Godfrey A. Walters

A genetic programming approach to Rainfall-Runoff modelling

Planning for sustainable development of water resources relies crucially on the data available. Continuous hydrologic simulation based on conceptual models has proved to be the appropriate tool for studying rainfall-runoff processes and for providing necessary data. In recent years, artificial neural networks have emerged as a novel identification technique for the modelling of hydrological processes. However, they represent their knowledge in terms of a weight matrix that is not accessible to human understanding at present. This paper introduces genetic programming, which is an evolutionary computing method that provides a ‘transparent’ and structured system identification, to rainfall-runoff modelling. The genetic-programming approach is applied to flow prediction for the Kirkton catchment in Scotland (U.K.). The results obtained are compared to those attained using two optimally calibrated conceptual models and an artificial neural network. Correlations identified using data-driven approaches (genetic programming and neural network) are surprising in their consistency considering the relative size of the models and the number of variables included. These results also compare favourably with the conceptual models.

Improved design of “Anytown” distribution network using structured messy genetic algorithms

Abstract The recently introduced Structured Messy Genetic Algorithm model for optimising water distribution network rehabilitation is expanded to include not only pipe rehabilitation decisions but also pumping installations and storage tanks as variables. The formulation of the model is detailed with two approaches presented for handling the design of storage within the system. The application of the model to the benchmark “Anytown” problem is used as an example of its capabilities, with cheaper designs being produced than any previously published.

Multiobjective Genetic Algorithms for Pump Scheduling in Water Supply

Cost minimisation is the main issue for water companies when establishing pumping regimes for water distribution. Energy consumption and pump maintenance represent by far the biggest expenditure, accounting for around 90% of the lifetime cost of a water pump. This paper introduces multiobjective Genetic Algorithms (GAs) for pump scheduling in water supply systems. The two objectives considered are minimisation of energy and maintenance costs. Pump switching is introduced as a surrogate measure of maintenance cost. The multiobjective algorithm is compared to the single objective GA, with both techniques improved by using hybridisation with a local-search method.

A hybrid inverse transient model for leakage detection and roughness calibration in pipe networks

Leakage detection and calibration of hydraulic models are important issues for the management of water and other distribution networks. An inverse transient model based on a hybrid search technique is presented here. The inverse model is developed mainly for the detection of leaks in water distribution networks. The inverse transient procedure is formulated as a constrained optimisation problem of weighted least-squares type. Two optimisation techniques are tested: the genetic algorithm (GA) and the Levenberg-Marquardt (LM) method. After examining their performance, a new hybrid genetic algorithm (HGA) is developed to exploit the advantages of combining the two methods. The resulting HGA-based inverse transient model is compared with the GA and LM-based inverse transient models using two case studies. The HGA-based inverse transient model proved to be more stable than the LM-based model and it is more accurate and much faster than the GA-based inverse transient model.

EVOLUTIONARY DESIGN ALGORITHM FOR OPTIMAL LAYOUT OF TREE NETWORKS

A model for the optimal layout selection for a network with a tree structure is described. Such networks occur in sewerage, irrigation, water and gas supply and distribution, and in many other diverse areas of engineering. The model is based on Evolutionary Design and Genetic Algorithm principles. Novel features include the use of an efficient tree growing algorithm and the incorporation of redundant ‘genetic’ information within the ‘reproduction’ phase. Software performance is described and discussed using two examples.

Rehabilitation strategies for water distribution networks: a literature review with a UK perspective

Abstract Primarily, a rehabilitation strategy should aim to satisfy the regulatory requirements set down in respect of water distribution network operation. However, water companies in the UK have come to recognise that the business needs associated with the improvement of the deteriorating fabric of their distribution networks extend beyond these requirements. Extra economy can be gained by operating the networks efficiently based on a rehabilitation strategy which considers the associated costs over an extended period. Economic, hydraulic, reliability and water quality performance criteria must be optimised as part of an effective strategy. Numerous rehabilitation decision making approaches have been presented. However, many have adopted flawed economic approaches and have been based inadequately on one or two selected performance criteria. Few models have considered the extended planning horizons associated with a whole-life costing approach to this problem. However, the multi-objective optimisation approaches which have been developed recently have the potential to be developed into the required whole-life costing model based on the appropriate economic model and performance criteria.

OPTIMAL LAYOUT OF TREE NETWORKS USING GENETIC ALGORITHMS

Two alternative Genetic Algorithm methods for the optimal selection of the layout and connectivity of a dendritic pipe network are presented and compared. Both methods assume that the layout is selected from a directed base graph defining all feasible arcs. The first method uses a conventional binary string to represent the network layout, with the second method using a more efficient integer representation. Comparison with an exact Dynamic Programming formulation is made. The Genetic Algorithm techniques are shown to be very effective search procedures for the class of network optimization problem investigated.

Symbolic and numerical regression: experiments and applications

This paper describes a new method for creating polynomial regression models. The new method is compared with stepwise regression and symbolic regression using three example problems. The first example is a polynomial equation. The two examples that follow are real-world problems, approximating the Colebrook-White equation and rainfall-runoff modelling. The three example problems illustrate the advantages of the new method.

Optimal sampling design for model calibration using shortest path, genetic and entropy algorithms

Abstract Before hydraulic network models are to be used for predictive purposes with any degree of confidence, they need to be calibrated against field data. The selection of field test locations in a water distribution system for collection of data for such calibration, also called sampling design, is often done by subjective judgement. While providing adequate data and calibration in many applications such approaches do not ensure optimal or near optimal data collection and parameter estimation. There is therefore merit in establishing more objective and rational criteria for determining the most appropriate placement of monitoring points. Three different sampling design methods applied to selecting pressure monitoring point locations for estimating pipe roughness coefficients are presented in this paper. The first two methods rank the sampling locations based on shortest path algorithms logic. The third sampling design method attempts to identify the optimal set of monitoring points by maximizing the Shannon entropy function using a genetic algorithm (GA) search method. Results of application of the different sampling design methods to an example network are presented and a comparison made with a sampling system designed by an expert.

AN EVOLUTION PROGRAM FOR OPTIMAL PRESSURE REGULATION IN WATER DISTRIBUTION NETWORKS

Leakage losses in a water distribution network increase significantly for higher pressures and an obvious way of reducing losses is by reducing network pressures. This paper presents a methodology for pressure regulation in a water distribution network using an evolution program, encompassing the principles of evolutionary design and genetic algorithms. The optimization problem of minimizing the pressure heads is formulated with the settings of isolating valves as decision variables and minimum allowable pressures as constraints. The algorithm developed incorporates a steady-state network hydraulic analysis model based on the linear theory method. Computational results for two example networks demonstrating the effectiveness of the methodology are presented.