Software Development for the Water Sector

The water sector in the last 20 years has undergone radical paradigm shifts arising from the crisis of global proportions that have characterized the sector, prompting many international fora, including the Dublin conference in January 1992. One of the responses from academic institutions to this crisis is the development of computer-based predictive tools for better and more accurate prediction of the variables that affect water use and management. In the School of Civil and Environmental Engineering at the University of the Witwatersrand, attempts have been made to develop software to aid planning, management, and decision making in the water sector. Two of such software are Wadessy a water distribution network design software, and a groundwater flow modelling software GEMFLOW that is based on the Green element method (GEM). Although their engines are quite robust and have been applied in field studies in Botswana and Zimbabwe, and compare favourably with published models, their elegance in terms of graphical user interface (GUI) is still rudimentary. The cost for their development has been mainly in the training of postgraduate students who have assisted in their development. Industry uptake has been very limited, which is one of the reasons why their GUIs are still rudimentary. With greater investment into the development and marketing of these and many other software, the potential exists to have “made-in-Africa” software with capabilities comparable, if not better than, those developed in more advanced countries. This paper reports on these software, compares these with similar initiatives in more advanced countries, and discusses the challenges in development, funding, and uptake by industry. The experiences described herein are most likely to be similar with other software development initiatives in subSaharan Africa. Introduction The design of water distribution systems (WDSs) has received a great deal of attention because of its importance to industrial growth and water's crucial role in society for health, fire-fighting, and quality of life, particularly in light of increased urban development and water use [1]. Water reticulation networks (WRNs) are essential components of all WDSs as they convey potable water from the source, pump station or storage to the consumers. The cost of these networks may amount to as much as 60% of the entire WDS [2] and as a result, operation and maintenance costs may soar higher if WRNs are ill designed [3]. WRNs also account for the largest costs in municipal maintenance budgets [1]. Despite often scarce resources, all governments are obligated to provide this resource. Since WRNs are composed mostly of pipes, pipe-sizing decisions have become critical in designing cost effective WDSs that are capable of handling varied demand loadings and satisfying minimum pressure head requirements [3]. Optimisation therefore provides an appropriate tool for achieving the best WRN designs. To arrive at an optimal solution, an iterative simulation-optimisation algorithm is employed. Efficient hydraulic simulation (both static and dynamic) is based on modelling the continuity, conservation of energy and pressure head difference equations and determining the unknown variables using the Newton-Raphson iterative procedure on simultaneous equations generated using the nodal method. The Choleski Decomposition technique is employed to generate the matrix for computing node residual pressure heads. Based on either the Darcy-Weisbach or Hazen-Williams pipe friction equations, continuity is checked at each network node and if a violation exists, the Advanced Materials Research Vols. 18-19 (2007) pp 543-548 online at http://www.scientific.net © (2007) Trans Tech Publications, Switzerland Online available since 2007/Jun/15 All rights reserved. No part of contents of this paper may be reproduced or transmitted in any form or by any means without the written permission of the publisher: Trans Tech Publications Ltd, Switzerland, www.ttp.net. (ID: 130.203.133.33-17/04/08,14:22:01) entire simulation process is repeated. Output from the simulation include pipe flows and orientation, pipe head losses, friction factors, node residual pressure heads, draw-off at each source node, pumping heads and valve head losses. Solution optimization is accomplished by Genetic Algorithms (GAs) which are adaptive search algorithms premised on the evolutionary ideas of natural selection and genetics. They are amenable to especially problems in small and large systems that employ multiple variables, are stochastic in nature and operate under varying loading conditions such as is found in WRNs. Not only does GAs provide an effective method to solving such problems, they consistently out-perform other traditional methods. GAs proceed from the theoretical principle of implicit parallelism. This principle enables highly fit members of a population to receive increased numbers of offspring in successive generations and thus leads to fitter generations – more like the survival of the fittest. The GA design tool presented in this paper (hereafter referred to as Wadessy – an acronym for Water Decision Support Systems) is employed to optimise WRN pipe sizes. For each pipe link, Wadessy recommends, at the most, two pipe segments which may have different pipe sizes and lengths. The individual lengths of the two segments must however sum up to the total length prescribed for that pipe link. The objective function for each Wadessy optimisation run is as summarised below: Minimise CWRN = Minimise(CostWRN + Costpressure violation at each WRN node, n) (1) where Costpressure violation at each WRN node,n = ∑ = n i 1 (Pressure violation x node demand) While Wadessy’s modelling capability is with WRNs, GEMFLOW is a simulation model for groundwater resource evaluation. To achieve judicious management of a groundwater system requires not a good understanding of its hydrogeology but also suitable predictive model that captures the response of the aquifer to various recharge and pumping stresses that are as a result of various human activities. GEMFLOW is designed to serve as a predictive tool for proper management of aquifers. It is based on the singular integral theory of the boundary element method (BEM) that is implemented in a finite element sense. It has the capability of modelling confined and unconfined aquifers with parameters that vary in space (heterogeneous aquifers). The time-dependent second-order differential equations which governed flow in aquifers are solved in GEMFLOW by transforming those equations into integral ones using Green’s identity, and the resulting integral equations are solved in a discrete manner in elements which are used to discretise the flow region. GEMFLOW gives the liberty to dicretise the aquifer by a variety of elements: either linear or quadratic rectangular or triangular elements. For the unconfined flow problem which is nonlinear, the Newton-Raphson algorithm is used to linearise the nonlinear discrete equations. Wadessy: Numerical experiments Wadessy was employed to determine the least cost design of three popular WRN examples in literature. Below, results from this exercise are compared with other published WRN design models. 544 Advances in Materials and Systems Technologies