~lection des sections inutilcs.Le problkme est CnoncC sous forme de logicicl non IinCaire utilisant commc variables de conception le diamktre des tuyaux, le dCbit et la charge d'arrkt (s'il y a lieu) pour chaque section. On assume que Ic rcfoulcment de la pompe est dCcrit par une courbe parabolique Q-h,,, dont Ics coefficients pcuvent Ctre dCfinis par I'utilisateur. Un progiciel non IinCaire de grande puissance (MINOS) est utilisC pour rCsoudre Ic problkme. La solution continue obtenue a I'aide de MLNOS est modifiCe par la rnCthode de discrktisation pour en arriver a la solution discrkte optimalc. La matrice des contraintes est produite autornatiquement.L'utilisateur doit foumir un fichier de donnCes simpleet concis qui sera intcrfact par un prCprocesseur pour produire la fichier cornplexe et ClaborC requis par MINOS....

Genetic Algorithms Compared to Other Techniques for Pipe Optimization

“…Researchers have reported a number of applications of nonlinear optimization to pipe network problems (EI-Bahrawy andSmith 1985, 1987;Su et al 1987;Lansey and Mays 1989a;Lansey et al 1989;Duan et al 1990). EI-Bahrawy and Smith (1985) applied MINOS to the design of water collection and distribution systems.…”

Reliability based optimal design of water distribution networks considering life cycle components

“…Alperovits and Shamir (1977) proposed a linear programming gradient method (LPG) for optimisation of a looped water distribution system, which has been improved upon by several researchers to attain a global maximum (Quindry et al 1981, Featherstone and El-Jumaily 1983, Fujiwara and Khang 1987. Due to the nonlinear nature of the distribution system infrastructure, numerous researchers have also applied non-linear programming techniques to optimise the water distribution network problems (Shamir 1974, El-Bahrawy and Smith 1987, Su et al 1987, Lansey and Mays 1989, Duan et al 1990). The non-linear optimisation procedure does not necessarily yield a global optimum and the final solution depends on the initial input for the solution.…”

Robust multi‐objective optimization for water distribution system design using a meta‐metaheuristic

“…Several researchers have applied nonlinear optimization techniques to WDS design problems, including Shamir (1974), Ormsbee and Contractor (1981), El‐Bahrawy and Smith (1987), Chiplunkar et al (1986), Su et al (1987), Lansey and Mays (1989), Duan et al (1990), Cullinane et al (1992), and Loganathan et al (1995). The limitations of nonlinear optimization techniques are that they frequently become trapped at local optima and that they typically employ continuous variables.…”